University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA CONSUMER BEHAVIOURAL RESPONSES TO ONLINE DISPLAY ADVERTISING IN GHANA – THE EFFECTS OF AD CHARACTERISTICS, CONSUMER ATTITUDE AND INTERNET USER MODE BY PRISCILLA MENSAH (10441687) THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON, IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF PHD MARKETING DEGREE DECEMBER 2019 University of Ghana http://ugspace.ug.edu.gh DECLARATION I do hereby declare that this thesis is the result of my own research and has not been presented in whole or in part for any academic award in this or any other university. All references used in this thesis have been fully acknowledged. I bear sole responsibility for any shortcomings. …………………………………….. ………………………….. PRISCILLA MENSAH DATE (10441687) i University of Ghana http://ugspace.ug.edu.gh CERTIFICATION I hereby certify that this work was supervised in accordance with procedures laid down by the University. ………………………………………. ………………………….. PROFESSOR BEDMAN NARTEH DATE (PRINCIPAL SUPERVISOR) ………………………………………… ………………………….. DR. STEPHEN MAHAMA BRAIMAH DATE (CO-SUPERVISOR) ii University of Ghana http://ugspace.ug.edu.gh DEDICATION I dedicate this work to my terrific husband, Dr. Raphael Odoom for his unwavering support and understanding all through the years. Thanks for giving me the peace of mind I needed to complete this task. You are my God-given solace. To my adorable son, RJ, for being my “muse” - I needed to wrap up efficiently so I could meet you. To my mum, Madam Elizabeth Aryee, to whom I remain eternally grateful for the gift of education. To the memory of my late dad, Mr. Francis Feller Mensah for his timeless counsels that set my course and still drive my decisions in life. iii University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENT I am tremendously grateful to God Almighty, the invisible hand that has sustained me throughout my study. I am also particularly thankful to my research supervisors, Professor Bedman Narteh and Dr. Stephen Mahama Braimah for their guidance and assistance which aided in the successful completion of this thesis. The confidence you both had in me was a key driving force. I further acknowledge the faculty members of the Marketing and Enterpreneurship Department of UGBS for their thoughts, criticisms and contributions toward shaping this research work. I am truly appreciative of my husband, Dr. Raphael Odoom for being the first point of call for my “wild ideas” and for being a listening ear and a shoulder in the good and bad “PhD moments” as we have come to call them. Thanks, for being extremely supportive. I’m also very grateful to my entire family. First, my mum - Madam Elizabeth Aryee, for giving me much more than she ever had. Second, to my big sister and brother - Linda Dairo and Godswill T.K. for their sacrifices and contributions toward my education. Then to Henrose Samuella, Eugene Kobby Elikplim, Ephraim Elinam, Susan Naa Ameley, and Elmira Elorm for spurring me on in their own special ways. I wish to express my gratitude as well to Mrs. Matilda Adams, with whom I shared several thoughts along the way; and to my colleagues, Mrs. Victoria Mann, Mr. Atsu Nkukpornu, Mr. Kofi Aning Jnr. and all the others for the moot and preparatory seminars we had. I further wish to acknowledge particular persons, Dr. Ebo Afful and Mr. George Kwabena Asamoah both of the Ghana Institute of Journalism, as well as Maxwell Ocloo, Irene Kafui Esi Asinyo and Richel Akosua Selorm Quarshie for helping in one way or another. Finally, to every individual that made time to particpate in the survey, I’d like to say - your responses made this research possible and for that, I’m more than grateful. Thanks a million. iv University of Ghana http://ugspace.ug.edu.gh ABSTRACT The unrivalled strength of the internet in terms of reach, richness of information, targeting and interactivity has made it a vital medium for advertising which has caused a remarkable growth in online/internet advertising. Over the past decade, online display advertising (ODA) has emerged as the fastest growing category of online advertising spurring several firms and businesses to invest heavily in display ads in order to connect with and keep their brands in front of consumers. While ODA holds various benefits for firms, both the literature and practice point to advertising clutter as a major challenge that has accompanied its growth. This has left advertisers struggling to stand out among the clutter to capture consumer attention and also, gain insights into what types of display advertising work best in the online environment. This thesis sets out to provide a theoretical and practical understanding of consumer behavioural responses to online display advertising by offering insights into how ODA characteristics, consumer-specific factors like attitude toward online advertising (ATOA) and user mode (internet usage motive) as well as how the nature of the advertised brand (product or service) enhance ad acceptance and minimise ad avoidance behaviours of consumers in an emerging market setting – Ghana. The study draws on the stimulus organism response (SOR) model and the reversal theory to propose a conceptual framework to empirically examine and explain the interrelationships among these variables. Adopting a positivist paradigmatic stance, the thesis employed a quantitative cross-sectional survey to collect data from 592 internet users in Ghana. Data gathered was analysed using Structural Equation Modelling (SEM) as well as ANOVA and Binary Logistic Regression. The study finds that, interactivity, informativeness and personalisation were the relevant ODA characteristics that serve direct stimuli functions in eliciting approach behaviours (ad acceptance). In addition, attitude toward online advertising (ATOA) emerged as a significant mediator (facilitator) of the positive relationship and the negative relationship between these ODA characteristics and ad acceptance as well as ad avoidance respectively. Also, user mode significantly moderated the relationship between personalisation and ad acceptance as well as informativeness and ATOA, and there were differences in behavioural responses of consumers based on the nature of the advertised brand such that, ad avoidance was higher for service-featured ODAs and ad acceptance was higher for product-featured ODAs. These findings bring to the fore knowledge that, reliance on ODA characteristics although may be quite adequate in eliciting positive behavioural responses, may not be sufficient in lessening avoidance behaviours toward display ads; rather how these ODA stimuli generate positive consumer ATOA is more crucial. Findings also point to understanding consumers’ motive for internet usage and the nature of the brands firms seek to promote as vital issues for advertisers and publishers if the appropriate ad features are to be selected in designing display ads that will suit the brands as well as appeal to the various user groups in order to generate the required attitude and responses. Further practical and theoretical implications of the study are discussed in the thesis. v University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENTS DECLARATION ................................................................................................................ i CERTIFICATION ............................................................................................................ ii DEDICATION ................................................................................................................. iii ACKNOWLEDGEMENT ............................................................................................... iv ABSTRACT ........................................................................................................................ v TABLE OF CONTENTS ................................................................................................. vi LIST OF TABLES ............................................................................................................ x LIST OF FIGURES ......................................................................................................... xi CHAPTER ONE: INTRODUCTION .............................................................................. 1 1.1 CHAPTER OVERVIEW ........................................................................................ 1 1.2 BACKGROUND TO THE STUDY ....................................................................... 1 1.3. PROBLEM FORMULATION AND RESEARCH GAPS ..................................... 4 1.3.1 Issue Gaps ............................................................................................................... 5 1.3.2 Theoretical and Contextual Gaps ............................................................................ 9 1.3.3 Methodological Gaps ............................................................................................ 11 1.4 RESEARCH QUESTIONS & OBJECTIVES ...................................................... 12 1.5 SIGNIFICANCE OF THE STUDY ..................................................................... 13 1.5.1 Theoretical Contributions ..................................................................................... 14 1.5.2 Practical Contributions ......................................................................................... 15 1.6 OUTLINE OF THE THESIS & CHAPTER DISPOSITION ............................... 15 1.7 CHAPTER SUMMARY ...................................................................................... 18 CHAPTER TWO: CONTEXT OF THE STUDY ........................................................ 19 2.1 INTRODUCTION ................................................................................................ 19 2.2 A BRIEF PRIMER ON GHANA’S ECONOMY ............................................... 19 2.3 INTERNET USAGE IN GHANA ........................................................................ 21 2.4 THE ADVERTISING SECTOR IN GHANA ...................................................... 23 2.5 ONLINE ADVERTISING IN GHANA ............................................................... 25 2.6 CHAPTER SUMMARY ...................................................................................... 27 CHAPTER THREE: THEORETICAL FRAMEWORK ............................................ 28 3.1 CHAPTER OVERVIEW ...................................................................................... 28 3.2 INTRODUCTION ................................................................................................ 29 3.3 JUSTIFICATION FOR THE UNDERPINNING THEORIES ............................ 30 3.4 STIMULUS ORGANISM RESPONSE MODEL ................................................ 32 3.4.1 Overview of the Theory ........................................................................................ 32 3.4.2 Components of the Theory ................................................................................... 34 3.4.2.1 Stimulus ............................................................................................................... 34 3.4.2.2 Organism ............................................................................................................ 35 3.4.2.3 Response ............................................................................................................. 36 3.4.3 Applications and Relevance of the theory to this Thesis ...................................... 38 3.5 REVERSAL THEORY ........................................................................................ 44 3.5.1 Overview of the Theory ........................................................................................ 44 3.5.2 Telic Versus Paratelic Perspective of Reversal Theory ........................................ 45 3.5.3 Applications and Relevance of the Theory to this Thesis .................................... 47 3.6 OTHER THEORETICAL PERSPECTIVES ....................................................... 49 3.7 CHAPTER SUMMARY ...................................................................................... 52 vi University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR: REVIEW OF ODA RESEARCH ................................................. 54 4.1 CHAPTER OVERVIEW ...................................................................................... 54 4.2 INTRODUCTION ................................................................................................ 54 4.3 PROCEDURE FOR THE REVIEW .................................................................... 56 4.3.1 Scope of the Review ............................................................................................. 56 4.3.2 Literature Search, Article Selection and Information Extraction ......................... 57 4.3.3 Distribution by Database, Journal and Year ......................................................... 58 4.4 ADVERTISING AND ONLINE ADVERTISING .............................................. 60 4.4.1 Typologies of Online Advertising ........................................................................ 62 4.4.1.1 Online Display Advertising (ODA) ..................................................................... 64 4.5 MAPPING ODA RESEARCH: MAJOR THEMES, ISSUES & EVIDENCES . 65 4.5.1 Antecedents to Online Display Advertising Effectiveness ................................... 66 4.5.1.1 Ad-related Issues: Design features or Executional characteristics ................... 67 4.5.1.2 Consumer-Related Issues .................................................................................... 82 4.5.2 Assessing Online Display Advertising Effectiveness ........................................... 88 4.5.2.2 Conative/Behavioural Measures ........................................................................ 90 4.5.3 Consumer Attitude toward Online Advertising .................................................... 93 4.6 THEORETICAL APPROACHES IN ODA RESEARCH ................................... 95 4.7 METHODOLOGICAL APPROACHES IN ODA RESEARCH ......................... 99 4.8 GEOGRAPHICAL AND ECONOMIC DISPERSION OF ODA RESEARCH ..... 102 4.9 RESEARCH GAPS AND FUTURE RESEARCH AVENUES ........................ 103 4.9.1 Gaps in Issues and Evidences ............................................................................. 104 4.9.2 Theoretical and Contextual Gaps ........................................................................ 108 4.9.3 Gaps in Methodological Approaches .................................................................. 111 4.10 CHAPTER SUMMARY .................................................................................... 113 CHAPTER FIVE: CONCEPTUAL FRAMEWORK ................................................ 115 5.1 CHAPTER OVERVIEW .................................................................................... 115 5.2 INTRODUCTION & ASSUMPTIONS UNDERLYING THE FRAMEWORK .... 116 5.3 HYPOTHESES DERIVATION ......................................................................... 118 5.3.1 Conceptualising Behavioural Responses ............................................................ 118 5.3.2 Online Display Advertising (ODA) Characteristics as Stimuli .......................... 120 5.3.2.1 Interactivity ....................................................................................................... 120 5.3.2.2 Placement ......................................................................................................... 122 5.3.2.3 Informativeness ................................................................................................. 124 5.3.2.4 Personalisation ................................................................................................. 125 5.3.2.5 Exposure Condition .......................................................................................... 126 5.3.3 Attitude toward Online Advertising ................................................................... 129 5.3.4 The Mediating Role of Attitude toward Online Advertising .............................. 131 5.3.5 The Moderating Role of Internet User Mode ..................................................... 132 5.3.6 Control Variables ................................................................................................ 135 5.4 CHAPTER SUMMARY .................................................................................... 136 CHAPTER SIX: RESEARCH METHODOLOGY ................................................... 137 6.1 CHAPTER OVERVIEW .................................................................................... 137 6.2 INTRODUCTION .............................................................................................. 137 6.3 RESEARCH PARADIGM AND PHILOSOPHICAL VIEWPOINTS .............. 138 6.3.1 Positivism as the Paradigmatic Stance of this Study .......................................... 141 6.4 RESEARCH PURPOSE/DESIGN ..................................................................... 144 6.4.1 Explanatory Research as the Chosen Design ...................................................... 145 6.5 RESEARCH APPROACH ................................................................................. 146 vii University of Ghana http://ugspace.ug.edu.gh 6.5.1 Quantitative Research ......................................................................................... 146 6.5.2 Qualitive Research .............................................................................................. 147 6.5.3 Mixed-Methods Research ................................................................................... 148 6.6 RESEARCH STRATEGY .................................................................................. 149 6.6.1 Survey as the Chosen Method for this Study ...................................................... 151 6.7 DATA COLLECTION METHODS ................................................................... 153 6.7.1 Population, Sampling Technique and Sample Size Determination .................... 153 6.7.1.1 Sampling Technique for the Study .................................................................... 154 6.7.1.2 Sample Size for the Study ................................................................................. 156 6.7.2 Survey Instrument Development and Administration ........................................ 157 6.7.2.1 Pre-testing and Questionnaire Modification .................................................... 159 6.8 MODE OF DATA ANALYSIS .......................................................................... 161 6.8.1 Factor Analysis ................................................................................................... 161 6.8.1.1 Assessing the Appropriateness of the Data Set for Factor Analysis ................ 162 6.8.1.2 Factor Extraction, Rotation and Interpretation ............................................... 163 6.8.2 Structural Equation Modelling (SEM) ................................................................ 164 6.8.2.1 Two-Stage SEM ................................................................................................ 165 6.8.2.2 Evaluating the Fitness of the Model ................................................................. 167 6.8.2.3 Mediation Analysis ........................................................................................... 168 6.8.2.4 Moderation Analysis ......................................................................................... 170 6.8.3 Analysis of Variance (ANOVA) ........................................................................ 172 6.8.4 Logistic Regression ............................................................................................ 172 6.8.5 Unit of Analysis .................................................................................................. 173 6.9 RELIABILITY AND VALIDITY ...................................................................... 174 6.9.1 Reliability of the Research Instrument ............................................................... 174 6.9.2 Validity of the Research Instrument ................................................................... 176 6.10 ETHICAL CONSIDERATION ......................................................................... 177 6.11 CHAPTER SUMMARY ................................................................................... 178 CHAPTER SEVEN: DATA ANALYSIS AND PRESENTATION OF RESULTS ..... 179 7.1 CHAPTER OVERVIEW .................................................................................... 179 7.2 SAMPLE CHARACTERISTICS ....................................................................... 179 7.3 DESCRIPTIVE STATISTICS ............................................................................ 182 7.4 EXPLORATORY FACTOR ANALYSIS (EFA) .............................................. 185 7.4.1 Extraction, Rotation, Reliability and Re-specification of the EFA .................... 186 7.4.2 Examination of Common Method Variance ...................................................... 188 7.5 STRUCTURAL EQUATION MODELLING (SEM) ........................................ 189 7.5.1 Confirmatory Factor Analysis/Measurement Models ........................................ 190 7.5.1.1 Reliability and Validity of the Final Measurement Model ............................... 191 7.5.2 Structural Model ................................................................................................. 194 7.5.2.1 Assessment of the Direct Structural Model ...................................................... 195 7.5.2.2 Test for Mediation Effects ................................................................................. 198 7.5.2.3 Test for Moderation Effects .............................................................................. 201 7.6 ONE-WAY ANALYSIS OF VARIANCE (ANOVA) ....................................... 203 7.7 LOGISTIC REGRESSION ................................................................................ 204 7.8 CHAPTER SUMMARY ................................................................................... 207 CHAPTER EIGHT: DISCUSSION OF FINDINGS .................................................. 209 8.1 CHAPTER OVERVIEW .................................................................................... 209 8.2 INTRODUCTION AND DISCUSSION OF FINDINGS .................................... 209 8.2.1 The Stimuli Effect of ODA Characteristics on ATOA and Behavioural Responses ..... 210 viii University of Ghana http://ugspace.ug.edu.gh 8.2.1.1 Personalisation ................................................................................................. 211 8.2.1.2 Interactivity ....................................................................................................... 213 8.2.1.3 Informativeness ................................................................................................. 215 8.2.1.4 Exposure Condition .......................................................................................... 217 8.2.2 Mediating Effect of ATOA in the ODA Characteristic-Response Relationship ....... 219 8.2.3 The Moderating Effect of Internet User Mode ................................................ 221 8.2.4 Behavioural Responses to Product and Service-featured ODAs ..................... 223 8.3 CHAPTER SUMMARY .................................................................................... 225 CHAPTER NINE: SUMMARY, CONCLUSIONS AND IMPLICATIONS ........... 226 9.1 CHAPTER OVERVIEW .................................................................................... 226 9.2 SUMMARY OF THE RESEARCH AND MAJOR FINDINGS ....................... 226 9.3 REFLECTIONS .................................................................................................. 230 9.3.1 Reflection on Theories ........................................................................................ 230 9.3.2 Reflection on Conceptual Framework ................................................................ 232 9.3.3 Reflection on Methodological Approaches ........................................................ 234 9.4 CONTRIBUTIONS AND IMPLICATIONS OF THE RESEARCH ...................... 235 9.4.1 Theoretical Contribution and Implications ............................................................... 236 9.4.2 Implications for Online Advertising Practice ........................................................... 239 9.5 CONCLUSIONS ..................................................................................................... 244 9.6 RESEARCH LIMITATIONS AND AVENUES FOR FUTURE RESEARCH ...... 245 REFERENCES .............................................................................................................. 251 APPENDICES ................................................................................................................ 282 Appendix A1 Online Advertising Thematic Areas & Methodological Approaches ....... 282 Appendix A2 Numbered Articles used for the Review ................................................... 283 Appendix A3 Publications by Journals and Years .......................................................... 284 Appendix A4 Summary of Study, Measured Variables & Gaps in ODA Research ........ 285 Appendix B Survey Instrument ....................................................................................... 295 Appendix C Ethical Approval ......................................................................................... 299 ix University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Table 2.1 Selected Indicators of Economic Growth in Ghana .......................................... 20 Table 2.2 Internet Usage and Penetration Rate in Ghana .................................................. 22 Table 3.1 Summary of Selected Articles Underpinned by the SOR Model ...................... 39 Table 4.1 Selected online Advertising Definitions ............................................................ 62 Table 4.2 Online Advertising Forms Studied .................................................................... 67 Table 4.3 Distribution of Articles by Geographical Setting & Theoretical Approaches ... 98 Table 6.1 Major Philosophical Paradigms in Social Science Research ........................... 141 Table 6.2 Summary of Study Constructs and Sources .................................................... 158 Table 7.1 Demographic Profile of Respondents .............................................................. 180 Table 7.2 Internet Usage Profile of Respondents ............................................................ 181 Table 7.3 Descriptive Statistics – ODA Characteristics .................................................. 183 Table 7.4 Descriptive Statistics – Attitude and Behavioural Responses ......................... 185 Table 7.5 KMO and Bartlett’s Test ................................................................................. 186 Table 7.6 Rotated Component Matrix and Scale Reliability ........................................... 187 Table 7.7 Modification and Model Fit Summary of Measurement Model ...................... 191 Table 7.8 CFA Results for Final Measurement Model .................................................... 193 Table 7.9 Correlation Matrix for Discriminant Validity ................................................. 194 Table 7.10 Structural Model Assessment Results – Direct Paths .................................... 196 Table 7.11 Structural Model Assessment Results – Mediated Paths ............................... 200 Table 7.12 Invariance Test Results: Multigroup Analysis .............................................. 202 Table 7.13 Structural Model Results for Multigroup (User Mode) Moderation ............. 203 Table 7.14 ANOVA Results ............................................................................................ 204 Table 7.15 Logistic Regression: Likelihood Ratio for Behavioural Intensity ................. 206 Table 9.1 Summary of key Findings ................................................................................ 229 x University of Ghana http://ugspace.ug.edu.gh LIST OF FIGURES Figure 1.1 Structure and Flow of the Thesis ...................................................................... 17 Figure 2.1 Population and Internet Usage Growth ............................................................ 22 Figure 2.2 Internet Advertising Revenue ........................................................................... 26 Figure 3.1 SOR Model ....................................................................................................... 34 Figure 4.1 Distribution of Articles by Year ....................................................................... 59 Figure 4.2 Cited Executional Characteristics in ODA Research ....................................... 69 Figure 4.3 Distribution of Articles by Methodology ....................................................... 102 Figure 4.4 Distribution of Articles by Economic Setting ................................................ 103 Figure 5.1 Conceptual Framework .................................................................................. 117 Figure 6.1 Structure of the Methodology ........................................................................ 138 Figure 7.1 Final Measurement Model ............................................................................. 192 Figure 7.2 Structural Model Results for Direct Paths ...................................................... 195 Figure 7.3 Structural Model Results for Mediated Paths ................................................ 199 Figure 9.1 Post-study Framework .................................................................................... 233 xi University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION 1.1 CHAPTER OVERVIEW This chapter provides a window into the study and an overview of the direction of the entire thesis. The chapter starts off with the background of the study which discusses current issues in online advertising, and notable challenges that have attended its growth leaving practitioners in want of indicators that may help elicit favourable consumer responses to their online display advertisements. The problem and the corresponding gaps in the literature the study seeks to address are presented along with key questions and objectives outlined to guide the study. The chapter then discusses the expected contributions of this thesis to both theory and practice. Lastly, to provide a roadmap for the entire thesis, the chapter presents the structure or flow of the study. 1.2 BACKGROUND TO THE STUDY Since its advent, the internet has earned a special place as a medium of choice in present day advertising, attributable to its unequalled strength regarding reach, targeting, richness of information, and interactivity (McCoy et al., 2012; Nihel, 2013; Liu & Mattila, 2017). More so, the tremendous surge over the past decade, in the number of internet users and the amount of time they spend online (Souiden et al., 2017) speak to the internet’s standing and potential as a pertinent advertising medium. In 2018, about half the global population are said to be online (i.e. are internet users) (Internet World Statistics, 2018). And so today, to create relevance among, reach out to, connect with, and keep brands in front of their customers and target segments, several firms and businesses do not only have online presence, but have also shifted most of their marketing strategies online and are intensifying their advertising efforts in the online environment (Kim, 2018). 1 University of Ghana http://ugspace.ug.edu.gh As a result, Internet advertising (online advertising), has grown exponentially, contesting the prevalence of traditional media such as television and print, and has become a sizable portion of the total advertising market (Belanche et al., 2017; Liu-Thompkins, 2019). In 2017, for instance, online advertising represented 41 percent of global advertising spending and reached $209 billion (Kafka & Molla 2017). It is also projected that by 2020, the global market for online advertising, dominated by search and display advertising will reach $265 billion (Companiesandmarkets.com, 2014; IAB, 2016). At this instant, it is relevant to point out that the focus of this thesis is on online display advertising (ODA), which represents graphical ads of different formats that are hosted on social networking sites, news sites, blogs, and commercial websites etc. (Chapelle et al., 2014) that internet users or website visitors see along with other content. Growing at 17% every year, ODA currently accounts for 49.7% of the online advertising market and is anticipated to reach $222.3 billion by 2022, making it the fastest growing category of online advertising (eMarketer, 2018). This is spurring marketers/advertisers and web publishers to invest heavily in display advertising (Hof, 2013; Fridgeirsdottir & Najafi-Asadolahi, 2018). Because most display ads are incorporated with mechanisms, that allow consumers to clickthrough, they create a high probability of advertising engagement. Also, internet-targeting technologies allow companies to tailor their ODA to consumers based on previous browsing history and personal information among other things (Sridhar et al., 2016). From an industry standpoint, all these are strong indicators of the steady growth and relevance of online display advertising in the today’s marketplace. Paradoxically, this increasing growth of online display advertising presents marketers and advertisers with some challenges; the most significant of which is the perception of advertising clutter. Given its pervasive nature, the internet has a high advertising density (Ha & McCann, 2008; Goldfarb, 2014). At every visit to the internet, users are exposed to 2 University of Ghana http://ugspace.ug.edu.gh torrents of ODAs that compete on the medium for their attention causing them to pay selective attention to ads as well as avoid them (Rejòn-Guardia & Martinez-Lopez, 2014). Also, the rapidly changing online advertising landscape, driven by new and evolving technology-enabled features, and the consistent emergence of diverse ODA formats from static banners to interactive audio-visual ads (Li & Lo, 2015), complicate advertisers’ decisions regarding which formats will yield the best results for their campaigns (de Pelsmacker & Neijens, 2012; Belanche et al., 2017). Because ODAs are capable of appealing to and irritating consumers depending on certain ad-related and consumer-related factors (Brajnik & Gabrielli, 2010; Janssens et al., 2012), marketers and advertisers are grappling to stand out among the clutter of copious online advertising to not only capture consumer attention, but also hold it for elaboration of the advertising content which may translate into other positive behavioural effects (Kuisma et al., 2010; Nihel, 2013). There is, therefore, a mounting need to understand consumer reactions and responses to online display advertising in order to provide pertinent indicators and actionable knowledge on how firms can effectively employ it to enhance their advertising goals. From an academic perspective, online advertising in general has gained scholarly attention following its rapid global increase and corresponding practice-related challenges, (Smit et al., 2014; Hoban et al., 2015; Nasir, 2017). Yet, research in this area is considered the third most studied form of advertising after print and television (Kim et al., 2014). Although this discrepancy may be ascribed to the relative nascence of the phenomenon given its brief history which only began in the mid-1990s (Tutaj & Reijmersdal, 2012), some compelling reasons have been put forth by scholars regarding why the quantum of research work in the area should experience an upsurge and catch up with the pace of conventional/traditional advertising research. Foremost, ad networks and host websites continually track consumers’ online actions, providing a remarkable volume of data at individual levels which makes 3 University of Ghana http://ugspace.ug.edu.gh online advertising highly measurable even in international contexts, unlike traditional advertising (Lewis et al., 2011; Sridhar et al., 2016). Additionally, online advertising has a pronounced interdisciplinary nature that transcends the fields of marketing, communication and advertising (Cho & Kang, 2006; Knoll, 2015), and is investigated by scholars in the field of information systems or management, economics and operations as well (e.g. Kim et al., 2012; Johnson, 2013; Tang et al., 2014). Nonetheless, in spite of the potential the field holds for academic investigations, there is a gap in advertising literature and by extension marketing literature. This study is thus, opportune as it responds to scholarly calls (e.g. Wojdynski & Evans, 2015; Liu-Thompkins, 2019) for further academic enquiries aimed at progressing the advertising field by providing theoretical and practical understanding of the value of online display advertising and how firms can use it appropriately. 1.3 PROBLEM FORMULATION AND RESEARCH GAPS In a saturated media environment where consumers are overwhelmed with advertisements and are gaining more control over their advertising exposure, advertisers are struggling to stand out among the clutter, capture consumer attention as well as generate favourable reactions. For this reason, there is a growing need for marketers and advertisers to gain insights into what types of advertising work best in online settings (Bright & Daugherty, 2012; Hoban et al., 2015). Over the past two decades, there has been a well-articulated stream of research on online advertising (e.g. Schlosser et al., 1999; Rodgers & Thorson 2000; Robinson et al., 2007; Ha, 2008; Goldfarb & Tucker, 2011; Goodrich, 2013; Aguirre et al., 2015; Parra-Arnau et al., 2017). Particularly, the possible antecedents to online display advertising effectiveness and the approaches for assessing the efficacy of online display ads have emerged as a timely focus of academic research aimed at offering insights that may suggest strategic ways to enhance positive online advertising effects and decrease 4 University of Ghana http://ugspace.ug.edu.gh negative outcomes (Martín-santana & Beerli-Palacio, 2012; Lambrecht & Tucker, 2013; Eshghi et al., 2017). In spite of the plethora of available studies, the empirical evidence regarding the underlying factors that drive online display advertising effectiveness, remain rather inconclusive and less disciplined (Bleier & Eisbensen, 2015a; Auschaitraku & Mucherjee, 2017). Especially, the measurement in terms of consumer responses need to be resolutely grounded in the literature. Pursuant to this, a review of the online display advertising literature was conducted to understand the major themes, issues and evidences in ODA studies, in order to identify gaps, and define a research route for this study. Following from the discourses in the review, a number of gaps are apparent in the literature that require scholarly deliberation. For the purpose of this study, these gaps are ordered along three major strands (issues, theories and context, and methodology), and are discussed in the subsequent subsections. 1.3.1 Issue Gaps Academic attention paid to online display advertising effectiveness has been considerably documented in the literature with most of the extant studies focusing on various ad formats and their possible effects (Tujat &Reijmersdal, 2012; Belanche et al., 2017; de Pelsmacker & Neijens, 2012; Auschaitraku & Mucherjee, 2017). These ad formats have similarities and distinctions based on certain executional factors and elements that characterize them (Burns & Lutz, 2008; Li & Meeds, 2007). Nonetheless, most of the studies that focus on the ad formats do not effectively delineate the underlying characteristics that drive the effect of such formats (e.g. Tutaj & Reijmersdal, 2012). Much more relevantly, studies that explicitly focus on these executional characteristics as antecedents to ODA effectiveness are limited (Rejón-Guardia & Martínez López, 2014). Of the several noteworthy ad characteristics studied (such as animation size, colour, sound, repetition, entertaining, language, and appeal 5 University of Ghana http://ugspace.ug.edu.gh etc.) issues of interactivity, informativeness, placement (context-congruency), and exposure condition among others, are still high on the agenda of expected research (Segev, 2014; Mahmoud, 2014; Kim, 2018). Additionally, scholarly attention to such others as, salience, and personalisation, which typify some relatively nascent ad formats (e.g. native ads and skippable videos) has been even more scant (Agarwal et al., 2011; de Pelsmacker & Neijens, 2012; Wojdynski & Evans, 2015). Although consumers’ perceptions and responses are driven by these various features that make up the ads or formats, relatively few studies have focused on these individual features, and so examining the effects of these features is, therefore, a pertinent call in the literature (Tang et al., 2014; Liu & Mattila, 2017) and the focus of this thesis. Furthermore, several consumer responses to online advertisements, considered measures of online advertising effectiveness, have been of critical interest to practitioners and academics (see Rosenkrans, 2009; Kireyev et al., 2015; Auschaitraku & Mucherjee, 2017). In the stream of literature reviewed, such behavioural responses as website visit (Hoban et al., 2015), ad acceptance (Belanche et al., 2017) ad clicking or mouse rollover (Wang & Sun 2010a; Bleier & Eisenbeiss, 2015a; Liu & Matila, 2017), ad avoidance (Seyedghorban et al., 2016), and purchase intention (Jung et al., 2011; Goodrich et al., 2015) among others, have been examined. These empirical investigations notwithstanding, the literature is still fragmented and inconclusive on which responses serve as useful measures of online display advertising effectiveness, and which ad characteristics generate favourable and unfavourable conative responses (Resnick & Albert, 2014; Belanche et al., 2017). The inconclusive results have been attributed to the investigation of different ad stimuli, methods, measures, audiences and contexts (Kuisma et al., 2010), which make such studies hardly comparable. Particularly, the literature depicts a lack of studies that consider both direction (positive and negative) as well as the intensity (active and passive) of behavioural 6 University of Ghana http://ugspace.ug.edu.gh responses in a single investigation. If advertisements are expected to generate positive effects of value to the advertiser and consumer but could have negative effects too (Tang et al., 2014), then the question remains as to what ad features cause consumers to respond positively or negatively toward ODAs. In order to contribute to the literature in these foregoing regards, this study investigates the effects of a multiplicity of pertinent ODA characteristics on behavioural response in order to provide an understanding on their influence, and under what circumstances they may be more or less effective (i.e. generate positive or negative responses). Also, while ad-related factors may influence consumer responses to online advertising, several empirical evidences have demonstrated that this relationship is not always direct and simple (Song et al., 2011; Liu & Mattila, 2017). For this reason, available literature abounds in investigations on the intervening, confounding and complimentary roles that consumer- related and product-related factors play in this relationship (e.g. Ching et al., 2013; Lambrecht & Tucker, 2013; Souiden et al., 2017). However, attitude as a consumer-related factor though very topical in the literature, has received research attention (e.g. Goodrich, 2013; Eshghi et al., 2017; Auschaitraku & Mucherjee, 2017) mostly as an outcome of online advertising, or an antecedent to other effects. While these studies have made valuable contributions to improve understanding of consumer attitude as a precursor to other outcomes or as a response in itself, they are deficient in two ways. First, with the exception of a few (e.g. Valaei et al., 2016; Souiden et al., 2017), existing studies have commonly examined attitude toward specific ads based on the scope of their research, rather than general attitude toward online advertising. Second, these studies are also limited in providing insights and explanations on how attitude functions as a facilitating variable in transmitting/enhancing the effect of advertising stimuli onto consumer responses. What is more, the few studies that have examined the intervening role of attitude toward online 7 University of Ghana http://ugspace.ug.edu.gh advertising such as Wang et al., (2009) and Wang and Sun (2010a; 2010b) also lack theoretical cohesion. Hence this study contends that further insights may be acquired if the mediating role played by attitude is examined from a general online advertising perspective grounded on a theoretical notion. It is also worth noting that because the internet is a goal-oriented medium, one of the recognised predictors of consumers reactions to online advertisements is “user mode” or the goal-directedness/orientation of consumers which is driven by their internet usage motive (Zanjani et al., 2011). However, online advertising studies that have examined its role in explicating responses to online display ads are wanting (Jung et al., 2014). An aspect that has particularly, been overlooked is how user mode may function as a contingency factor causing variations in behavioural responses (Bleier & Eisenbess, 2015b; Seyedghorban et al., 2016). This study focuses on user mode as a very relevant situational or consumer- related factor that may confound consumer responses to display advertising on the internet (as a goal-oriented evnironment). Moreover, although, evidence from the online advertising literature points to disparities in consumer processing and responses to online display ads of several product categories owing to product type peculiarities (e.g. Flores et al., 2014; Eshghi et al., 2017), the current stream of literature is deficient in evidence of such variations in the product-service context. Since these product-related and consumer-related variables offer explanations to how the effects of online display advertising may vary or be enhanced, they deserve further research and discussion in order to legitimise prior claims as well as provide new insights (Aguirre et al., 2015; Seyerghorban et al., 2016). As such, this thesis as well examines the moderating effect of user mode on the online ad characteristics- attitude-behavioural response relationships, and also assesses the disparities (or lack thereof) that might exist in consumer responses to online display advertising of services and products. 8 University of Ghana http://ugspace.ug.edu.gh Following these issue gaps, the current study attempts to understand the relationship between online display advertising and consumer behavioural responses, by arguing that though the effects (behavioural responses) of ODAs are reliant on ad-related features, certain consumer-related variables may mediate and/or moderate these relationships (van Reijmersdal et al., 2016; Belanche et al., 2017), and the nature (product/service) of the advertised brand may also cause variances in these responses. There is thus, the need to clearly delineate these conditions, and to ascertain the degree to which they influence, enhance and/or confound behavioural responses to online display advertising. 1.3.2 Theoretical and Contextual Gaps From a theoretical perspective, the literature documents a considerable amount of online advertising studies that have adopted atheoretical (e.g. construct/category-based and model- based) approaches, relative to theory-based approaches. Amid the theories that have been used in ODA research, the stimulus organism response (SOR) model and the reversal theory have been sparsely applied in spite of their unique perspectives on how ad-related and consumer-related factors may influence online advertising effects or outcomes (Jung et al., 2014; Bleier &Eisenbess, 2015b). Although a number of the studies that were reviewed referred to features and executional elements of ODA as relevant advertising stimuli antecedental to the effectiveness of display ads, little to no attention is given to the stimulus organism response model which can provide some insights on ODA characteristics and their relationships with behavioural responses of consumers. It is fair to point out that while a considerable number of online empirical works have employed this theory, the focus has been on online retailing and shopping (e.g. Kim & Lennon, 2012; Kamboj et al., 2018), leaving the stimulus role played by the executional features that characterise online display advertising unexplored. The SOR model depicts how external stimuli affect an individual’s internal state and subsequently, their behaviour (Mehrabian & Russell, 1974). Because the 9 University of Ghana http://ugspace.ug.edu.gh internet is a stimulus rich medium, with commercial websites featuring several advertisements on a single webpage (Kuisma et al., 2010), examining how advertisers could leverage their ODA features as stimuli that may generate positive attitude toward online advertsing and behavioural responses from the viewpoint of the SOR model could help provide theoretically and emperically grounded insights for theory progression (Tang et al., 2014; Bleier & Eisenbess, 2015b). This will also allow for sturdy claims to be made about the stimuli function of ODA characteristics and their link with consumers’ attitude toward online advertising and behavioural responses. The stimulus organism response model helps explain the persuasive functions of advertising and how ODA characteristics act as stimuli generating internal (attitude) as well as external (behaviour) responses. Nevertheless, it falls short of explaining how user mode (goal orientation/internet usage motive) may confound these focal relationships (Bleier & Eisenbess, 2015b). In effect the reversal theory which offers explanations on how a contingency factor like “user mode” may moderate or cause variations in behavioural responses, has also been largely ignored in the ODA context (Seyeghorban et al., 2016), as such, the possible link between these two theory streams has not been established. In light of the above, this study demonstrates empirically how consumer behavioural responses to online advertising are influenced by both ad-related and consumer-related factors in a coherent manner from these theoretical viewpoints. In so doing, the current study extends the explanatory power of these theories into the online advertising context and contributes to the theory integration call in the broader advertising literature (Jung et al., 2014; Faber, 2015; Johar, 2016). From a contextual viewpoint, the study, identifies a gap as empirical works on online advertising are greatly skewed towards data emanating from developed nations (e.g. Martin- 10 University of Ghana http://ugspace.ug.edu.gh Santana & Beerli-Palacio, 2012; Hoban et al., 2015; Hussain et al., 2018). While some studies (e.g. Wang & Sun, 2010a; 2010b) have provided comparative understanding of online display advertising and its effects from both developed and developing market settings, the emerging market perspectives appear limited (Valaei et al., 2016; Eshghi et al., 2017). Emerging markets differ in several ways from developed markets. Also, differences exist among emerging markets as well because, such markets within Eastern Europe for instance, vary extensively in culture, economic development, internet penetrations, online advertising uptake, as well as the population of online consumers and their online behaviours from those in for instance, sub-Saharan Africa (Boone et al., 2010). The study, therefore, considers the paucity of research into online advertising effects from emerging markets or developing contexts as significant, and attempts to bridge this gap by providing perspectives from a developing market setting – Ghana, where internet penetration and online advertising uptake though progressively picking up, lags behind other developed economies that have seen much representation in extant literature. 1.3.3 Methodological Gaps The methodological gaps to be addressed are along two strands; the increased dominance of experimental studies, and high use of student samples. First, of the quantitative approaches, online advertising research is dominated by experimental (field and controlled) designs. While experiments provide the leeway to collect data unobtrusively and are free of self-reporting predisposition in real media atmospheres (Chang, 2017), laboratory experiments are plagued with external validity issues (Goldfarb, 2014), and the behaviour of respondents in laboratory experiments are considered falsified, given their knowledge of the research activity. It is also relevant to point out that in “normal” daily situations, unlike in controlled laboratory environments, people’s behaviours have a greater degree of variations, and such behaviours can be examined through observations or self-report data 11 University of Ghana http://ugspace.ug.edu.gh (Tang et al., 2014). More surveys and field experiments are, therefore, encouraged as they offer more uncontrived objectivity (Lewis & Rao, 2015). Particularly, since internet users who are exposed to online display advertisements may eventually engage in certain activities (e.g. offline purchase and behavioural intentions) that may not be observable in experimental situations, surveys offer more suitable approaches to measure such. Second, the essence of sampling in any research is generalisability, and this is even more so for the advertising field because of its closely-knit link with practice. Sampling has been an essential methodological problem ailing online advertising research (Ha, 2008; Knoll, 2015) as is reflected in the reliance on college students by the majority of studies in the reviewed ODA literature. Given the stream of evidence that variations exist in consumer attitude toward advertising in terms of certain demographics, it is essential that studies use diverse and broader respondent populations as possible, especially when the explanatory power of predictors are interacted with respondents’ characteristics (Chang, 2017; Eshghi et al., 2017). Albeit college students may constitute a sizable fraction of internet users and are an essential segment for advertising, they undoubtedly are not representative of all internet users on the bases of some demographics (e.g. age, education, outlook on technology, familiarity with the internet and usage rate etc.). This study will therefore focus on a broader population of online consumers beyond the borders of college students in order to curb validity issues and provide a more representative picture from the study setting. 1.4 RESEARCH QUESTION AND OBJECTIVES Following from the gaps raised with respect to issues, theories and methodologies, the study broadly seeks to examine how consumers respond to online display advertising by assessing the influence that specific ODA characteristics exert on these responses under different consumer-related conditions. Thus, the key question the study seeks to answer is What 12 University of Ghana http://ugspace.ug.edu.gh effects do ad characteristics, attitude toward online advertising and internet user mode have on consumer behavioural responses toward online display advertising of product and service brands? Specific objectives culminating into this broader intent are to: 1. Determine the relationship between online display advertising (ODA) characteristics and consumer behavioural responses. 2. Examine the intervening role of consumer attitude toward online advertising in the relationships between ODA characteristics and behavioural responses. 3. Assess the moderating effect of user mode on the relationships between ODA characteristics, and attitude toward online advertising as well as behavioural responses. 4. Explore the likely variations in consumer behavioural responses based on the nature (product vs service) of the advertised brand. 1.5 SIGNIFICANCE OF THE STUDY This study holds relevance along several strands for practical and theoretical developments. There is a lengthy debate in the advertising literature on the effectiveness of online advertising and current research on the issue has produced inconclusive results (Auschaitraku & Mucherjee, 2017). For this reason, there have been several academic calls for further research into both antecedents and measures of online display advertising effectiveness (Martin-santana & Beerli-Palacio 2012), as well as how to enhance user responses to online display ads (Bleier & Eisenbess, 2015b). This current study is thus, expected to provide theoretical and practical contributions which are briefly discussed next. Further detailed contributions from the thesis are provided in the concluding chapter (see pages 243-251). 13 University of Ghana http://ugspace.ug.edu.gh 1.5.1 Theoretical Contributions This thesis seeks to strengthen existing knowlwdge on advertising by enhancing understanding of the ODA features that are essential in driving consumer behavioural responses in online environments through the examination of some pertinent characteristics. More importantly, the study’s focus on a multiplicity of ad characteristics will help eliminate certain eccentric effects and provide deeper understanding of the individualised influence of the various characteristics to further enhance the online advertising effectiveness literature (Belanche et al., 2017; Breuer and Brettel, 2017). Also, because this thesis examines both the direction (acceptance and avoidance) as well as the intensity (active and passive) of behavioural responses, it will add further insight and improve the general understanding among academics and practitioners on how approach behaviours toward online display ads can be enhanced and avoidance behaviours minimised. Further, by relying on the SOR model and the reversal theory, the current study expands the borders and explanatory power of these theories from mainstream marketing into the online advertising context. What is more, these theories also allowed for the clear delineation of two consumer-related variables (attitude toward online advertising and internet user mode) which have not been adequately studied, and the possible integrative role they play with ODA features in engendering behavioural responses of consumers to ODA. The study also contributes to the existing literature on online advertising effectiveness by providing evidence from an emerging or developing middle-income country in sub-Saharan Africa. Contextual issues are essential in management and more specifically, marketing research (Boso et al., 2013), and given the roles cultural differences are known to play in consumer attitude and responses to advertising (Wang & Sun, 2010b; Lascu et al., 2016), providing this perspective will bring further enlightenment to the literature since evidence from the sub-Saharan African perspective are lacking. 14 University of Ghana http://ugspace.ug.edu.gh 1.5.2 Practical Contribution Practically, the study advances online advertising practice by offering insights on the applicability and suitability, as well as aid in subsequent selection of appropriate features in display advertising designs for more effective outcomes. More relevantly, given the apparent unlimited possibilities of ODA formats, insights into consumer perceptions and attitudes regarding, and behavioural responses toward their underlying features will guide practitioners in their future design, placement or execution and improvement of ODA. By this, the study offers insight on how practitioners can harness these factors to shape and enhance their online advertising goals.With the international scope of business activities today, and the cross-border nature of online advertising, this study provides insights to help advertisers and marketers in Ghana, those in settings with similar economic and customer characteristics as well as those in the international front seeking to promote their businesses and brands in such settings to fine-tune their online advertising efforts and ODA designs to enhance effectiveness and positive consumer reactions. The outcomes of this study will also aid practitioners (advertiser, internet publishers etc.) to understand better how to, more seamlessly, employ online media and related substructures to advance their advertising goals. 1.6 OUTLINE OF THE THESIS AND CHAPTER DISPOSITION This study comprises nine chapters. Chapter one introduces the study by providing the background and argument for this thesis, the research problem and gaps, as well as the research objectives. Chapter two focuses on the study context by highlighting some developments in internet usage and sheding light on the advertising sector and online advertising practices in Ghana. Chapter three discusses the theoretical perspectives that underpin this thesis. Specifically, the chapter reviews and discusses the SOR model and reversal theory in order to establish how this thesis may contribute to knowledge through these theories. Chapter four 15 University of Ghana http://ugspace.ug.edu.gh reviews extant literature in the study areas of online display advertising, discusses issues and evidences in ODA research under themes, and assess the theoretical and methodological approaches used in the reviewed ODA literature. This gives an understanding on the extent of research progression on this topic as well as aids in the identification of pertinent gaps in the literature. Following from this, chapter five incorporates the key issues arising from the review of literature and theoretical foundations into a conceptual framework that better explains the various relationships the study intends to examine in consort with corresponding testable hypotheses. The essence of this is to provide a graphical view of the study as well as guidance for the empirical investigation. Chapters six to nine form the empirical aspect of the thesis. Chapter six outlines the methodology for the study, thus discussing how the study was conducted in order to achieve the objectives outlined in the first chapter. The chapter examines methodological issues such as research paradigm, purpose, approach, and strategy. The chapter also reviews the instrumentation used for data gathering and the modes by which data was analysed. Chapter seven then reports the data analysis, presentation and interpretation of the results of the hypothesis tests. Chapter eight provides thorough discussions on these results/findings, based on the stated hypotheses and the study objectives. These discussions are done in relation to existing literature and the study’s context. Chapter nine which is the final chapter of the study covers five key areas. This chapter summarizes the study and findings, provides reflections, discusses implications for research and practice, draws conclusions and discusses limitations of the research, and points out avenues for further research. The structure of the entire thesis is illustrated in Figure 1.1 below. 16 University of Ghana http://ugspace.ug.edu.gh Figure 1.1 Structure and Flow of the Thesis Chapter One Background; Problem Statement & Research Gaps; Objectives and Study Context Chapter Two Context of the Study Chapter Three Theoretical Framework Stimulus Organism Response Model Reversal Theory Chapter Four Literature Review on Online Display Advertising Definitional Scope Key Themes in ODA Research: Theoretical and of ODA Issues and Evidences Methodological Approaches Chapter Five Conceptual Framework & Hypothesis Development Que stionnaire Design Chapter Six & Data Collection Research Methodology Chapter Seven Data Analysis and Presentation Chapter Eight Discussion of Findings Chapter Nine Summary, Conclusions & Implications 17 University of Ghana http://ugspace.ug.edu.gh 1.7 CHAPTER SUMMARY This chapter provides a general introduction to this thesis. The chapter puts forth the argument that the increasing growth of online advertising is attended by incessant emergence of ODA formats, increased advertising clutter, and consumers’ selective attention to online display ads. These challenges have heightened the need to understand the relevant ad-related and consumer-related factors that influence consumer behavioural responses to ODA so actionable indicators can be provided for eliciting positive consumer behavioural responses. Discussions within the chapter point out that the explicit examination of the executional characteristics of ODA, and the mediating and moderating role of attitudes toward online advertising, and user mode respectively as well as the nature of the advertised brand (product vs. service) will allow for a better understanding of how they influence both the direction (ad acceptance and ad avoidance) and intensity of consumers’ behavioural responses to online display advertising. The chapter concludes with a discussion on the significance of the thesis to practice and theory as well as how the entire thesis is structured. 18 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO CONTEXT OF THE STUDY 2.1 INTRODUCTION This chapter describes the setting within which the study was conducted in order to provide a contextual frame for the thesis. The chapter first, presents a brief overview of Ghana’s economy. Discussions on Internet usage in Ghana and the penetration rates over the years are presented. The chapter also highlights the current outlook of the advertising sector and throws light on the prevailing practices in online advertising in the country. 2.2 A BRIEF PRIMER ON GHANA’S ECONOMY Ghana is a sub-Saharan African (SSA) country geographically bordering Togo, Cote d’Ivore, and Burkina Faso, and is divided into sixteen (16) administrative regions, with the capital city, Accra, located in the Greater Accra Region. Ghana attained middle-income status in November 2010 and began commercial production of crude oil in 2011 (ISSER, 2017). The revenues that accrued in these years bumped the country to the global list of fastest growing developing economies and was ranked by the World Bank as the 2nd in West Africa and 85th among the world’s largest economies with a 40.7 billion US Dollars GDP in 2012 (World Bank, 2014). Over the past decade, the economy is thriving, and the country has experienced advances in business activities, political stability, and heightened foreign investments which have improved economic growth. Ghana’s economy keeps progressing and in the first quarter of 2019, the growth of gross domestic product (GDP) was estimated to be 6.7%, that is a 1.3% increment over the previous year (World Bank, 2019). This increased quarterly growth in 2019 was driven by a strong recovery in the services sector which grew by 7.2% relative to 1.2% in 2018. 19 University of Ghana http://ugspace.ug.edu.gh The economy is made up of three main sectors; Agriculture, Industry, and Services. The service sector is the largest of the three sectors and contributes about 54.2% to Ghana’s GDP, making it the most dominant contributor. Selected indicators of economic growth and sector contributions are presented in Table 2.1. Ghana Statistical Service (GSS) categorises the service sector of the country as comprising financial and insurance activities; real estate; professional activities; administrative and support service activities; trade; transport and storage; repair of vehicles; household goods; hotels and restaurants; public administration and defense; social security; education; health and social work; community, social and personal service activities, and information/media and communication. Information/media and Communication is one of the subsectors contributing to the service sector growth, and is the sector under which the study falls. Ghana enjoys a considerable degree of media freedom, with radio as the most popular medium followed by television, and new media, particularly the internet which is also beginning to gain a foothold. Table 2.1 Selected Indicators of Economic Growth in Ghana Indicators 1990 2000 2010 2017 Population (in millions) 14.60 18.94 24.51 30.21 Gross Domestic Product (in billion $) 402.59 263.11 1,312.61 42,689.78 Gross Domestic Product (annual growth in 3.33 3.70 7.90 3.58 %) Inflation (annual in %) 31.17 27.23 16.60 17.42 Foreign Direct Investment (in million $) 14.80 165.90 2,527.35 3,485.33 Sectorial Percentage Contribution to GDP Agriculture 45.07 39.41 30.83 19.60 Industry 38.08 32.20 49.36 52.24 Services 16.86 28.39 19.81 28.16 Imports 16.88 48.80 29.48 40.74 Exports 25.85 67.25 45.90 47.86 Source: World Bank, 2018 As at 2019, the population of Ghana was about 30 million, making it a sizeable emerging market in the SSA Region (World Bank, 2019). As one of the most densely populated African countries, Ghana has a huge consumer market made up of an urban population (55% 20 University of Ghana http://ugspace.ug.edu.gh of Ghanaians are urban dwellers) averagely higher than most African countries, and a young educated generation with literacy above 90% compared to a 76% literate adult population (Nordea, 2019). 2.3 INTERNET USAGE IN GHANA Internet usage rates are growing globally and an emerging middle-income country such as Ghana is no exception. Ghana was the second country within the sub-Saharan African Region to have complete internet connectivity in 1995, however, penetration did not improve speedily until the mid 2000s (Quarshie & Ami-Narh, 2012). Ever since, the country’s access to, and use of the internet has been increasing steadily, albeit the progress is below the global, regional and African average. Table 2.2 and Figure 2.1 show internet usage and penetration rates over the past decade. In Ghana although there are about 54 authorised internet service providers (NCA News, 2016), the telecommunications sub-sector is the largest provider of internet services, and the sector as at November 2017 was made up of 5 major operators (i.e. MTN, Vodafone, Airtel-Tigo, Glo and Expresso) who provide both voice and data services. These telecommunication companies with the exception of MTN (who provides 4G) operate with 3G mobile data (Citifmonline, 2016). Besides the telecom providers, other major internet service providers in the country include Surfline, Busy Ghana, and Blu Telecoms who also operate with 4G data. As at the end of the second quarter of 2019, internet users in Ghana amounted to 11,400,732 which has been credited to the proliferation of smart mobile phones, personal computers, and easy access to internet connectivity in the country (IWS, 2019). Internet penetration in Ghana in 2019 therefore stands at 37.9%, and averagely Ghanaians spend nearly four hours online using any device (Africa news, 2019). It has also been recounted by Graphic Online (2018) that of the devices used, Ghanaians preferred mobile phones in their internet usage 21 University of Ghana http://ugspace.ug.edu.gh activities compared to computers (laptops and desktops), and tablets. Specifically, 75% of internet traffic is said to come from mobile phones, 22% from PCs and 3% from tablet devices. Table 2.2 Internet Usage and Penetration Rate in Ghana Year Internet Users Population Penetration Usage Source 2009 997,000 23,887,812 4.2 % ITU 2010 1,297,000 24,339,838 5.3 % ITU 2011 2,085,501 24,791,073 8.4 % ITU 2012 2,707,724 25,544,565 10.6% ITU/IWS 2013 3,218,225 26,164,432 12.3 % ITU/IWS 2014 5,062,667 26,786,598 18.9 % ITU/IWS 2015 5,171,993 26,327,649 19.6% IWS 2016 7,958,675 28,206,728 29.6% IWS 2017 7,958,675 28,833,629 27.8% IWS 2018 10,110,000 29,463,643 33.6% IWS 2019 11,400,732 30,096,970 37.9% IWS Source: International Telecommunications Union and Internet World Statistics Figure 2.1 Population and Internet Usage Growth 35,000,000 30,000,000 25,000,000 20,000,000 Population Growth 15,000,000 Internet Users 10,000,000 5,000,000 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 With the substantial improvement in internet availability and mobile data services, several electronic commerce sites are shooting up that provide platforms for goods to be bought online and delivered to the doorsteps of customers. Although the electronic commerce market in Ghana is still in its early phases of development, it is more appealing compared to other West African countries (Nordea, 2019). Activities of the Pan-African online 22 University of Ghana http://ugspace.ug.edu.gh platform, Jumia, and the emergence of other online retail outfits such as Konga, Kaymu, and Tonaton among a host of others, coupled with Ghana’s Interbank Payment and Settlement Systems (GhIPSS) which kickstarted internet payment gateway, have also increased people’s participation in online purchasing, as well as information seeking regarding purchases (itnewsafrica.com, 2015). With the recent introduction of mobile money payments interoperability system in May 2018, and the full PayPal compliance to be established by 2020, electronic commerce activities in the country are expected to improve. In view of this, various companies also now have online presence, especially, websites and social media accounts, and utilise the internet for the transaction of basic business activities and promoting their brands. 2.4 THE ADVERTISING SECTOR IN GHANA The advertising sector in Ghana is a booming one as advertising is arguably, the most utilised marketing and communication tool employed in diverse sectors like education, politics, religion, fast moving consumer goods (e.g. electronic appliances and textiles), banking and insurance, real estate, agribusiness and hospitality (Nexus, 2013; The Report: Ghana 2012). Over the past two and a half decades, economic and media liberalisation has resulted in the growth of a highly competitive advertising industry with several new multinational and local agencies springing up (The Report: Ghana 2012) equipped with varying degrees of creative refinement and industry specialisation. Currently, there is no reliable database detailing the firms within the sector which makes it very challenging to know the number of registered agencies there are, as well as their specialties and focus. However, it is presumed that the sector has over 100 advertising agencies, and anecdotal accounts suggest that, as the general sector grows, and the internet penetration rate increases, more firms specialised in digital and internet advertising are joining the market. Currently there are about 10-15 renowned digital advertising agencies in Ghana (Digital 23 University of Ghana http://ugspace.ug.edu.gh content Africa, 2017). From a traditional media perspective, radio and television remain the most popular mediums for advertising, although, outdoor advertising is also heavily patronised with billboards ranking first in that category. In Ghana, there are no acts or governing bodies ordering and regulating advertising practice in the country, and so, advertising agencies operate without significant restrictions. An Advertising Council Bill which establishes the Advertising Council of Ghana, as the official government body charged with the duty of regulating advertising, registering practitioners in the industry and licensing advertising companies, has been presented to Parliament in 2016 and is yet to be passed into law. Presently, the Food and Drugs Authority is the formal body granted the mandate to vet and approve advertisements promoting pre-packaged foods, alcoholic and energy beverages. The Advertising Association of Ghana (AAG) is the industry body and professional institute in Ghana and is to a large extent the umbrella organisation representing agencies in the country. AAG is a non-profit organisation that represents the interests of advertising and marketing firms in the country and sets operational standards for the industry which signifies to a large extent, self-regulation. The AAG has as its members, eighty-two advertising agencies as at March 2020, and membership is voluntary. AAG is led by a thirteen-member executive council with three ex-officio members. As its aim and objectives, implemented through the Code of Professional Conduct and Ethics, the AAG seeks: o to promote public confidence in the Advertising profession. o to safeguard the common interests of those engaged in or using advertising for the promotion of common action and the institution of protective measures. o to encourage the study of the theory and practice of advertising, and the improvement of its techniques, by the institution of study, examination and award of certificates. 24 University of Ghana http://ugspace.ug.edu.gh o to establish that efficient advertising in an essential factor in the marketing of goods and services, and in the economic life of the country. o to demonstrate the efficiency of the services, that advertising and its associated interests can give to government, industry and the public. o to further the adoption of standards or practice in the business relations between media owners, advertising agencies and advertisers. 2.5 ONLINE ADVERTISING IN GHANA Consistent with the growth of internet usership, and the advertising sector, online advertising in Ghana is growing but trails behind developed settings and even other developing sub-Saharan African countries like South Africa and Nigeria. In spite of this, its forms have evolved from websites and emails to social media, and display ads which seem to be showing high potential. As the internet usage and penetration rates, the explosion of online retail outlets, and internet payment systems provide indications of the potential of the internet as a vital medium to reach and communicate with consumers, online advertising expenditures in the country are growing. Particularly, given the growing mobile device (e.g. smart phones and tablets) ownership and usage in the country, and mobile data connectivity, marketers are making a concerted effort to reach their target customers while they go about their daily activities (Graphic Online, 2018). Presently, most firms in Ghana are spending between 8-12% of their budget on digital advertising and the growing trend in digital spending on display ads and social media is led by international brands and telecom operators (Digital Content Africa, 2017). The Figure 2.2 below shows the internet advertising revenue in Ghana from 2013 to 2018, with future estimations up to 2023. According to Statista (2019), total internet advertising revenue in Ghana is projected to grow from 15 to 29 million U.S. dollars by 2023. The expected increase in revenue growth, 25 University of Ghana http://ugspace.ug.edu.gh provides indications of corresponding likely surge in online advertising spending among firms in the coming years. Figure 2.2 Internet Advertising Revenue 29 30 27 25 25 22 20 18 15 15 12 9.4 10 7.1 4.6 5.3 5 0 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Source: Statista (2019) Owing to the borderless and pervasive nature of the internet, online advertising campaign design is not the sole domain of ad agencies. Online ad maker software and applications (e.g. bannersnacks) allow individuals to design ads ranging from simple banners to video ads. Also, besides recognised large publishers, and social media sites (such as Facebook, Linkedin, Instagram and YouTube) informational sites in Ghana such as online news portals (e.g. Myjoyonline, Citifmonline, Yen.com.gh etc.), blogs (e.g. AmeyawDebrah.com, GhBase.com) and commercial websites (e.g. jumia.com.gh, Tonaton.com etc.) host display ads of various product and service brands from within and outside the country. These advertised brands range from automobile, fashion (e.g. apparel, footwear, sportswear, accessories, cosmetics, textiles, and watches etc.), electronics (e.g. computers, mobile phones, televisions etc.), education, hospitality (e.g. lodging, restaurants/food, events etc), and auto repair among others. 26 Revenue in million U.S. Dollars University of Ghana http://ugspace.ug.edu.gh However, unlike traditional advertising, because online advertising is still taking shape in the country, very little is known from an academic perspective about its uptake among practitioners, how Ghanaian consumers perceive and respond to, as well as form attitude toward online advertising in general. While several studies have been conducted in developed countries on the subject, findings from these studies cannot be generalised to developing settings. Particularly on the basis that advertising is considered a social actor and a cultural artefact that conveys socio-cultural values and beliefs (Frith, 1995), individuals’ attitude and responses to advertising and their online behaviours are context- dependent and vary extensively by country (Wang & Sun, 2010b; De Mooji & Hofstede, 2010). For instance, according to Nordea (2019) compared to other African countries, Ghanaian consumers are more receptive to advertisement and show stronger interest in promotions and are also more willing to try new brands. Given that online advertising is a relatively budding phenomenon in Ghana, and consumers are often exposed to online display ads from within and outside the country for diverse categories of product and service brands, it would be enlightening to understand what issues impact consumer behavioural responses toward ODA in the Ghanaian market place, and how marketers and advertisers can better strategise their advertising designs. 2.6 CHAPTER SUMMARY The chapter discussed the context in which the study was conducted by providing a brief overview of Ghana’s profile and economic development over the years. The chapter discusses Internet usage and penetration trends in the country, and presents a synopsis of the advertising sector, throwing light on the lack of regulatory framework to order activities within the sector. Discussions on prevailing online advertising activities are also discussed in order to situate the study. The next chapter presents the theoretical framework that underpins the study. 27 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE THEORETICAL FRAMEWORK 3.1 CHAPTER OVERVIEW Following from the previous chapters which present the research problem, gaps and objectives as well as the context of the study, this chapter discusses the theoretical perspectives that underpin this current study. The review of extant ODA literature presented in the next chapter (see chapter three) unearthed five theories that stood out as used by researchers to provide understanding on key issues in online display advertising. These theories include the elaboration likelihood model (ELM) by Petty et al. (1983), reactance theory by Brehm (1966), mere exposure effect theory by Zajonc (1968), stimulus organism response model (SOR) by Mehrabian and Russell (1974), and reversal theory by Apter (1984). After a careful consideration of these theories, which all offer some perspectives and insights on ad-related and/or consumer-related issues and how these issues affect online display advertising outcomes, the study chose the stimulus organism response (SOR) model and the reversal theory as the main theories to underpin this current study. These two theories as opposed to others that emerged from the literature review, complement each other in examining how relevant ODA characteristics, influence consumer behavioural responses to online advertising under different consumer-related situations but have sparsely been applied in ODA research. By so doing, discussions in the chapter help to theoretically position the study in light of its contribution to knowledge. In line with this, the chapter is organized as follows; introduction and justification for the chosen theories are presented in section two and section three respectively. Sections four and five then provide an overview of the underpinning theories, their assumptions and elements, applications, as well as their relevance to this thesis. Section six briefly discussed 28 University of Ghana http://ugspace.ug.edu.gh the three other theories that emerged from the review in order to shed some light on their use in ODA research and the insights they could provide into consumer responses toward ODA as well as justifications for not using them as the main theoretical foundation for this study. A summary of the chapter is then presented in section seven. 3.2 INTRODUCTION Theories are well-established principles that are used to explain, forecast, and understand occurrences and, in most instances, to challenge and extend prevailing knowledge within the limits of critical bounding assumptions (Abend, 2008). They are considered statements of constructs and their interconnections that depict the ‘how and why’ of an occurring phenomenon (Gioia & Pitre, 1990). For Miller (2005), “they are the nets with which we catch the world, or the ways in which we make sense of social life” (p.22). Theories are known to help researchers organize their thoughts to achieve better understanding of a given subject matter; hence, they form the footing on which most scholarly research works are grounded (Hambrick, 2007; Corley & Gioia, 2011). It is argued by Cai and Mehrai (2015) that any research work not guided by theory is deficient in interpreting and understanding the observed phenomenon. For this reason, discussing the theoretical parameters within which this thesis is conducted is essential as it helps define the specific viewpoints that the study takes in examining the overarching research question that need to be answered to increase understanding of the study as well as analysing and interpreting the data gathered in order to provide answers to the stated research objectives. For this reason, the subsequent section provides a rationalisation for the choice of SOR model and the reversal theory that are fundamental to the study’s aim of understanding consumer behavioural responses to online display advertising in Ghana. 29 University of Ghana http://ugspace.ug.edu.gh 3.3 JUSTIFICATION FOR THE UNDERPINNING THEORIES Given the multidisciplinary nature of advertising, several theories from psychology, communications, advertising as well as marketing have been applied in studying online display advertising and its effects. In spite of this, compared to atheoretical approaches, there is still relatively low theory-based approaches to understanding the effects of online advertising. Evidence from both practice and the ODA literature suggests that owing to the high advertising density of the internet, advertisers are struggling to stand out among the clutter, to capture consumer attention as well as elicit favourable responses from consumers (Nihel, 2013; Hoban et al., 2015). On account of this, the current study seeks a theory-based and practical understanding of consumer behavioural responses to online advertising, which the study proposes may be influenced by both ad-related and consumer-related factors. ODA features or characteristics have been identified as essential factors that drive the effectiveness of display ads and influence online advertising responses (Brajnik & Gabrielli, 2010; Tang et al., 2014). Through these features, advertisiers are able to get consumers to be attentive to, and receptive toward online ads (Kuisma et al., 2010; Wang et al., 2013; Bruce et al., 2016). Also, attitude toward online advertising has been pointed out as an essential consumer-related factor that influences consumers’ reactions to display ads and is considered a major determinant of advertising efficiency. (Zha et al., 2015; Shaouf et al., 2016; Souiden et al., 2017). What is more, the task/goal orientation of consumers are known to affect how they respond to persuasive communication (Wang et al., 2013). Because internet users’ online activities typically start with a task-oriented plan, user mode which depicts the degree to which internet usage is goal-directed is suggested to exert a certain degree of influence on the attention alloted, and/or interaction users have with online ads (Simola et al., 2011; Kim et al., 2014; Seyedghorban et al., 2016). 30 University of Ghana http://ugspace.ug.edu.gh As mentioned before, since the focus of this thesis is to examine consumer behavioural responses to online display advertising by enhancing understanding of how consumer attitude toward online adveritising and their user mode enhance or mitigate the influence of ODA characteristics on behavioural responses, it became evident that a single theory may not sufficiently and coherently explain the interconnections the study seeks to explore (see Figure 5.1). The reasoning is that, in spite of the SOR model providing support for the stimuli function of ODA characteristics, and asserting that internal states of consumers intervene between the stimuli and behavioural responses, the theory is deficient in offering explications on how other consumer-related factors cause variations in behavioural responses. Alternately stated, the SOR model is used to explain how ODA characteristics represent essential stimuli that advertisers use to make their display ads stand out among the numerous online ads to which consumers are exposed. The model is also used to highlight the role of attitude toward online advertising as a facilitating/intervening factor that transmits the effect of the ODA characteristics on behavioural responses of consumers. However, the SOR model does not provide detailed accounts on how user mode may vary the direction and magnitude of the relationship between the stimuli and behavioural responses, thus necessitating input from another theory to offer understanding of situations in which ODA features may generate varying degrees of positive or negative consumer responses. And so, in addition to the SOR model which helps in explaining the link between the online ad characteristics, the mediating variable (attitude toward online advertising), and the outcome variables (behavioural responses), the study as well draws on a second theory, reversal theory, to help provide possible explanations to the contingencies of user mode and the variations it may cause in the baseline (ODA characteristics-behavioural response and ODA characteristics-attitude toward online advertising) relationships. The complementary 31 University of Ghana http://ugspace.ug.edu.gh use of these theories in this study provides an avenue to comprehensively examine: how consumers perceive specific ODA characteristics and how these perceptions influence their behavioural responses; how attitude toward online advertising intervenes between the ODA characteristic-behavioural responses relationship; and how their goal directedness (user mode) may vary the direction and extent of these behavioural responses. In line with the aim of this research, the SOR model in consort with the reversal theory is reasonably the most appropriate and effective theoretical framework to explore the interrelationship among the variables in this thesis, and the key hypothesis postulated. Particularly because, according to Saunders et al. (2012), theories should not just comprise cause and effect components to examine fundamental variables but should as well be useful in describing the nature of the relationships among variables and provide rational explanations for the existence of such relationships. The next section discusses the SOR model. 3.4 STIMULUS ORGANISM RESPONSE (SOR) MODEL 3.4.1 Overview of the Theory The Stimulus Organism Response (SOR) model as conceptualised by Mehrabian and Russell (1974) and later modified by Jacoby (2002), outlines how external stimuli affect people’s internal state and subsequently their behaviour. The model which has its foundation in environmental psychology, proposes that cues or features in an environmental setting incite a person’s emotional and cognitive conditions leading to certain behavioural outcomes (Donovan & Rositer, 1982). According to the SOR model, stimuli (S) in any surrounding may cause changes in an individual’s internal/organism states (O), which then leads to approach or avoidance behavioural responses (R) (Mehrabian & Russell, 1974). Stimulus within the borders of the theory are factors that influence an individual’s internal state and stimulate the individual (Eroglu et al., 2001). Organism denotes internal processes and structures that intervene between the stimuli and the final reactions produced by the 32 University of Ghana http://ugspace.ug.edu.gh individual (Chang et al., 2011). These intervening processes come in the form of perceptions, physical, emotional and cognitive activities, and the original model focused on pleasure, arousal, and dominance (PAD) (Chang et al., 2011). Then, the ultimate outcome or the conclusive decision of the consumer which may be approach or avoidance behaviours signify ‘response’ in the S-O-R framework (Sherman et al., 1997). Essentially, the theory hinges on the premise that the relationship between any environmental stimuli and responsive behaviour is intervened by the internal processes of the organism exposed to such stimuli (Chen & Yao, 2018). Since its introduction into marketing by Donovan and Rossiter (1982), extant marketing literature depicts a considerable use of the theory in the past years and researchers have broadened the array of circumstances in which the said interrelationships may occur (Islam & Rahman, 2017; Kamboj et al., 2018). Stated alternately, the SOR model has been noticeably used to study several phenomena, and marketing literature has validated its applicability in predicting consumer behaviour in several settings. These studies have examined and extended the model to impulse buying behaviours (Chen & Yao, 2018), website experience (Mollen & Wilson, 2010; Kim & Lenon, 2012), online brand communities in social media (Islam & Rahman, 2017; Kamboj et al., 2018), consumption experiences in m-commerce (Li et al., 2012), tourist motivations (Rajaguru, 2013), as well as offline and online retail settings (Chang et al., 2011; Goi et al., 2014). With internet penetration and online consumption or purchasing behaviour surging, SOR has become the prevalent theoretical model in the impulse buying literature (Chan et al., 2017) and has seen extensive use in the retail literature (Chen & Yao, 2018). The key notion of studies underpinned by the model lie in the submissions that design characteristics, features of websites as well as online brand community characteristics influence consumers emotional and cognitive states and subsequent behaviours either negatively or positively. In the 33 University of Ghana http://ugspace.ug.edu.gh marketing and consumer behaviour literature, it is contended that stimuli that consumers are exposed to may be environmental factors or any of the marketing mix variables (Bagozzi 1986; Manganari et al., 2009; Kim & Lennon, 2012). 3.4.2 Components of the Model The key concepts and elements in the SOR model have been explained and illustrated by several researchers (Jacoby, 2002; Vieira, 2013). As reflected in the name these elements include the stimulus, organism, and the response. The Figure 3.1 provides a graphical view of the original SOR model and its primary elements are successively discussed. Figure 3.1 SOR Model Primary Emotional The Environment (S) Responses (O) Behavioral Responses Approach-Avoidance (R) Se nse modality variables (e.g., colour and Temperature) • Pleasure • Arousal (Which includes a physical • Dominance approach, exploration, Information rate affiliation, performance, or (C haracterizing the spatial and other verbal and non-verbal temporal relationship among the communications of stimulus components of an Characteristic preference) environment) emotions associated with personality Source: Mehrabian and Russell (1974, p.8) 3.4.2.1 Stimulus The SOR model has at its core, the idea that stimuli are cues or any environmental factor that influences or arouses an individual (Eroglu et al., 2001). Such stimuli are projected to be features or qualities of the stimulus object capable of generating some organism response (Chang et al., 2011). Mehrabian and Russell (1974) at the initial development of the model proposed a general measure of environmental stimulation which could be applied within various physical and social settings to describe the information rate or load of the environment, and was reflected in dimensions such as complexity, unity, diversity, 34 University of Ghana http://ugspace.ug.edu.gh congruity, artificiality, crowding, symmetry, novelty and meaningfulness. Following this perspective, researchers ascertain that the attractiveness of any stimulus is dependent on its information load (Steenkamp & Baumgartner, 1992). Within the context of consumer behaviour, however, Bagozzi (1986) and other researchers assert that these stimuli are external to consumers and may comprise marketing mix elements along with other environmental factors. According to Goi et al. (2014), it is the onus of markets, therefore, to custom and project suitable information, actions and cues that effectively affect the internal (cognitive and emotional) state of customers in order to facilitate positive responses from them. Consequently, extant studies have examined diverse collections of characteristics as stimuli in various settings and acknowledge that ambient factors, design factors and social factors are essential stimuli in online environments (see Table 3.1). 3.4.2.2 Organism The SOR model as well suggests that consumers respond to stimuli in two stages. The first is an internal response, referred to as “organism response” and the other is an external response referred to as “behavioural response” (Chen & Yao, 2018). Chang and Chen (2008) define organism as, “cognitive and affective intermediary states and processes that mediate the relationships between the stimulus and the individual’s response” (p.820). As the second element in the model, these internal responses are integral to enhancing behavioural responses, and are directed towards altering the connection between the organism and the stimulus object (Eroglu, 2003). It has been suggested by the original proponents of the model that the organism response mirrors the emotional states of individuals and are clustered in three independent dimensions namely: pleasure, arousal, and dominance (PAD). “Arousal-non-arousal refers to the degree to which an individual feel, excited, stimulated, alert or active in the situation; pleasure-displeasure refers to the degree to which an individual feels good, joyful, happy, or satisfied in the situation; and finally, dominance- 35 University of Ghana http://ugspace.ug.edu.gh submissiveness refers to the degree to which an individual feels control over or free to act in the situation.” While Donovan and Rossiter (1982) reasoning along the lines of the original proponents, submit that people’s response to a stimulus is intervened by their emotional reaction to that stimulus, other perspectives (e.g. Mazaheri et al., 2011) suggest that response to stimulus is based on cognitive processes such as rational evaluation of informational cues. These two contrasting but complementary views provide a two-pronged perspective to organism response which supports Jacoby’s (2002) earlier position that emotions, attitudes and beliefs are all part of the organism element of the model. Particularly, studies that apply the framework to research in online settings argue that the online environment is a “cognitive landscape” typified by high levels of cognitive stimuli (Demangeot & Broderick, 2007), and as such studies in online environments should consider the bi-lateral view of organism response. In recent streams of studies underpinned by the SOR model (Kim & Lennon, 2012), the organism is represented by affective and cognitive intermediary states and processes that mediate the association between stimuli and responses, and others have also argued that a specific affective reaction satisfies the condition of organism response (Chang et al., 2011). 3.4.2.3 Response Response is the last element of the model and reflects behavioural reactions or the final action toward the stimulus object driven by the mediating organism response (Sherman et al., 1997; Eroglu et al., 2001). The response component of the model has been categorized into approach avoidance behaviours based on the idea that people’s behavioural reactions generally occur in either one or two of these directions (Chang & Chen, 2008;). Per the 36 University of Ghana http://ugspace.ug.edu.gh model, the level of internal changes experienced by individuals exposed to a stimulus will determine their approach-avoidance response (Vieira, 2013). Approach and Avoidance Behaviours From the early viewpoint of Donovan and Rossiter (1982), behavioural responses toward a stimulus setting show four features representative of approach or avoidance predispositions: “(1) a physical desire to stay in the presence of or remove oneself from the presence of the stimulus object; (2) a willingness to dedicate resources to enhance the relationship between the organism and the stimulus object or a tendency to avoid interaction between the organism or individual and stimulus object; (3) a willingness to engage with other stimuli associated with the stimulus object or a tendency to avoid additional stimuli related to the stimulus object and; (4) an elevation or reduction of personal satisfaction when engaged in stimulus-related task completion.” Thus, on the one hand, approach behaviours concern all favourable behaviours directed at the stimulus of interest (Rajaguru, 2013). These affirmative behaviours toward the stimuli may manifest in several forms such as a longing to remain within the stimulus environment, explore and connect with the stimulus object (Donovan et al., 1994). From empirical marketing perspectives, other suggestions have pointed to amount of time spent within the environment (Mummalaneni, 2005), purchase within the environment (Chang & Chen, 2008; Chang et al., 2011; Chen & Yao, 2018) satisfaction with the environment (Kim & Lennon, 2012) as well as a yearning to return to the environment (Jeong et al., 2009; Kamboj et al., 2018) as forms of approach behaviours. On the other hand, avoidance behaviours within the parameters of the SOR model are actions contrary to approach behaviours (Tang et al., 2014). Avoidance behaviours describe 37 University of Ghana http://ugspace.ug.edu.gh the absence of interest or desire that the consumer may display toward the stimulus (Sweeney et al., 2002). Researchers have espoused other explanations and manifestations of avoidance behaviours to include lack of intention to revisit the stimulus setting (Spangenberg et al., 2006); an inclination to move away from the stimulus object, and to indulge in other activities (Kim & Moon, 2009). 3.4.3 Applications and Relevance of the Theory to this Thesis Although the SOR model has been used extensively in the online context of consumer behaviour, the focus has been on retailing, online impulse purchasing (Islam & Rahman, 2017) as well as other internet-related behaviours. For instance, Kim and Lennon (2012) examined the effects of reputation and website quality on online consumers’ emotion, perceived risk and purchase intentions. In their study, website quality dimensions and reputation were considered stimuli that affected consumers emotions and perceptions of risk which influenced their purchase intentions. In a similar study, Sheng and Joqinapelly (2012) investigate the influence of web atmospheric cues on internet users’ emotional responses and purchasing intention in an e-commerce setting. Their study revealed that interactivity and vividness positively influence users’ valence and arousal states. Also, Kamboj et al. (2018) assessed branding co-creation in brand communities on social media and found brand trust as a mediator between customer participation and brand loyalty, trust and co- creation. Aside these online studies, the SOR model has also been employed in studies investigating the purchase behaviour of high technological products. Lee et al. (2011) in their study for instance found that stimuli such as innovativeness of technology, visual appeal. Prototypicality, and self-expression significantly influence behaviour as mediated by attitude and pleasure. Extant literature abounds in empirical works underpinned by the SOR model in the online retailing, and online consumer behaviour context, and Table 3.1 provides a summary of a selected few of these applications. 38 University of Ghana http://ugspace.ug.edu.gh “Table 3.1 Summary of Selected Research Articles Underpinned by the SOR Model Study Stimulus Variables Organism Variables Responses Variables Jeong et al. (2009) -Product Presentation Features -Four Experience Realms -Website Patronage o Entertainment Intention o Educational o Escapist o Esthetic Consumer Emotion o Pleasure o Arousal Koo & Ju (2010) -Graphics -Pleasure -Intention -Colours -Arousal -Links -Menus Wang et al. (2010) -Perceived Web Aesthetics -Satisfaction -Purchase o Aesthetic Formality -Online service quality -Re-purchase o Aesthetic Appeal -Arousal -Loyalty -Complaint -Service switch Chang et al., -Ambient characteristics -Positive emotional -Impulse buying (2011) -Design characteristics responses behaviour -Social characteristics Lam et al. -Ambience -Customer Satisfaction -Desire to Stay (2011) -Navigation -Cognitive -Intention to revisit -Seating Comfort -Affective -Interior décor Wong et al. (2012) -Mall/store quality ---- --- --- -Shopping enjoyment -Quality of merchandise -Convenience -Enhancements -Price orientation Kim & Lennon -Reputation -Emotion -Purchase intention (2012) -Website quality -Perceived risk o Website design o Fulfillment/Reliability o Customer service o Security/Privacy Dong & Siu -Substantive staging -Service experience -Experience (2013) o Background functional evaluation intensification -Communicative Staging -Experience extension o Employee behavior o Employee image o Cultural atmospherics Rajaguru (2013) -Visual effect -Tourism intentions Visitation -Vocal effect -Celebrity effect Goi et al. (2014) -Exterior -Experience -Cognitive -General interior -Mood -Affective -Store layout -Emotions -Behaviour -Interior displays -Human variable -Value *Tang et al. -Online design features ---- --- --- -Approach behaviour (2014)* o Ad content (active & passive) 39 University of Ghana http://ugspace.ug.edu.gh o Ad form Avoidance behaviour o Ad action (active & passive) *Bleier & -Ad personalisation Usefulness Click- Through Eisenbess o Depth Reactance (2015b)* o Breadth Privacy concerns Islam & Rahman -Online brand community -Customer engagement -Brand loyalty (2017) characteristics o Information quality o System quality o Virtual interactivity o Rewards Chen & Yao -Website architectural quality -Impulse buying -Impulse buying (2018) o Ubiquity tendencies behaviour o Ease of use -Normative evaluation o Information exchange -Positive affect -Promotional campaigns o Discount o Scarcity Kamboj et al. -SNSs participation motivations -Customer participation -Brand trust (2018) in Brand communities on -Brand loyalty SNSs -Branding co-creation Nunthiphatprueksa -Intrinsic quality -Destination Image -Behavioural response & Suntrayuth -Contextual quality (2018) -Representatinal quality -Social quality Note: Studies on online display advertising are in asterisks (*)” The SOR model is espoused to be applicable in a wide variety of settings, and so, this current study applies it to the ODA context. The focus of this thesis is to examine the effects of an array of online display advertising characteristics as essential stimuli in online advertising and how they influence consumer attitude toward online advertising and behavioural responses of consumers who are exposed to such advertisements. From a general advertising perspective, advertisers anticipate that following exposure to advertising stimuli, consumers will reckon the likelihood of advertised brands to satisfy their needs. From this perspective one would expect a considerable amount of advertising studies to employ the SOR model and extend arguments as well as assertions into the advertising or online advertising domain. However, out of the 63 empirical papers reviewed on online display advertising (see Appendix A2) and presented in the next chapter, the study identified only two studies (i.e. Tang et al., 2014; Bleier & Eisenbess, 2015) underpinned by the SOR model. 40 University of Ghana http://ugspace.ug.edu.gh In their study, Bleier and Eisenbess (2015b) investigated how trust moderates the effect of ad personalisation on consumers internal and external responses. Using a two-dimensional (breadth and depth) conceptualisation of personalisation, and usefulness, reactance and privacy concerns as internal responses, the researchers demonstrated that “more trusted retailers can heighten the perceived usefulness of their ads through a combination of high depth and narrow breadth of personalisation without eliciting increased reactance or privacy concerns. On the other hand, for less trusted retailers, banners with higher depth are not perceived more useful but instead trigger increase reactane and privacy concerns regardless of their personalisation breadth. These effects directly translate into consumers’ click- through intentions.” Also, quite recently, Tang et al. (2014) expanded the approach- avoidance dimensions of the model through an exploratory study in the online advertising context. The authors argued that prior studies underpinned by the SOR have mostly focused on the direction of the behavioural response, thus overlooking the intensity of the behavioural response or efforts. The study then combined the behavioural direction and intensity into a two-dimensional framework. The said online behavioural responses are summarised into four main dimensions: “active approach, passive approach, active avoidance and passive avoidance”, and behavioural responses toward online ads may be one of the four types. This categorisation of course, requires further research given that the intensity level of behavioural responses have received sparse research attention. The internet is considered a stimulus-rich environment with several formats of online display ads struggling to capture consumer attention. Online display advertisements unlike their search counterparts are push-based, and their executional characteristics have been posited to elicit consumer responses and influence consumer reactions or behaviours toward ads as well as the products/services and brands advertised (Goodrich 2011; Goodrich et al., 2015). Since decisions regarding interactivity, appeal, animation, placement, 41 University of Ghana http://ugspace.ug.edu.gh personalisation, and exposure conditions among others, convey both informative and persuasive intents of the advertiser, and are implemented to indicate high-quality ads that may attract consumer attention, researchers are focusing more on how these executional or ad-related features influence consumer perceptions, attitudes and behaviour (Bright & Daugherty, 2012; Goodrich et al., 2015). For instance, van Doorn and Hoekstra (2013) examined the effects of personalisation on perceptions of intrusiveness as well as purchase intentions. The results of their study signify that, while high degrees of personalisation in the form of using names and transaction information increase perceptions of intrusiveness, and influence purchase intention negatively, these negative influences are lessened if the ad is tailored to fit the consumer’s present needs. The studies that have considered these executional features have pointed to a paucity of research works to answer the question of which of these features may serve as useful or effective stimuli in the domain of online display advertising. Discussions so far, have pointed out advances made by marketing researchers toward understanding the dependencies espoused in the SOR model leading to behavioural responses. Nonetheless researchers have proffered some suggestions for future studies in order to extend the application of the theory to other marketing phenomena. Some of these suggestions are briefly reflected on in the ensuing paragraphs as they pertain to this thesis. To begin with researchers in the domain of the SOR model have articulated the need for future studies employing the model to examine diverse categories of stimuli as used by firms in various settings (research and industry) particularly in the online context using various online applications (Kim & Lennon, 2012; Islam & Rahman, 2017; Chen & Yao, 2018). Though the theory has been applied in the online context, its application has been biased toward retail and consumer behaviour in such settings (see Viera, 2013). In applying the theory to online advertising, Tang et al. (2014) noted that future research into what specific 42 University of Ghana http://ugspace.ug.edu.gh design elements/features of online display advertising may generate both approach and avoidance behaviours from consumers was needed. Other SOR researchers also point to the need for future studies to assess how consumers meaningfully combine stimuli and then assign meanings to these combinations which may results in diverse internal (organism) and external (behavioural) responses for various consumers based on these groupings (Kim & Lennon, 2012). According to Islam and Rahman (2017), exploring and examining the different types of stimuli used by firms, could spur academic enquiries into how these assortments can be managed in order to amplify their integrated efficacy. Again, SOR researchers have called for future research that would investigate how various stimuli interact with one another and concurrently influence consumers’ perceptions and subsequent behaviours (Chang et al., 2011) as well as explore the contingencies of several consumer-related variables that may cause divergences in their perceptions, internal and external responses (Goi et al., 2014; Bleier & Esenbeiss, 2015b). On the basis of these discussions, and in line with the aim of this study, the SOR model is applied in explaining the influence of ODA characteristics on consumer behavioural responses. Consistent with the model, the study examines the effect of specific ODA characteristics (interactivity, informativeness, placement, personalisation, and exposure condition) considered to be the stimuli, on consumer attitude toward online advertising as the organism response or mediator, and ad acceptance and ad avoidance as behavioural responses. Quite importantly, which ODA characteristics may serve as relatively useful stimuli, and the extent to which consumer attitude toward online advertsing may play an intervening (organism) role between such stumli and the behavioural response of consumers are questions this current study seeks to answer. This thesis theorises and contends that to the extent that the effectiveness of online display ads depend on consumers’ exposure to the 43 University of Ghana http://ugspace.ug.edu.gh ad message content and ad-related features (ODA characteristics) such as interactivity, placement, personalisation, informativeness, and exposure conditions which serve as mechanisms by which awareness is raised and product quality is indicated in an attempt to influence consumer behaviour, then they provide significant stimulation for certain behavioural responses in the context of online display advertising. Also, in line with the ODA literature, attitude toward online advertising is an essential affective response (Souiden et al., 2017) that could play a facilitating role between the ODA stimuli and the behaviours evoked by such stimuli because attitude is a vital internal process that preceeds behaviour (Wang & Sun, 2010b). What is more, in the online advertising contexts, advertisiers face the likelihood of consumers showing both approach and avoidance behaviours toward their display ads owing to consumers favourable or unfavaourable perceptions of such ads (Wang & Sun, 2010b). As a result, attitude toward online advertising is operationalised as an organism/internal response which can enhance the stimuli effect of ODA characteristics on behavioural (ad acceptance and ad avoidance) responses also operationalise as approach and avoidance behaviours. The thesis further introduces a consumer-related factor, ‘user mode’, as the moderator in the relationship between ODA characteristics and attitude toward online advertising as well as behavioural responses, and the theoretical rationale for this is discussed next. 3.5 REVERSAL THEORY 3.5.1 Overview of the Theory Reversal theory focuses on people’s motivation, emotion, and dispositions based on an analysis of daily experiences (Apter, 1984; 2001). The theory suggests that the things individuals experience and their responses to these occurrences are “shaped by a set of alternative ways of seeing the world” (Apter, 2007). The fundamental assumption of the 44 University of Ghana http://ugspace.ug.edu.gh theory is that every individual is a different sort of person at different times, and so, their dispositions alter during the course of day-to-day life (Portell & Mullet, 2014). Proponents of the theory aver that the subjective quality of any activity is determined by a person’s current frame of mind referred to as meta-motivational state within the boundaries of the theory. The theory postulates eight meta-motivational states that are categorised into four contradictory pairs of domains, with each pair denoting opposite ways of experiencing activities in some basic way (Piotrowski, 2011). These pairs of meta-motivational states comprise: o Conformist (rule-abiding) vs. Negativistic (rebellious) o Mastery (domination-oriented) vs. Sympathetic (relationship-oriented) o Autic (concern for self) vs. Alloic (concern for others) o Telic (serious minded, goal-oriented) vs. Paratelic (playful, process-oriented) Reversal theorists advance that people will “reverse”, or switch between these contrasting states in the course of an activity or their daily lives, and only one state in each pair can be active at a time (Apter, 2001; Mackenzie et al., 2011). Although all eight meta-motivational states are important, the telic and paratelic states are essential for explaining variances in consumer behavioural responses to online display advertising (Seyedghorban et al., 2016), and so are the focus of this thesis. This pair of meta-motivational state is discussed further in the next section. 3.5.2 Telic versus Paratelic Perspective of Reversal Theory Given that reversal theory explicates the complexity and variability of consumers’ cognitive experience at diverse points and times (Jung et al., 2014), the telic/paratelic state is considered a “means-end” domain because individuals in these states are either focused on how they can enjoy themselves (the ‘means’) or achieve a goal (the ‘ends’) (Apter, 2001). The rationale Apter (2007) provides is that, individuals in a telic state focus on future goals, 45 University of Ghana http://ugspace.ug.edu.gh and those in a paratelic state focus on the present moment and the specific activity they are engaged in instead of the goal. The telic state is one in which the goal is experienced as being paramount, and activities are simply means towards that end. The paratelic state, however, is one in which the ongoing activity is experienced to be of vital significance, and goals are there to merely make the activity more interesting (Wright et al., 2012). Essentially, the point of view of the telic state is serious and that of the paratelic state is playful (Jung et al., 2014). Extant literature on reversal theory therefore classifies the telic state as typified by serious-mindedness, goal and future-orientedness, planning ahead, arousal-avoidance, preference for important activities and focus on activity/task completions (Apter, 2007). The paratelic state is also characterized by values such as spontaneity, arousal-seeking, sensation-orientedness, playful-mindedness, present- orientedness, preference for trivial activities, and attempting to prolong an activity (Fruchart et al., 2018). An essential feature that separates this pair of meta-motivational states is known in reversal theory as a “protective frame” (Apter, 1992; Kerr, 2005). This experiential frame, when present, causes an individual to feel cut off from significant consequences. This is one of the aspects of playfulness and activity-orientation and is, therefore, a defining feature of the paratelic state. Conversely, the goal-oriented telic state is aware of consequences, including the serious consequences of things going wrong owing to a lack of the protective frame. Accordingly, within the context of reversal theory, irritating situations may be perceived as unpleasant in the telic state, and as pleasant in the paratelic state, because of the protective frame (Portell & Mullet, 2014). 46 University of Ghana http://ugspace.ug.edu.gh 3.5.3 Applications and Relevance of the Theory to this Thesis Following its original proposition in the mid 1970s, reversal theory has been applied in diverse fields, and several research works have employed the theory to explain the complex behaviours of people in the areas of sports (e.g. Sit & Lindner, 2006; Fruchart et al., 2018), personal training or exercise (e.g. Kerr, 2001), recreation (e.g. Kerr, 2007), stress-regulating effects (Martin et al., 1987), behavioural counselling (Blaydon et al., 2004), management (Carter, 2005), as well as consumer behaviour (Davis, 2009). Within the context of online advertisings Rodgers and Thorson (2000) was the first post-millenial study to incorporate reversal theory, specifically, the telic and para-telic dyad into their Interactive Advertising Model (IAM). Recent extant studies that have also underpinned their work with reversal theory include the works of Jung et al. (2014), and Seyedghorban et al. (2016). Generally, these studies indicate that the effects of predictors on consumer reactions to online advertising vary across user modes/groups. In their study, Jung et al. (2014), proposed that internet users meta-motivational state, that is, telic versus paratelic mind frame defines the effectiveness of advertising interactivity. Specifically, the authors examined the moderating role of the telic/paratelic user mode in persuasion. Through a field experiment, they demonstrated that internet users in a serious- minded (telic) state form highly positive attitudes toward online ads with low levels of interactivity whereas those in a playful-minded (paratelic) state develop positive attitudes toward ads with low interactivity levels. The work of Jung et al. (2014) reinforced the suggested mechanism of reversal theory and appears to be one of the first online advertising study and by extension marketing studies to have used reversal theory. A bit more recently, seyedghorban et al. (2016), replicated and extended a model of internet advertising avoidance originally proposed by Cho and Cheon (2004). Working with the supposition that while the explosion of the Internet significantly improves ways in which firms target their 47 University of Ghana http://ugspace.ug.edu.gh customers via online advertising, consumers have also become very capable at avoiding such advertisements (Johnson, 2013), Seyerghorban et al. (2016) used a survey of 339 Iranian internet users to validate Cho and Cheon’s model. Relevantly, the extension phase of the study found that user mode moderates the associations acknowledged in the initial model. Consistent with the tenets of the reversal theory, the study showed that the effects of predictors such as perceived goal impediment, and prior negative experience on advertising avoidance differed among telic and paratelic users. In spite of reversal theory's recognition in diverse fields, the marketing literature hardly explores this noteworthy behavioural theory even in the consumer behaviour context (see Rodgers & Thorson, 2000). Based on these earlier submissions and their findings, Jung et al. (2014) as well as Seyerghorban et al. (2016) suggest the need for future studies to explore how reversal theory explains consumers’ online behaviours, particularly, given the extensive growth of internet usage and online advertising. Relative to conventional media, the Internet is considered a more goal and task-oriented medium (Cho & Cheon 2004). Though consumers visit the Internet to achieve their goals, the nature of these goals (e.g. research, entertainment, shopping, socialization etc.) may put internet users in different states of mind at different times. Because the telic/paratelic meta- motivational states form a “goal directedness continuum” (Apter, 1984), the strength of behavioural responses to online advertising may fluctuate for users at the conflicting points. Hence, following from earlier discussions in this section, and the aim of this thesis, there is room to respond to some scholarly calls through the integration of reversal theory in examining consumer behavioural responses to ODA in Ghana. This thesis argues that stimulus interpretations are contingent on the “internet user mode” in which consumers may be at the time of exposure, and so the study examines the moderating effect that 48 University of Ghana http://ugspace.ug.edu.gh telic/paratelic meta-motivational states, conceptualised as user mode may have on internet users’ attitude toward online advertising and behavioural responses (ad acceptance and ad avoidance) to online display advertising driven by specific ODA characteristics/stimuli (interactivity, placement, informativeness, personalisation, and exposure conditions). 3.6 OTHER THEORETICAL PERSPECTIVES As earlier stated, three other theories stood out in the literature review (see chapter three). Although these theories are less useful within the parameters of this thesis relative to the underpinning SOR model and the reversal theory, they do provide some added insights on the research problem and objectives the study seeks to address. For instance, considered a dual process model, the elaboration likelihood model (ELM) suggests that, there are two paths to consumer persuasion – a central route and peripheral route (Petty et al., 1983). The “central route” is used when consumers are extremely motivated to process persuasive information. In such instances, more mental resources are deployed to scrutinise and elaborate on message content or argument. This results in the formation of an enduring attitude which is expected to be a relatively strong predictor of subsequent behaviour. The “peripheral route” is taken in situations of low motivation or involvement, where recipients consider a persuasive message as personally irrelevant or “are engaged in distracting task during their exposure to the appeal” (Cacioppo & Petty, 1984, p.673). In situations of low elaboration (peripheral route), the acceptance or rejection of a persuasive message is not based on the careful consideration of issue-relevant information, but on superficial analyses. For example, consumers may be attentive to exterior features like, colour, music, and image etc. in an ad (Wang et al., 2009). This causes a provisional change in attitude which is less predictive of behaviour. From the perspective of online display advertising researchers (e.g. Rappaport, 2007; Spilkerattig & Brettel, 2010), the variant push and pull online advertising 49 University of Ghana http://ugspace.ug.edu.gh forms, and characteristics of online advertising such as interactivity, make the ELM an appropriate framework for examining online advertising effectiveness. Also, the theory of mere exposure effects which concerns how individuals form favourable attitudes from brief exposure to a stimulus, posits that the more exposure a person has to a stimulus, the more likely the person is to like the stimulus object (Zajonc, 1968),). It has been argued that in most cases than not, exposure to advertisements occur under incidental situations when the individual’s attention is directed toward something else (e.g. browsing the internet, reading a magazine, or playing a game). In situations of this sort, any stimulus (such as an ad) that is incorporated into the context of the individual’s focal activity, are regarded as background issues and are not actively processed (Wang et al., 2013). Nonetheless, because most affective reactions and learning occur outside awareness, such advertisements may still induce changes in the subconscious minds of people (Zajonc, 1980). More precisely, repeated exposures to these ads may cause preference and heightened liking ‘merely’ because viewers are subconsciously conversant with them (Janiszewski, 1993). In the online display advertising milieu, some scholars (Briggs & Hollis, 1997) have asserted that by merely viewing a banner ad without clicking on it, viewers may still develop brand recognition which may impact their attitude toward the brand. Although the empirical evidences from studies that have applied the theory seem to differ (Goodrich, 2014), the bulk of the evidence seems to affirm high likelihood of mere exposure effects, nonetheless under certain conditions than others. For instance, Yeu et al. (2013) found that high achieving gamers (i.e. those more focused on the game) reported higher levels of implicit memory for banner ads imbedded in the game, than those less focused/involved in the game. Similarly, in an earlier study, Schneider and Cornwell (2005) using a banner ad in a car 50 University of Ghana http://ugspace.ug.edu.gh racing game, found significant recall, and recognition effects when consumers were familiar with the brand compared to unfamiliar brands. The key notion of this theory lies in the assumption that independent of considerable rational information processing, consumers become more familiar with an ad and the advertised brand just by being repeatedly exposed to it, and this acquaintance may result in the development of positive emotions (Zajonc, 1968). Another theory that surfaced in the literature and adds up to the discussion is the theory of psychological reactance. With a rich history in research on persuasive communication, this theory has been used to understand why persuasive messages or campaigns elicit behaviours opposite to what message senders anticipate (Hornik et al., 2008). From the viewpoint of the reactance theory, attempts to influence behaviour involve both persuasion and coercion, and the extent to which these efforts intrude on or limit the target audience’s freedom to choose or engage in a specific activity determines their response (Brehm & Brehm, 1981). Fundamentally, the theory terms such response psychological reactance, which is a motivational state of arousal to regain the reduced freedom; resulting in a heightened attractiveness of the threatened activity, and in a downgrading of the imposed alternative (Heinberg et al., 2015). Because online display ads are mostly push-based, they may be perceived as intrusive depending on several ad-related factors that deprive consumers of the liberty to enjoy the contents of the medium that interest them. These perceptions may cause irritation, annoyance etc. leading consumers to react negatively towards such online ads (Edwards et al., 2002; Tang et al., 2014). While these alternative theories offer some perspectives on the issues this thesis seeks to address, the viewpoints are limited. In the case of the ELM, the theory considers attitudes formed following exposure to persuasive messages as precursors to behavioural reactions; 51 University of Ghana http://ugspace.ug.edu.gh its major emphasis, however, is on attitude formation and change based on processing styles. Furthermore, the mere exposure effect theory also focuses on cognitive and affective outcomes of advertising exposure. Since behavioural response is the major outcome variable this thesis seeks to evaluate, applying the ELM or the mere exposure effects theory would have provided a partial support to the issues the study seeks to address, thus restricting the contributions they thesis seeks to make, that is enhancing understanding of behavioural response issues in ODA research. The theory of psychological reactance, also though valuable in explaining the negative or avoidance behaviours of consumers towards display ads, gives attention to only one aspect of behavioural direction. According to the research objectives, this study seeks to examine the interrelationships among ODA characteristics, consumer attitude toward online advertising, and their behavioural responses from a two directional (ad acceptance and ad avoidance) viewpoint. Because the SOR model maps the nexus among these three interconnected levels of variables it provides the most suitable baseline theoretical parameter within which the study can be conducted. And because the SOR model also falls short of providing explications to the influence of situational consumer-related variables that may cause variations in the magnitude of the behavioural direction, it was complemented with the reversal theory in order to provide a wholesome view of the study. 3.7 CHAPTER SUMMARY This chapter has presented the theoretical foundations for this thesis. The chapter provided an overview of the SOR model and the reversal theory, discussed their assumptions and dimensions, applications as well as their relevance to the current thesis. Other theoretical viewpoints that could have been applied are also briefly discussed and the rationale for choosing the SOR model and the reversal theories in lieu of the alternatives are provided, The choice of the SOR model and the reversal theory stem from their integrative ability in 52 University of Ghana http://ugspace.ug.edu.gh explaining the effect that ODA characteristics may have on consumer behavioural responses to online display advertising, how attitude toward online advertising may enhance the effect of these characteristics, and the moderating role that internet user mode may play in the baseline relationships. By so doing the arguments raised in the chapter theoretically position the study in making contributions to knowledge as far as online display advertising and consumer behavioural response issues are concerned. The succeeding chapter presents a review of extant research on ODA which assesses the state of research in the area as well as pinpoints existing research gaps in need of scholarly attention, some of which provided the springboard for this current study. 53 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR REVIEW OF ONLINE DISPLAY ADVERTISING (ODA) RESEARCH 4.1 CHAPTER OVERVIEW The previous chapter discussed the theoretical foundation of the thesis, and how applicable the SOR model and the reversal theory are to examining consumer behavioural responses to online display advertising (ODA). This chapter reviews extant literature on ODA published over a 10-year period (2009 to 2018) in order to unearth relevant issues on the subject matter. Thus, the specific purpose of this chapter is to provide an analysis of current state of ODA research. To do this, the chapter identifies major themes in ODA research, evaluates evidences from studies under these themes, assesses the theoretical focus and methodological approaches used in these studies, identifies gaps, and suggests insights or avenues for future research. By doing this, some of the gaps identified are used to establish the parameters of this thesis in order to contribute to knowledge in the field. The chapter is organized in the following order: section two introduces the chapter and spells out in details the purpose of the review; section three presents the procedure used for the literature review; and section four discusses the definitional scope of advertising, online advertising and ODA as they pertain to this thesis. Section five discusses the issues and evidences from the review in themes. Sections six, seven and eight present findings on the methodological and theoretical approaches identified in extant ODA research, and geographical or economic dispersion of the reviewed articles respectively. Section nine discusses the gaps identified, and section ten provides a summary of the review. 4.2 INTRODUCTION Reviews are essential roadmaps for knowledge advancement, theory development, closure of saturated research areas, and unearthing of novel avenues for research (Webster & 54 University of Ghana http://ugspace.ug.edu.gh Watson, 2002). From the literature, three reviews exist on online advertising research. They include the works of Kim and McMillan (2008), who provided a general picture of internet advertising research with 113 articles published in four dominant advertising journals spanning 1996-2003 by examining influential authors and papers, and co-citation patterns to provide a trajectory from the past to the future. Similarly, Ha (2008) analyses articles published in six advertising journals between the period of 1996-2007 to examine the conceptual underpinnings, and practical contributions of online advertising research, as well as suggest a research agenda for future studies. Notably, some of the issues raised by Ha (2008) have received academic responses, with new trends and topics emerging from the studies that spawned afterwards. Based on such developments, a recent review by Liu- Thompkins (2019) assesses how online advertising research has evolved since the works of Kim and Macmillan (2008) and Ha (2008) were published. Liu-Thompkins’s (2019) work discussed six thematic areas in online advertising research and pointed out some future research questions to be addressed. As valuable as these reviews have been, they focused broadly on online advertising, and were limited to specific advertising and marketing journals. Considering the interdisciplinary nature of the phenomenon as well as current trends procreated in recent times particularly, regarding the various typologies of online advertising, this chapter using an expanded journal set, reviews literature on online display advertising. As the fastest growing category of online advertising, academic literature on ODA is correspondingly maturing; and so, by examining the empirical issues and evidences, conceptual and methodological approaches used in these studies, this chapter pinpoints knowledge gaps that may guide potential contributions in the area. 55 University of Ghana http://ugspace.ug.edu.gh 4.3 PROCEDURE FOR THE REVIEW A fairly systematic approach is adopted in carrying out this review, as the approach has been advocated for its rigorous methodology, wider literature coverage, objectivity and coherent presentation of discourse (Okoli, 2015; Anees-ur Rehma et al., 2016). A number of parameters were set in identifying, selecting and categorizing the literature sources. These criteria and the results from the literature search are delineated in subsections one to three. 4.3.1 Scope of the Review The review was limited to peer-reviewed articles published in English language academic journals. This is in view of scholarly suggestions that the thoroughness of the peer-review process offers a high level of quality to research (Bornmann, 2011). Moreover, journal articles are conduits for evidence gathering and disseminating new findings to both practitioners and researchers (Ngai 2005), and are, therefore, regarded as sources of authenticated knowledge that have maximum influence within the academic setting (Podsakoff et al., 2005). On these bases, unpublished working papers, conference proceedings, essays, books and doctoral dissertations were excepted from this review. Additionally, in an attempt to enhance and balance the current state as well as avoid overlays with extant reviews on online advertising research (e.g. Ha, 2008, Kim and McMillan, 2008, Liu-Thompkins, 2019), the current review focuses on ODA. For the same reason, articles on social media advertising were also excepted since Knoll (2015) has reviewed research in that area. This review was also time-restricted to identify and focus on studies that have been published between 2009 and 2018 since the period is said to have witnessed several changes and dynamism to online advertising particularly in the area of display advertising (Auschaitraku & Mucherjee, 2017). 56 University of Ghana http://ugspace.ug.edu.gh 4.3.2 Literature Search, Article Selection and Information Extraction Pursuant to the aforementioned criteria, six online databases that accommodate diverse peer-review journals, as well as search engines (google scholar) were used for the search. The databases include Taylor and Francis Online, Emerald Fulltext, Science Direct, JSTOR, Wiley Online and Inderscience Online, and EBSCOhost. These databases were used for the search because they encompass a wide range of peer-reviewed journals in management science, communications and economics among others; more importantly, they index key advertising and marketing journals (Liu-Thompkins, 2019). The literature search was conducted using search keywords/phrases to enhance identification of information relevant to the study area. Keywords or phrases such as “online display advertising,” “display ad,” “web banner advertising,” “banner advertisings” and “internet advertising” were used one at a time. The databases were interrogated with these key phrases in titles, abstracts and keyword list in March 2018, which initially produced several articles. Through a detailed examination of the abstracts of these articles, duplications, as well as papers that made reference to online advertising but did not have display advertising as the focus of the paper, were eliminated, resulting in 106 articles. These 106 articles were subsequently assessed to determine their relevance to the study. In the process, some articles were superficially unclear; hence the full paper had to be read before a decision was made to exclude or include them. Relevance was determined on the grounds that online display advertising, was at least, the main or one of the main forms of online advertising under discussion, and its features and significances were discussed even if its usage was not deliberate. Following from this process, 63 articles met the study parameters and were thus used for the review. The final articles used in this review are provided in the appendix section (see Appendix A2), and also asterisked (*) in the list of references. 57 University of Ghana http://ugspace.ug.edu.gh After creating a working list of 63 articles, certain preliminary information, such as year of publication, publishing journal, type of articles, geographic setting of the study, measured variables, and study theme, were recorded from each article. In addition to these, the theoretical focus and methodological approaches adopted were extracted, categorized and documented. The unit of analysis was, therefore, the full-text research article. The publishing journal and year of publication were essential for supporting claims of online advertising as a multidisciplinary field, gauging how much attention has been given to the area, as well as estimating growth trends. The study themes border on the major online display advertising subjects researched as well as the issues and evidences that emerged. Regarding theoretical focus, these studies were classified as using theoretical or atheoretical (i.e. frameworks, models, concepts and category-based) approaches (Heeks & Bailur, 2007; Kim et al., 2014). The methodological approaches were categorized as qualitative, quantitative, and mixed methods. Studies classified as qualitative studies are those that gathered and analysed data using thematic analysis and interviews while quantitative studies are grouped as utilizing surveys and experimental strategies as well as secondary market data. Mixed methods studies utilised both qualitative and quantitative approaches while non-empirical studies passed as conceptual papers. Results from the literature search are presented in the subsequent sub-section highlighting the distribution of the articles by journals, and yearly publication frequency. 4.3.3 Distribution by Journal and Year The literature search carried out in the six databases produced differing number of articles with Taylor & Francis Online producing the highest number. Considering the ten-year period under review, a majority of the articles (approximately 71%) were published between 2009 and 2014. Relatively, there was a decline in the number of published articles in the following years. This could among other things, be attributed to the attention of some 58 University of Ghana http://ugspace.ug.edu.gh advertising journals within these periods to social media and native advertising which are relatively nascent forms of online advertising. Considering that 3 articles were identified at the time of the review (March 2018), it seemed fair to assume an increase in the number of publications for the study’s cut-off period. This observation provides a signal that the relevance of online display advertising as a subject of scholarly enquiry appears to be growing. The Figure 4.1 depicts distribution of the articles by year. Figure 4.1 Distribution of Articles by Year 10 9 8 7 6 5 4 3 2 1 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year Regarding the journals (see Appendix A3) that published these articles, a bulk of them (23 out of 39 journals) were situated in the field of marketing, contributing approximately 70 percent of the reviewed articles; 5 journals were within the field of information management/systems, and the remaining 11 spread across economics, operation studies and general management, as well as area and sector studies. Journals that made significant contributions to ODA research in terms of quantity of articles published within the period under review include International Journal of Advertising which published seven articles; Journal of Marketing Communication and Journal of Advertising Research published four articles each; and Journal of Marketing Research and Computers in Human Behaviour published three articles each. Two articles each were published by Journal of Advertising, Journal of Interactive Advertising, Journal of Interactive Marketing, Journal of Promotions 59 Number of Articles University of Ghana http://ugspace.ug.edu.gh Management, Psychology & Marketing, Marketing Science, International Journal of Internet Marketing and Advertising, and Electronic Commerce Research and Application. These 12 Journals accounted for approximately, 60 percent of the articles published. The remaining 25 articles (approximately, 40%) were published in varied journals. The diversity of the journals speaks to the multidisciplinary and varied nature of online advertising. A detailed outline of the journals and the articles published in them are presented in Appendix A3. In order to provide a frame for the review, in the next section, the definitional parameter of online advertising and online display advertising is discussed. 4.4 ADVERTISING AND ONLINE ADVERTISING To put discussions in this chapter in the right perspective as well as delimit the boundaries of the entire study, this section spells out the definitional scope of online advertising and narrows down to online display advertising which is the focus of the thesis. Harker (2008) observes that online advertising lacks a formal conceptualisation and this observation is depicted by the ad-hoc definitions in the literature, mostly based on the remit of such scholarly enquiries. The definitions of online advertising available in the literature, as enumerated in Table 4.1, are aligned with two perspectives of advertising definition. Generally considered a marketing communication tool, advertising has been defined as “a paid, non-personal communication from an identified sponsor using mass media to persuade or influence an audience” (Wells et al., 1992, p.21). Following several changes to the advertising field, which rendered the definition of Wells et al. (1992) a tad conservative, the American Marketing Association (2005) defines advertising as “the placement of persuasive messages in any of the mass media by organisations in order to inform and/or persuade people of a specific target market about their products, services, organisations or ideas” (Van de Waldt et al., 2007, p.186). 60 University of Ghana http://ugspace.ug.edu.gh Closely aligned with the AMA’s definition, is Schlosser et al. (1999) early definition of online or internet advertisement as “any form of commercial content available on the internet that is designed by businesses to inform consumers about a product or service” (p.36). Meyers and Gerstman (2001) coming from a similar angle, offer a definition that considers it as “a form of promotion that uses the Internet and World Wide Web for the expressed purpose of delivering marketing messages to attract customers”. Situated between the two general advertising perspectives, Barnes (2002, p.400) also considers it as “any paid advertisement, from banners to sponsorships, which appears on the web or other Internet channels, including e-mail”. Ha (2008, p.31), however, defines it as “deliberate messages placed on third party web sites including search engines and directories available through Internet access” – a definition that appears closely linked to the perspective of Well et al. (1992). While, in the view of Schlosser et al. (1999), online advertising could be delivered via any medium in any form as well as offer information at any extent of depth, Ha’s (2008) definition excepted self-promoting platforms or channels such as websites. Ha (2008) argues that if online advertising is to enhance advertising theory, its definitional scope should derive from the “traditional” definition of advertising, and permit research findings to work in partnership with online ad spending research and industry statistics since these data do not take into consideration self-promoting websites or e-mails. However, Johar (2016) reviewed the changing advertising scene and points out that the definition of advertising has fundamentally altered. According to Johar (2016), advertising is basically about persuading consumers, and so, definitional issues regarding whether or not it is paid for, non-personal, and/or comes from an identified sponsor, are obsolete. Further, drawing inspiration from Dahlén and Rosengren (2016) who defined advertising as “brand-initiated communication intent on impacting people”, Johar argues that in today’s business 61 University of Ghana http://ugspace.ug.edu.gh environment, even blog posts can qualify as advertising. Notwithstanding the absence of unanimity on an overarching umbrella definition of the scope and dimensional confines of online advertising, one commonality in the definitions in table 4.1 below is that any message on the internet intended to inform consumers and promote a brand can be considered online advertising. Table 4.1 Selected Online Advertising Definitions Author(s) Definition Schlosser et al. (1999) “Any form of commercial content available on the internet that is designed by businesses to inform consumers about a product or service.” Meyer & Gerstman (2001) “A form of promotion that uses the Internet and World Wide Web for the expressed purpose of delivering marketing messages to attract customers.” Barnes (2002) “Any paid advertisement, from banners to sponsorships, which appears on the web or other Internet channels, including e-mail.” de Pelsmacker et al. (2004) “Spreading a commercial message in standardized formats on rented spaces on websites of other companies.” Ha (2008) “Deliberate messages placed on third party web sites including search engines and directories available through Internet access.” Hanafizadeh & Behboudi (2012). “Internet-based process by which advertisers communicate, interact with and persuade online users in order to position a brand, which allows a company to promote both consumer awareness and preference in a customized and personalized way, and to decrease the time needed to make a buying decision.” 4.4.1 Typologies of Online Advertising According to Goldfarb (2014), although a consensus on a canopy definition for online advertising is lacking, the phenomenon can be clustered into three dominant typologies; 62 University of Ghana http://ugspace.ug.edu.gh classified advertising, search advertising, and display advertising. An additional form that is pointed out in recent literature is what has become known as native adverting. Classified advertising generally refers to geographically specific ads that are presented or featured on classified websites and news blogs that do not provide algorithmic search or other media content (Goldfarb, 2014). To put this differently, they are textual links included in specialised online listings or web catalogues (Xu et al., 2014). The term classified is used to signify the way in which the products and services being advertised are grouped under specific headings. Craigslist is considered the largest of such classified websites and online job sites as well fit this category of online advertising. In Ghana, GhanaAds.com is an example of a classified advertising site. Search advertising, also known as sponsored/paid search advertising, describes the practice of advertisers paying fees to Internet search engines such as Google, Yahoo, MSN, and Bing etc. to fashion text ads in response to keyword searches which appear or are displayed alongside organic (non-sponsored) web results (Goldfarb, 2014; Yang & Ghose, 2010). This type of online advertising is directly linked to consumers’ search requests which are “statements of intent” offering advertisers the right opportunity to expose these consumers to their ads because keywords and ad messages are typically matched with user-originated queries or behaviours (Goldfarb, 2014). In view of this, in relation to other forms of online advertising, search advertising scores extremely low on the consumer “intrusiveness spectrum” and is considered as leading to more fitting prospects for advertisers or firms (Yang & Ghose, 2010). Owing to the increasing popularity and reach of these search engines, search advertising represents one of the most rapidly growing forms of online advertising and is fast becoming a principal revenue stream for search engines (Katona & Savary, 2010; Yao & Mela, 2011). 63 University of Ghana http://ugspace.ug.edu.gh 4.4.1.1 Online Display Advertising (ODA) ODA is a form of online advertising in which advertisers pay relevant publishers or third parties (e.g. news sites, blogs, social networking sites, commercial websites etc.) to place graphical ads on their web pages (Chapelle et al., 2014). Online display ads are ads that come in graphic images varying in shape, size, animation, colour, and duration that internet users or website visitors see along with other content (Goldstein et al., 2014). Display ads can be placed on any type of website and are more frequently seen on commercial and social media sites as well as applications (apps) (Auschaitraku & Mucherjee, 2017). ODA includes such diverse formats as banners, wallpapers, interstitials, skyscrapers, floating ads, pop-ups, plain text ads, rich media ads, and video ads among others (Draganska et al., 2014; Goldfarb, 2014). Ad format denotes “the manner in which an ad appears” (Rodgers & Thorson, 2000, p.47) and so, banners, for instance, refer to horizontal, quadrilateral-shaped graphical elements found at the top of webpages while, skyscrapers, though akin to banners, are vertically located on the borders of a webpage (Burns & Lutz, 2008). Pop-up ads open another window over the user’s browser and can only be removed from the screen by closing or minimizing the window while floating ads use a combination of flash technology and Dynamic Hypertext Mark-up Language (DHTML) to create a translucent or shaded layer over the web page and then execute an animated ad within this layer and disappears after a specific time (typically 5-30 second). Interstitials are presented automatically to users when they move between two content pages, and once the requested page loads, the interstitial disappears, and rich media is an umbrella term used for ads improved with sound, video, motion and interactive features (Ha, 2008; Brajnik & Gabrielli, 2010). Although ODA is used as a catch-all term for all graphic ads users see online, it is contrasted from native advertising which is a relatively nascent form of online advertising. Also 64 University of Ghana http://ugspace.ug.edu.gh referred to as sponsored content, native advertising describes any paid customised message that is similar in form, tone, appearance, functionality and standards to the editorial and entertainment content as well as the autonomously produced materials of the publishing site (Tutaj & van Reijmersdal, 2012; Bakshi, 2015). Native ads are particularly popular on informational websites such as news sites and blogs as well as social networking sites (Becker-Olsen, 2003; Hoezel, 2014). While native ads are graphic in nature too, they are considered an independent category of online advertising because they are meant to blend in with the webpages on which they are placed, and not look like ads at all in order to circumvent consumer detection (Benton, 2014). Online display ads, however, are meant to stand out and attract user/consumer attention. ODA is an essential form of online advertising which currently accounts for 49.7% of the online advertising market and is its fastest-growing category (eMarketer 2018). Given its relevance as a form of online advertising, its increasing use, its push-based (compared to search advertising), prominent and ubiquitous nature; internet users are exposed more frequently to display ads on webpages (e.g. commercial, social media, and search engines) accessed on PCs and all forms of mobile devices than they are to other categories; and so, this thesis focuses and is limited to online display advertisements. Having spelt out the definitional boundary of ODA, in the subsequent sections five to eight, findings from the literature classification and analysis on ODA are presented. These findings have been discussed in four broad sections: major themes in ODA research, methodological approaches, theoretical approaches, and geographical/economic dispersion of studies. 4.5 MAPPING ODA RESEARCH: MAJOR THEMES, ISSUES AND EVIDENCES In mapping out the reviewed articles, some pertinent issues and sub-issues were identified in order to help comprehend dominant trends in the literature and put the discussion in a 65 University of Ghana http://ugspace.ug.edu.gh comely perspective. The mapping exercise examined thematic patterns or some fundamental imports of studies in ODA research communicated in the literature. Working within the parameters of the current study, three key themes (with sub-themes) seem to emerge. This section, therefore, discusses the issues and evidences from the reviewed papers within these three themes namely: antecedents to online advertising effectiveness, assessing online advertising effectiveness, and consumer attitude towards online advertising. 4.5.1 Antecedents to ODA Effectiveness Online media and, for that matter, advertising within online environments keeps evolving. This drives advertisers’ and marketers’ in their quest to gain insight into the types of advertisements that are suitable in the online domain. Given the diverse range of online ad formats, as well as inadequate and varied information regarding their usefulness, online advertising campaign choices become an intricate issue for advertisers (Burns & Lutz, 2008, de Pelsmacker & Neijens, 2012). There is, therefore, a growing body of literature on online advertising effectiveness, and this literature stream on one hand, describes the precursors to online advertising effectiveness. Studies that examine the effectiveness of ODA point to ad characteristics (such as format, design and execution elements and contents), context characteristics, and consumer characteristics as essential antecedents to online display advertising effectiveness (e.g. Jung et al., 2011; Martin-santan & Beerli-Palacio, 2012; Wang et al., 2013; Tang et al., 2014). This lends credence to Rodgers and Thorson’s (2002) Interactive Advertising Model (IAM) which stresses that both “advertiser-controlled” and ‘consumer-controlled” factors are central to internet users’ perceptions and processing of online ads. The ensuing sub-sections therefore probe into the issues and evidences from ad and consumer-related perspectives that surfaced in the literature analysis. 66 University of Ghana http://ugspace.ug.edu.gh 4.5.1.1 Ad-related Issues: Design features or Executional Characteristics Contributing to this sub-theme in the ODA literature are papers that address forms and types of ODA, and how they should be executed for optimum outcome. These studies examined varieties of online ad formats in order to ascertain their usefulness because, in the view of Burns and Lutz (2006), the nature of online ad format is an important issue that influences online advertising responses. Common ad formats and types appearing in these studies are shown in Table 4.2 below. Despite being the oldest form of online ad formats, banner ads are still the most commonly studied, and this could be attributed partly to the continuous development of the banner ads market (Goodrich, 2010) as well as the tendency of researchers to use banners as an umbrella term for ‘non-rich media’ display ads. Table 4.2 Online Display Advertising Forms Studied ODA Formats/Forms Articles o Banners Kuisma et al. (2010); Lee & Cho (2010); Heish & Chen (2011); Jung et al. (2011); Martin-Saantana & Beerli-Palacio (2012); Tutaj & van Reijmersdal (2012); Bright & Daugherty (2012); Flores et al. (2013); Goodrich (2013); Wang et al. (2013); Yeu et al. (2013); Segev et al. (2014) Bleier & Eisenbess (2015a; 2015b); Hussain et al. (2018) o Skyscrapers Kuisma et al. (2010); Goodrich (2011); Goodrich (2014) o Pop-ups Janssens et al. (2012); Chan et al. (2010) o Interstitials Ying (2009) o ODA (Blanket) Fulgoni & Mornn (2009); Kireyev et al. (2015); Goldfarb & Tucker (2011a) o Video Ads/Skippable Video Ads Pashkevich et al. (2012); Belanche et al. (2017); Goodrich et al. (2015) o Rich Media ads Rosenkrans (2009) Although research focus on online ad format is still ongoing and valuable, a growing trend in recent literature has been a concentration on ad characteristics or executional features following Burns and Lutz’s (2008) call. Over a decade ago, Burns and Lutz (2008) studied 67 University of Ghana http://ugspace.ug.edu.gh comprehensively the role internet users’ perceptions about six ODA formats play in their attitude formation towards these formats. Finding that user perceptions and attitudes varied across the various ad formats, the researchers assert that the formats can be “profiled” to inform their suitable selection based on the online advertising goals of a firm. Burns and Lutz (2008) thus, advocated for studies that will explicitly examine how characteristics inherent in the various ad formats influence attitude. And so, studies that spawned after that, have given attention to the design, content and placement of ads as essential considerations in the literature (Ying et al., 2009; Kuisma et al., 2010; Goodrich, 2011). Today, one cannot effectually discuss the effectiveness of online display advertising without bringing up issues of ad characteristics. As Tang et al. (2014) put it, “as … online ads have significant influence on consumers cognitive, affective and behavioural responses, design features of online ads is an area worth investigating” (p.2). Essentially, ad characteristics make up the ad formats and Tang et al. (2014) underscore the importance of examining the effects of these characteristics than simply focusing on the formats. By that, the essence of ad characteristics in driving the effects of display ads is recognised. Presently, existing formats show a variety of features, and according to Brajnik and Gabrielli (2010), some of these features are inherent, in the sense that they can be identified from the content of the ad; while others are “relational” – dependent on the how the ad appears as well as its location and the context in which it is utilised. Since one of the aims of this thesis is to examine the effects of ad-characteristics, on behavioural responses, Figure 4.2 provides a snapshot of the executional characteristics that have steadily appeared in ODA research reviewed, and the ensuing passages interrogate these features as essential antecedents to online advertising effectives. 68 University of Ghana http://ugspace.ug.edu.gh Figure 4.2 Cited Executional Characteristics in ODA Research Informativeness Placement Exposure Condition Entertainment Animation Personalisation Interactivity Size 0 2 4 6 8 10 Number of Appearances i. Informativeness The role of advertising as a basic information source has long been established in the literature, and research has illustrated that the informativeness of advertising is its main legitimising task. This is on the grounds that the ability of an advertisement to provide useful information and present a true picture of products, is the prime reason why consumers approve of it, and is a fundamental consumer belief underlying its innate economic benefits (Rotzoll et al., 1990). Informativeness is considered the ability of ads to inform consumers of products and alternatives so that purchases yielding the greatest possible satisfaction can be made (Ducoffe, 1996a). Affirming Ducoffe’s view of informativeness in the context of online advertising, Schlosser et al. (1999) argued that consumers’ attitude towards online advertising is a function of the informativeness of advertisement as well as its usefulness in stimulating purchasing behavioural decisions. Informativeness gained ground in the literature because as a goal-oriented medium, internet usage is driven by a number of motives, and information search is one of these motives (Rhoades et al., 2008). Informativeness as a feature is therefore, extensively explored in the literature on online advertising (Sun & Wang, 2010), and online display advertising 69 University of Ghana http://ugspace.ug.edu.gh researchers embrace the idea that it is a means to enhancing receptivity toward and interaction with advertisements since online ads with beneficial information are perceived more favourably by internet users (Goodrich et al., 2015). Prior studies on informativeness have assessed the extent to which display advertising aim to provide useful information, and how this influences consumers’ attitude and responses. For instance, Wang and Sun (2010a; 2010b; 2010c; 2010d) examined the relationship among consumers’ belief about online advertising, attitude toward online advertising, and their behavioural responses across multiple cultural setting, and found that informativeness as an essential belief factor among others, significantly predicts attitude toward online advertising, which then influences ad clicking and frequency of online shopping. These earlier studies and others (e.g. Li-Ming et al., 2013; Mahmoud, 2014) assert that the characteristic of informativeness satisfies internet users’ or consumers’ informational needs which moves them to attend to the advertisements. These assertions are corroborated by Goodrich et al. (2015) who stressed in their study that the inclusion of useful information among other features, can help overcome perceptions of intrusiveness which may have detrimental effects on advertising outcomes – e.g. ad abandonment. In their opinion, if internet users evade watching an advertisement, very minimal favourable outcomes can be anticipated for advertisers. In a web environment where advertisers aim to attract consumers’ attention and get them to interact with their online advertisements, a prerequisite to achieving this aim is to have consumers actively use online ads for information gathering (Mahmoud, 2014; Zha et al., 2015). Gleaning from the literature, though the influence of informativeness as a feature of online advertising seem to have been established, the extent and nature of this influence vary across cultural or economic settings. Wang and Sun (2010a) in their study, found that although 70 University of Ghana http://ugspace.ug.edu.gh Americans view online advertising as more informative, and were more likely to make online purchases as a result of this, relative to Chinese, they were less likely to click on ads. Driven by informativeness among other features, Romanians had the most positive attitude toward online advertising and tend to click on advertisements in relation to Americans (Wang & Sun, 2010b); and Chinese tend to purchase online more than Romanians do (Wang & Sun, 2010c). These findings birthed the need to examine these influences across diverse market or cultural settings to establish more firmly the effect of informativeness in the literature. ii. Placement Placement issues discussed in the ODA literature have been argued from various perspectives. Extant prevailing perspectives in the literature point to issues of ad position, ad-context congruency, and choice of webpage (e.g. Chan, 2010; Janssens et al., 2012; Belanche et al., 2017). From the first viewpoint, it is reasoned that ad position is integral to internet users’ evaluation of online advertising. According to an Eyetrack III study on ad placement in 2004, an ad at the foot of the screen encourages fewer clicking behaviour, while an ad near the top of a webpage garners more attention. This has somewhat been ascribed to internet users’ expectation that most relevant information are shown at the start of a page, and has also attributed to the notion that regardless of a viewer’s screen size, an ad at the top of the page has a higher likelihood of being seen (Nihel, 2013). However, Rosenkrans (2009) in an experimental study found that a rich media ad ran below the fold line generated more clicks. In a later study to examine the effect of ads on attention and reading, Simola et al. (2011) also found that ads positioned to the immediate right of the text region generated and captured more attention than ads placed over the text region. In adding further insight to the ad positioning debate, Goodrich (2014) found ads positioned to the left to be associated with more favourable attitudes than those to the right. The study 71 University of Ghana http://ugspace.ug.edu.gh as well reported significant variations in attitudes resulting from advertisement location based on gender. The author’s findings suggest that males had more favourable attitudes toward left-positioned ads and females had more favorable attitudes toward ads positioned to the right. Because advertisements do not appear in a vacuum, but are embedded in mediums or featured on webpages, consumers’ perceptions and evaluations of ads are as well contingent on contextual factors that are possibly outside their awareness, and so, the second perspective on ad placement addresses issues of ad-context fit (congruency). A number of researchers agree that a fitting media context engenders more positive ad perceptions among internet users because ads that are congruent (thematically similar) with the content of the webpage on which they are hosted, correspond with the specific interests of visitors to such pages thus, resulting in more favourable responses (Jeong & King 2010; Segev et al., 2014; Belanche et al., 2017). Although this is a prevailing view which has caused thematic congruence to become a standard in media choice for practitioners (Moorman et al., 2005), other contrasting findings exist. For instance, in an earlier study, Dahlen et al. (2008) find that placing ads in unexpected media locations compared to expect media environments improve the effectiveness of such ads. They argued that contrasting stimuli may draw more viewer attention and hence result in more processing. Contributing to the ad-context congruency literature, Bleier and Eisenbess (2015a) showed that the context in which an ad appears does not always influence its effectiveness since such influence is dependent on the online browsing mode of consumers. Others also pointed out that the effect of ad-context congruency may only be understood through individual and situational factors that may cause the differences in effect (Janssens et al., 2012). 72 University of Ghana http://ugspace.ug.edu.gh Other placement issues that emerged in the reviewed literature include the type of webpages on which ads may be featured (Goodrich, 2011; Hsieh & Chen, 2011), These studies suggest that webpages with different information type affect users’ attention to ODA and specifically, video-based and picture-based webpages have a relative edge in capturing and retaining consumers’ advertising attention than text-based and text–picture-based webpages. Quite recently, Auschaitraku and Mucherjee (2017) find that propelled by higher processing fluency, online display ads are more effective on commercial than on social websites, and on brand pages than on personal pages of the latter. While previous studies have extensively explored the role of ad placement from various viewpoints in the ODA literature, the results are divergent thus requiring future enquiries into placement effects. Particularly, recent calls made in the literature have asked for more research focus on how ad-context issues influence advertising effectiveness in order to legitimize prior claims and establish more strongly their effect (Stripp, 2018) especially on behavioural responses since most previous studies have frequently examined cognitive and affective outcomes (Janssens et al., 2012; Kim et al., 2018). This can offer insights into how advertisers can strategise their ODA placement more effectively given the emergence of diverse media platforms. iii. Exposure Conditions Advertising exposure refers to the presentation of an ad to target audiences, viewers or consumers. Exposure is an essential determinant of advertising effectiveness because for an advertisement to have effect on consumers, they must see/watch, pay attention to, and comprehend the advertising message being presented (Percy & Elliot, 2012). Therefore, determining exposure issues that influence advertising effects has been an essential concern for both practitioners and scholars (Fugoni & Morn, 2009; Chan et al., 2010). The most studied exposure subject in both traditional and online media contexts has been frequency or number of repetitions (Ying et al., 2009). This derived from Krugman’s (1977) ‘three 73 University of Ghana http://ugspace.ug.edu.gh exposure concepts’ which espoused that an initial exposure to advertisements causes viewers to ask, “what is it”, after a second exposure, consumers ask “what of it”, and the third exposure serves as a reminder among other things. From the viewpoint of mere exposure effect, when individuals are frequently exposed to a stimulus, their positive emotions toward the stimulus are enhanced, which then improves the effectiveness of persuasion (Zajonc, 1968). In other words, the more exposure people have to stimuli, the more they tend to like it. The literature on ad repetition thus suggest that, repeated exposures enhance the memory of viewers as well as their favourable attitudes, and behavioural intentions toward the ad (Li & Lo, 2015). Several researchers have supported the view that display advertisements presented over multiple exposures enhance ad recognition, favourable attitude and behaviours owing to heightened liking for the ad and the brand as well (Lee & Cho, 2010; Kim, 2018; Hussain et al., 2018). Conversely, some researchers have cautioned that repeated exposure is also a source of users’ unfavourable perceptions about online advertisements. Specifically, the number of exposures is said to affect internet user’s perceptions of intrusiveness (McCoy et al., 2017). Because consumers go online to achieve certain goals, excessive exposure to the same ad or too many ads at once delays their online experiences, hence causing a backlash (Yaverogly & Donthu, 2008). Another facet of exposure considered, is how exposure duration plays into consumer processing and responses to ODA. And so, drawing on the literature on processing fluency, other studies focused on the exposure duration as essential to online display advertising effects. For instance, driven by the idea that internet users may not always encounter an ad more than once during an online session, Wang et al. (2013) investigated the influence of exposure duration on viewers’ attitude toward the ad and the brand. It was revealed in their 74 University of Ghana http://ugspace.ug.edu.gh study that longer exposure durations may not always result in favourable influences, in the sense that the effect of exposure duration is dependent on the banner ad complexity. Specifically, when the ad is difficult to process, an increase in exposure duration resulted in an increased attitude toward the ad and advertised brand. Within the context of video ads, Goodrich et al. (2015) also found that lengthy ads have more positive effects; particularly they generated more ad recall and reduced perceptions of intrusiveness. Exposure to ads on the Internet is becoming more rampant, and advanced internet technologies now allow for display ads to be forced on internet users (Kim, 2018). This represents the third exposure issue discussed in the literature under review. In conventional media, television commercials depict instances of forced exposure because they disrupt the viewing process of audiences during or in-between programmes whereas newspaper ads exemplify voluntary exposure since readers may choose to view them or not when reading (Li & Leckenby, 2004). The internet, however, has the ability to display advertisements in both voluntary and forced exposure modes (Li & Meeds, 2007). Given the tendency of internet users to use ad filters to prevent the display of ads on their browser, advertisers and publishers have found means through evolving and complex technologies to generate varying levels of exposures (Hussain et al., 2018). These varying levels are explained by the extent to which consumers feel compelled to watch an ad (Kim, 2018). Exposure is thus forced when it takes away viewers’ freedom to choose what they watch and when they watch it. The few studies conducted into forced exposure have reported negative perceptions of consumers towards forced exposure ads driven by perceptions of intrusiveness (Hegner et al., 2015). Although exposure conditions have been a recurring theme in ODA research, most of the studies reviewed have mostly focused on one or at most two of these conditions (Wang et al., 2013). This study therefore integrates these three perspectives of exposure 75 University of Ghana http://ugspace.ug.edu.gh (repeated exposure, duration of exposure and forced exposure), and by so doing provides a broader and clearer view of their collaborative influence from a ‘single audience’ viewpoint. iv. Entertainment Extant research shows that several people use the internet in pursuit of fun and leisure, and for such people, entertainment gratifies the pleasures of being online; they, therefore, hold entertaining expectations of advertisements presented on the internet as well (Ching et al., 2013). According to Ducoffe (1996b), the entertainment value of online advertisement refers to the degree to which viewers perceive the ad as enjoyable or pleasing. As another precursor to online advertising effectiveness, entertaining ads have the ability to satisfy viewers’ desire for amusement, distraction, and visual pleasure (Bellman et al., 2014). Thus, such ads are able to create an affective connection between consumers and the advertising message (Wang & sun, 2010a). With the internet’s increasing advertising density and countless advertising messages competing for consumer attention, an online ad ought to be fascinating and enjoyable in an innovative way to not only capture but also retain consumer attention as well as stimulate favourable attitude toward the advertised brand (Baron et al., 2014). The efficacy of entertainment as a characteristic of online display ads is in its tendency to stimulate sentiments that are successively linked to the advertised product or service brand (Goh & Ping, 2014). Accordingly, an entertaining advertisement can influence consumer attitudes toward the ad and the brand and generate purchase intentions because it brings surprise and excitement to viewers (Bellman et al., 2014). Within the parameters of the reviewed literature, Jung et al. (2011) showed that when exposed to display ads with high entertainment value, consumers form more positive attitudes toward the brand as well as increased purchasing intentions. Wang and Sun’s (2010a; 2010b; 2010c; 2010d) research 76 University of Ghana http://ugspace.ug.edu.gh also suggest that entertainment is one of five belief factors that significantly predict attitude toward online advertising in both developed (e.g. USA) and developing countries (e.g. China). The authors assert that, in an evolving technological era, and an information-seeking society, the concept of ‘infotainment’ intimates that information and entertainment are often intertwined with each other, and online advertisements that epitomise this concept affect consumers more positively. v. Animation and Size Motion is the central concept underlying animation. In human-computer interactions, it is generally recognised that motion attracts attention, and relative to colour or orientation among other elements, user attention is more sensitive to motion (Nielson, 2000). Animation describes the use of motion in making advertisements noticeable and is a unique feature of banner advertising. Animation is one of the influential “attention-getting” features employed in online display advertising (Ying et al., 2009) because compared with “non- animated” objects, animated objects possess the edge of capturing the initial attention of users (Song et al., 2011). Scholars studying motion effects have submitted that people display an innate inclination toward moving objects; that is to say, when people are exposed to moving images, they pay greater attention to the source of the motion and process relevant information therein (Yoo & Kim, 2005). Because some level of attention is required prior to thorough elaboration of online advertisements, animation has become a key component for attracting viewer attention. While watching an animated ad is no assurance that users will click on the ad, animation does heighten user’s attention, a condition that must be fulfilled before clicking and other behavioural responses can be achieved (Kuisma et al., 2010). 77 University of Ghana http://ugspace.ug.edu.gh The effect of animation on advertising effectiveness has been indicated in the extant literature (Ying et al., 2009; Kuisma et al., 2010; Song et al., 2011; Bruce et al., 2016). Kuisma et al. (2010), examined the influence of animation on attention and memorization, and demonstrated that, its effects may vary on the basis of the ad format. Specifically, animation heightens attention to skyscrapers, but only improves recognition for banners. Similarly, Bruce et al. (2016) suggests that animated ads have greater carryover effects hence, they influence engagement over a lengthier period than static ads. Size also emerged as a design element or executional feature of ODA with three citations. Size is explained as the overall largeness or smallness of a display ad (Bruce et al., 2016). The evidence regarding the effect of size seem inconclusive in the literature. Earlier studies outside the scope of this review have reported no significant effect of ad size on consumer reaction (e.g. Cho, 2003); arguing that internet users learn to ignore ads, though the ads may have some effect through peripheral visions. Also, although larger display ads are more likely to be seen, and so have been associated with higher attention and memorization (Nihel, 2013), they are also perceived as more intrusive because they disrupt viewers activities and cognitive processes more (Ying et al., 2009). Ying et al. (2009) contends that, consumers are more receptive to small ads that occupy just a small area of a page or appear in another window on top of the browser because once viewers can still access a portion of their original content page, they would feel less disturbed. However, these types of ads stand the risk of being disregarded. vi. Personalisation Personalisation has become a key feature of online advertising and is increasingly used in display advertising owing to available consumer information. Consumers in online environments share vast amounts of personal data which marketers and advertisers utilise 78 University of Ghana http://ugspace.ug.edu.gh to make their advertising personalised to a huge degree (Aguirre et al., 2015). These data include personal identifiers, demographics, browsing history, purchase history and location information (Liu & Mattila, 2017). According to Chellapa & Sin (2005), personalisation refers to “the use of technology and customer information to tailor electronic commerce interactions between a business and each individual customer” (P.184). It has been described as a customer-oriented marketing strategy that adapts web content to deliver the right information to the right person at the right time, to increase instant and potential business opportunities (Tam & Ho, 2006; Maslowska et al., 2016). In the domain of online advertising, personalisation emphasises message contents tailored for internet users or based on their unique preferences or personal information (Baek & Morimoto, 2012). The prime purpose of personalisation is to expose internet users or consumers to advertisements that match their interest by suggesting that the advertising message is directed toward them (Aguirre et al., 2015). As a key feature that sets online advertising from its offline counterparts, personalisation is increasingly used by practitioners in their ODA as it is assumed to offer benefits to both advertisers and consumers (Bright & Daugherty 2012). The literature suggests that personalised advertisements are more engaging because they correspond with consumers’ preferences and interests (Goldfarb & Tucker, 2011; Lambrecht & Tucker, 2013), and are two times more effective than non-personalised advertisements (Tucker, 2014). Although the upshots of personalisation are argued out in the literature, its downsides have been pointed out too. Personalisation grants consumers the opportunity to acquire relevant advertising information without facing random, obtrusive, and irrelevant advertisements (Goldfarb & Tucker 2011), but the tracking of online activities, collection of behavioural data, and information dissemination, are at odds with consumer privacy concerns (Baek & Morimoto 2012). And so, there is a growing stream of studies to indicate that perceptions 79 University of Ghana http://ugspace.ug.edu.gh of intrusiveness are heightened for consumers when personalised advertisements reflect to a great extent their precise preferences (van Doorn & Hoekstra, 2013; Bleier & Eisenbeiss, 2015a) – a phenomenon Bleier and Eisenbeiss (2015a) referred to as “over-personalisation”. For van Doorn and Hoekstra (2013) personalisation is a two-edged sword since it induces higher purchase intentions, and at the same time higher perceptions of intrusiveness which then influences purchase intentions negatively. In order to circumvent the negative upshot and increase the efficacy of personalisation, some studies within the body of literature reviewed called attention to ways of heightening consumer approval of personalised ads. Some of these studies show that: personalisation only fuels purchase if consumers are actively involved in the advertised product/service category or have narrowly construed preferences (Lambrecht &Tucker, 2013) and; internet users with a low desire for control have a greater likelihood (i.e. behavioural intention) to interact with personalised ads than those with a high desire for control (Bright & Daugherty, 2012). Liu and Mattila (2017) also propose a personalisation strategy that targets the psychological motivations of consumers in lieu of using their personal information. As a result, much of the recent literature from diverse perspectives, in identifying the value of personalisation point out that its efficiency hinges on the interplay with placement and various consumer-specific factors (Bleier & Eisenbess, 2015a). What then comes to light is the need to examine some consumer-related conditions under which personalised display ads may be effective in order to sturdily establish its usefulness. vii. Interactivity Interactivity is a defining feature of the internet, and the online environment enables the application of interactive technologies to advertising in this environment. According to Liu and Shrum (2002), interactivity is “the degree to which two or more communicating parties 80 University of Ghana http://ugspace.ug.edu.gh can act on each other, and on the communication medium, and on the message, and the degree to which such influences are synchronised” (p.54). This definition provides a three- dimensional view of interactivity – active control, two-way communication, and synchronicity. For example, a common element in online advertising is the ability of internet users or consumers to click on hyperlinks provided in advertisements through to a brand website (Chatterjee, 2008) permitting them to exhibit control over the flow of content in accordance with their goals (Liu & Shrum, 2002). Two-way communication is also possible in online advertising as brand messages are transmitted to consumers, and advertisers gain feedback through several consumer actions (e.g. clicks, mouse rollovers). Synchronicity as an aspect of interactivity refers to the extent to which physical, spatial and distance impediments are removed in order for immediate responses to be provided to consumers (Liu & Shrum, 2002). Huge strides in technological advancements today, make it possible for advertisers to developed complex ad formats with automated immediate responses to consumer actions or requests (Baron et al., 2014). Online display advertisements with such interactive features are growing in popularity, and the academic literature has given attention to the issue of interactivity as a vital determinant of online display advertising effectiveness. In the stream of literature reviewed, few studies have examined the influence of interactivity on consumer attitudes, and behavioural responses (e.g. Rosenkrans, 2009; Jung et al., 2011; Pashkevich et al., 2012). Pashkevich et al. (2012) studied skippable video ads and established the importance of their interactive and controllable mechanisms in influencing consumer attitude and opinions, as well as reducing negative consequences. Jung et al. (2011) for instance, in an experimental study, included a puzzle game about the brand in an ad for the high interactive group, and the low interactive group surfed a music site that presented a banner ad. The authors established that consumer’ perception of advertising value, specifically entertainment value is influenced by 81 University of Ghana http://ugspace.ug.edu.gh the level of advertising interactivity. Over and above perceptions of value, Rosenkrans (2009) found that rich media ads that are interactive enhance mouse roll-overs and clicks. In the case of Rosenkrans (2009) viewers were asked to move their mouse to perform specific tasks which influenced the likelihood of interaction with the ad. While these studies provide interesting findings regarding the role of interactivity, the experimental designs and the advertising outcomes measured, provide difficulties regarding whether the responses were deliberate. The use of a survey with limited researcher interference would thus, provide further insight into the effect of interactivity (particularly when online ads are perceived as such) on behavioural responses. 4.5.1.2 Consumer-Related Issues Above and beyond ad characteristics, consumer-related factors have also been studied as critical to online advertising effectiveness (Hoban & Bucklin, 2015; Eshghi et al., 2017). Pertinently, the literature under this sub-theme is cognizant of the influence that several factors from the consumer angle may have on advertising effectiveness. Gleaning from the literature, three general factors (consumer involvement, web browsing/user mode, and perceived or perceptions of intrusiveness) represent broader umbrellas of consumer-related factors that have served as antecedents, mediators and moderators of online advertising effects in the reviewed papers and are detailed in the ensuing subsections. i. Consumer Involvement Involvement generally denotes the extent of an individual’s motivation to respond to stimuli as well as focus personal demand, conception and interest toward an object to achieve a desired outcome (Wang & Calder, 2006). According to the Elaboration Likelihood Model, an individual’s level of involvement in the course of “message processing” is a vital element in defining the path to persuasion, given that, compared to consumers with lower 82 University of Ghana http://ugspace.ug.edu.gh involvement, the highly-involved mostly display increased levels of cognition and further information-processing activities (Petty et al., 2005). Studies examining consumer involvement in the body of reviewed literature, often focus on consumers’ involvement with a product category (product involvement) or involvement with the advertisement (advertising involvement). Advertisement involvement is a person’s inherent drive to rationally process the contents of an advertising message (Laczniak & Muehling, 1993). Product involvement, however, describes a consumer’s lasting perception of the relevance of a product on the basis of their innate needs, values as well as interests (Zaichkowsky, 1985). The elaboration likelihood model suggests that involvement induces an attentive state of mind, making consumers more driven and able to process information (Yoo & Kim, 2005). Regardless of the type of involvement (product or advertising), consumers who are highly involved are said to cautiously examine the message’s claims, more determinedly scrutinize these claims, and persevere in their efforts regarding personal factors, such as personal relevance, beliefs and values (Ching et al., 2013). Users who are more involved with an advertised product pay more attention to the ad and process the commercial information more intensively. Extant literature confirms a positive influence of involvement on display advertising effectiveness, in terms of forming lasting attitudes toward the ad and the brand, and on click-through rates (Martin-Santana & Beerli-Palacio, 2012). It also serves as an arbitrating variable in determining the level of an ad’s influence on viewers (Te’eni-Harari et al., 2009). Several instances regarding the role of involvement are identified in the reviewed body of literature on ODA and include Belanche et al. (2017) who found that product involvement increases attitudes toward the ad and toward the brand and reduces the level of intrusiveness; 83 University of Ghana http://ugspace.ug.edu.gh Eshghi et al. (2017) who established that advertising message involvement enhances the influence of ad copy type and task orientation on brand attitude; and Wang et al. (2009) who asserted that variation and appeal strategies that are employed in consort with consumer characteristics, such as product involvement may result in higher advertising effectiveness. Consumer involvement reflects a person’s desire and drive to achieve an outcome, and therefore generally functions as an enhancer or a confounding factor in research on online advertising effects or effectiveness. ii. Web browsing/User Mode Earlier research suggests that the task orientation of consumers affects how they respond to persuasive communication (Rodgers & Sheldon, 1999). The relevance of this in the online environment is underscored by the knowledge that, people use the internet for diverse reasons. Internet usage motive defines “online users’ inner drive to put efforts to carry out any online activity” (Jung et al., 2014, p.2). These motives could come in various forms as research, shopping, entertainment, surfing, communication, or socializing among others. Depending on these motives, internet users deal with, understand, and form attitudes toward advertisements they are exposed to, in different ways (Rodgers & Thorson, 2000). According to the Interactive Advertising Model, information processing is activated when users switch their usage motives, stimulated by ads they see when they are online, and these motives determine their usage/user mode. Because internet users’ online activities mostly begin with a goal-oriented plan, user mode specifies “the extent to which internet activities are goal-directed” (Rodgers & Thorson, 2000, p.46). Within the literature reviewed using reversal theory as a guide, user mode has been broadly classified along a goal-directed spectrum with telic denoting “high goal-oriented, seriousness, and present-oriented” and paratelic representing “low goal-oriented, playful, 84 University of Ghana http://ugspace.ug.edu.gh and activity/experience orentied” (Seyedghorban et al., 2016). Illustratively, an internet user who is researching is debatably, serious-minded because “searching is guided by specific goals, and goal-directed visual search occurs when individuals are motivated to find specific information in order to complete a task” (Kim et al., 2014, p.815); while the other pursuing entertainment is playful and may tend to exhibit curiosity and exploration to satisfy present needs. Thus, the goal of web users can exert strong influence on attention allocation such that ads attract more user attention when users are casually browsing the internet, than when they are involved in for instance, a reading task (Simola et al., 2011). Recently, Jung et al. (2014) examine the role of user mode in consumers’ attitude formation toward interactive online display ads while Seyedghorban et al. (2016) extended and incorporated user mode into Cho and Cheon’s (2004) internet advertising avoidance model. Similarly, Eshghi et al. (2017) investigated the complementary role of individual task orientation on involvement and attitude toward online advertising, and Bleier and Eisenbeiss (2015a) who assessed variations in responses to personalised ads depending on consumers’ web browsing mode. Notably, a common argument echoed in these studies is the essential role user mode or motives play in consumer reactions to advertising in online environments. However, the results from these studies varied across contexts and product categories. Also, the limited empirical marketing or advertising research on the role of user modes in determining consumer behavioural responses to online advertising, provides a good reason to examine how it helps explain online behaviours of consumers with a focus on advertising responses. This is highlighted by the peculiarity of the concept to the online domain. iii. Perceived or Perception of Intrusiveness Perceived intrusiveness is extensively discussed in online advertising literature (McCoy et al., 2008; van Dorn & Hoekstra, 2013), and has been pointed out by Edwards et al. (2002) 85 University of Ghana http://ugspace.ug.edu.gh as an essential factor in determining consumers responses to online advertising. Li et al. (2002) define intrusiveness as “the degree to which advertisements in a media vehicle interrupt the flow of an editorial unit” (p. 39). Thus, explaining perceived intrusiveness as the extent to which users feel removed from their purpose for being online or from their reason for browsing a particular webpage by being “cut-off” by an ad (Rejón-Guardia & Mártinez-López, 2014). So, perception of intrusiveness is a subjective gauge of how distracting an ad is (McCoy et al., 2008), and the manifestation of a means by which the ad evokes irritation and generates emotional responses in users, conceivably compelling them to evade the ad (Edwards et al., 2002). Besides being approached from the perspective of cognitive process and task performance, perception of intrusiveness has also been considered from the viewpoint of interference with individual’s privacy (Rejón-Guardia & Mártinez-López, 2014). In such instances it is suggested that, perceptions of intrusiveness are caused if; consumers are unfamiliar with an advertiser, ads infringe on user’ privacy, and are shown or presented without users’ permission (Milne et al., 2004). Although perceived intrusiveness is a consumer-related issue that impinges on advertising effectiveness, it is a function of several ad-related features, and “describes the mechanism by which an ad induces negative emotional responses, but not the negative responses themselves. Researchers (e.g. Ying et al., 2009) in the field, therefore, assert that perceived intrusiveness of ads can be regulated through manipulation of certain aspects of the ad such as: its value to the user, placement and quality of execution. In their view, if “ads are informative, entertaining, related to the website’s content, and the frequency and quantity of ads are carefully controlled, they will be perceived as less intrusive.” As van Dorn and Hoekstra (2013) also point out in their study, respondents with higher levels of privacy concerns experience greater perceptions of intrusiveness when they are exposed to personalised display ads. Following these, Goodrich et al. (2015) suggest that, it is equally 86 University of Ghana http://ugspace.ug.edu.gh as helpful for advertisers as for website owners that ads are pretested for intrusiveness before full take-off so as to curb negative advertising outcomes. While the current study did not examine perceptions of intrusiveness, it serves as a foundational concept for explaining several of the relationships assessed in this study. iv. Other Consumer/Product-Related Issues There are also some studies within the body of literature reviewed that point out other consumer and product situation in which some advertising strategies may be more or less useful. The arguments put forward by these studies seem to espouse that design and execution of display ads should take into consideration consumer and product peculiarities in order to optimize advertising effects. They include the works of Lambrecht and Tucker (2013)who suggest that dynamic retargeted advertisements, though less effective than their generic counterparts, are more effective when consumers have narrowly constructed preferences – when consumers have detailed knowledge or ideas about the sort of product they intend to buy; Goldfarb and Tucker (2011) who found that context-based and obtrusive ads are effective when used independently, but become less effective when combined, particularly, for product categories considered private and customers who appear to guard their privacy. In the same vein others such as Eshghi et al. (2017) found that the technological intensity of the advertised brand/product influenced consumers advertising message involvement and subsequent attitude toward the brand/product; Goodrich (2013) also demonstrated the influence of age on attention and responses to ODA by indicating that older adults pay greater attention than younger adults to online banner ads as much as they form higher purchase intentions following exposure to such ads. Gathering from the analysis of the literature, and the discussions put forward under this theme, and its ensuing sub-themes, it became apparent that, in ensuring that the literature 87 University of Ghana http://ugspace.ug.edu.gh on online (display) advertising and its effects is progressed and strengthened, it is germane for studies to explore how variables related to both display ads, and consumers or internet users as well as the advertised product/brand may influence consumers behavioural responses toward ODA (Tang et al., 2014; Belanche et al., 2017). Drawing on these discussions, interactivity, placement, informativeness, personalisation and exposure conditions encapsulate the pertinent ad-related features that require further investigation in order that their influences may be resolutely established in the literature (Brajnik & Gabrielli, 2010; Bleier & Eisbensen, 2015a), and are the focus of the study at hand. Additionally, besides being the features that hold more significance for several contemporary ODA formats, they also lend themselves to non-experimental designs. And so, it is important that academic enquiries remain relevant to changes in practice and theoretical understanding of how these pertinent features are perceived by consumers. Additionally, focusing on user mode as the consumer-related (situational) mechanisms that may cause variations in the direction and intensity of their behavioural responses is another area in which this thesis seeks to make a contribution. 4.5.2 Assessing Online Display Advertising Effectiveness This theme concerns literature which examines the approaches for assessing the efficacy of online ads. Higher online advertising saturation and consumer negativity toward online advertisements are some key factors driving the need to assess the effectiveness of online advertising (Martin-Santana & Beerli-Palacio, 2012). On the basis that advertising channels, campaigns and formats may vary, no single measure is regarded as all-encompassing, and so the standards used in most studies are divergent. Studies that inquired into measurement of display advertising effectiveness largely used internet-related or direct-response measures (such as click-through rates, ad impressions etc.) or conventional measures (such as attention toward the ad, attitude toward the ad or brand, ad acceptance, ad recall and 88 University of Ghana http://ugspace.ug.edu.gh recognition, purchase intention and frequency etc.). They include works from authors such as Belanche et al. (2017); Auschaitraku and Mucherjee (2017); Bleier and Eisbensen (2015a; 2015b); Spalding et al. (2009); Rosenkrans (2009); Fulgoni and Morn (2009); Kireyev et al. (2015) as well as Xu et al. (2014) among others. According to Martin-santana & Beerli-Palacio (2012) the conventional measures of assessing the effectiveness of online advertising can be categorised under the traditional cognitive, affective and conative/behavioural (CAB) criteria of advertising effectiveness. Arguing from a similar angle, Hollis (2005) earlier asserted that although the internet-related measures are perculiar to the internet given that the medium has capabilities that extend the function of advertising far past that of traditional media, most of the direct internet-related measures could still be classified under the CAB criteria. He further argues that click- through rate (CTR) for instance, is a form of conative response since it depicts a behavioural reaction “indicative of a desire to check out claims made in the advertsing to ascertain their relevance and veracity” (p.266). From the perspective of the CAB, the review identified memory-based techniques as the most commonly utilised cognitive measures. These cognitive measures include ad/conent recognition, aided/unaided recall, attention and memorisation (e.g. Lee & Cho, 2010; Goodrich, 2011; Bright & Daugherty, 2012; Tujat & Reijmersdal, 2012; Nihel, 2013; Yeu et al., 2013; Goodrich et al., 2015). Studies that employed such measures demonstrate that conventional cognitive measurements offer vital information on internet advertising effectiveness that other forms of measures cannot detect. According to Wakolbinger et al. (2009) assessing online advertising effectiveness using recall for instance, gives advertisiers some assurance that viewers who recollect the ad or the brands and products advertised are more likely to purchase them. It is also argued by Kuisma et al. (2010) that attention and 89 University of Ghana http://ugspace.ug.edu.gh memorisation are essential cognitive responses to online display ads since these may result in other positive outcomes for advertisers. Regarding affect-based measures, the extant studies reviewed point to attitude toward the ad, attitude toward the brand, attitude toward the website and ad liking as key affective indicators of online advertising effectiveness (see Jung et al., 2014; Wang et al., 2013; Goodrich et al., 2015; Auschaitraku & Mucherjee, 2017; Belanche et al., 2017; Eshghi et al., 2017). Arguments raised by these studies are that, attitudes mirror the feelings of consumers during and after ad processing; as such, they improve brand evaluations which could in turn impact brand choice (Goodrich, 2011). Because “affection refers to both attitudinal and emotional aspects of meaning” (Li & Leckenby, 2004, p.5), other researchers (e.g. Van Reijmersdal et al., 2016) also used ad liking as affect-based measures of effectiveness. Van Reijmersdal et al. (2016) for one reasoned that, since persuasion relies on affect-based learning mechanisms such as affect transfer, in lieu of cognitively processing ads, consumers use feelings and liking as processing cues. The position of most studies that used affect based-measures stemmed from early submissions that, ads that are liked (toward which consumers have favourable attitudes) are more successful at persuading consumers than those that are not, thus making affect-based measures essential to marketers and advertisers (Goodrich et al., 2015; Eshghi et al., 2017). 4.5.2.1 Conative/Behavioural Measures According to Li and Leckenby (2004) conation or behaviour is the observable act of consumers or at least their stated desires to act (behavioural intentions). Brajnik and Gabrielli (2010) explain that they are actions or intended actions that suceed changes in cognitive and affective states of consumers exposed to advertising stimuli. Evidence from the reviewed ODA literature shows predominant use of CTR or ad clicking as well as 90 University of Ghana http://ugspace.ug.edu.gh purchase intentions as conative or behavioural measures (see Rosenkrans, 2009; Chan, 2010; Goldfarb & Tucker, 2011; Bleier & Eisenbeiss, 2015b; Van Reijmersdal et al., 2016; Liu & Mattila, 2017; Kim et al., 2018). Other studies also utilised shopping frequency (e.g. Wang & Sun, 2010a; 2010b; Nasir, 2017), intentions for ad interaction (Bright & Daugherty, 2012), ad viewing (Simola et al., 2011; Bleier & Eisenbeiss, 2015a), and ad acceptace (Belanche et al., 2017). Regarding the predominantly used behavioural measures, the prevailing argument has been that when consumers are persuaded by advertising message, they form the desire or intention to purchase the advertised brand; and since purchase intention is a commonly used and well established conative measure of advertising effectiveness particularly in traditional media, extending it to the online context further provides added insight regarding its role in gauging the effectiveness of ads (Wang & Sun, 2010c; 2010d). The other prevalent behavioural measure, CTR or ad clicking is also lauded as a key measure for assessing online advertising effectiveness (Bleier & Eisenbeiss, 2015b) because relative to other internet-related measures (e.g. ad impression), CTR is considered a more relevant and performance-centered measure (Rosenkrans, 2009; Nihel, 2013); and for this reason, has been used as one of the major behavioural response measures by majority of the reviewed studies. In spite of all these studies, there are still significant challenges in gauging online advertising effectiveness (Goldfarb, 2014). The extreme use of the CTR for instance, has been probed by some researchers (Yaveroglu & Donthu, 2008) who argue that it focuses singularly on an active response to online ads and not possible exposure to it. Other authors such as Lavrakas et al. (2010) have also pointed to the inadequacy of CTR on the basis of its inablility to measure cognitive and affective effects as well as other behavioural responses 91 University of Ghana http://ugspace.ug.edu.gh consumers may engage in without clicking on ads. This explains the rationale behind studies that balance results gained by CTR, with those obtained using conventional advertising outcome measures (e.g. Spilker-Attig & Brettel, 2010; Song et al., 2011; Martin-santan & Beerli-Palacio, 2012; Liu & Mattila, 2017). Empirical support for this position is provided by Fugoni and Morn (2009) who demonstrate that notwithstanding lack of clicks, display ads could positively induce consumers to visit advertiser websites, perform search queries using advertiser brand name, and make both online and offline purchase of advertised brands. In providing further explanations to issues plaguing measures of effectiveness, Lewis et al. (2011) argue that consumers engage in several activities (watch ads, search for products and purchase items among other things) while online. In their view, therefore, studies that measure online advertising effectiveness by correlating user responses and online ads stand the risk of overstating the efficacy of such ads because there is bound to be a particular type of selection bias (activity bias). Following this perspective, Xu et al. (2014) developed a model that underscores clicks and purchases as dependent arbitrary occurrences in continuous time and, hence, suggest the need to deemphasize the “last click” effect and focus on the indirect contributions which other ads (or formats) may make toward final conversion. Kireyev et al. (2015) in a recent study, along the same line of reasoning, contend that simple stationary measures are plagued with a fundamental attribution problem and, therefore, suggested dynamic versions of the classic metrics. While numerous researchers suggest their own conceptualizations of how consumers react to online advertising (which are considered indicators of online advertising effectiveness), some others (e.g. Ha, 2008; Spalding et al., 2009; Wakolbinger et al., 2009) recommend that the suitability of effectiveness measures should be done on the basis of a firm’s 92 University of Ghana http://ugspace.ug.edu.gh advertising objectives or goals. Behavioural response is not only one of the major objectives of online advertising, but also one of the three dimensions of the hierarchy of effects’ CAB criteria of effectiveness (Li & Leckenby, 2004) as discussed early on. However, the reviewed ODA literature depicts a prevalence of affective measures which seems consistent with Kim et al. (2014) who also found an increasing prominence of affective effects in advertising research. This current research, therefore, considers it expedient to examine consumers’ responses to online display advertising from a behavioural perspective. Aligning with researchers such as Lavarakas et al. (2010) and Pavlou and Stewart (2000), who assert that because individuals play active roles in the online environment, the effectiveness of online advertising needs to be examined from perspectives that measure not only the response to the ad but also interactions with it, this thesis examines behavioural responses to ODA by focusing on consumers’ approach and avoidance behaviours toward online display ads. Specifically, in this current study behavioural response was measured as “ad acceptance and ad avoidance” which examined the extent of consumer attentiveness, focus and interaction with ODA in the case of ad acceptance and lack of attentiveness, focus and interaction in the case of ad avoidance. In so doing this study takes into consideration both the direction and intensitiy of the behavioural response. 4.5.3 Consumer Attitude toward Online Advertising (ATOA) The stream of studies under this theme, concerns consumers’ perception, beliefs and general predisposition toward online advertising. Described as the evaluative sum of perceived attributes, as well as affective and cognitive benefits of online advertising (Wang & Sun, 2010b), attitude toward online advertising is topical in the literature owing to its role in predicting consumers’ responses and behaviours to online advertising after exposure (Souiden et al., 2017; Wang & Sun, 2010d). This literature stream diverges from earlier submissions that consider attitudes as analogous to and substitutable with beliefs (Mehta, 93 University of Ghana http://ugspace.ug.edu.gh 2000); and agrees with assertions that attitudes toward online advertising are outcomes of consumers’ beliefs about online advertising (Bracket & Carr, 2001). Subsequent to the initial work of Bracket and Carr (2001), several authors have made further contributions toward the latter assertion which holds that consumers’ attitudes toward online advertising are formed or conditioned by their beliefs (e.g. Mahmoud, 2014; Nasir, 2017). Beliefs about online advertising from the perspective of this body of literature, therefore, are specific statements about the attributes of online advertising (Wang et al., 2009). This position finds empirical backing in studies (such as Wang et al., 2009; Wang & Sun, 2010a; 2010b; 2010c; 2010d) which investigated five belief dimensions (entertainment, information seeking, credibility, economy and value corruption) and found, in most instances, that all are significant predictors of attitude toward online advertising with variations across several cultural contexts. Other works (such as Li-Ming et al., 2013; Sun &Wang, 2010) have assessed other antecedents such as familiarity, usability and trust. Beyond studying attitudes toward online advertising as an outcome of antecedents like beliefs, familiarity, usability and trust as well as other ad-related features, these studies further investigated the influence of attitudes toward online advertising on consumer responses such as ad clicking, brand attitude and preferences, and online shopping intention and frequency (e.g. Wang et al., 2009; Sun & Wang, 2010; Wang and Sun, 2010b; Goodrich, 2011). Others examined the attitudinal and behavioural differences across web user segments, cultures, genders, age as well as personality traits (Goodrich, 2013; 2014; Valaei et al. 2016; Nasir; 2017; Souiden et al., 2017). In essence, studies within this theme recognize that while consumers’ attitudes toward online advertising is considered a key outcome of certain pertinent precursors, it is also a vital antecedent to several behavioural responses, and these interrelationships are regulated by individual, demographic, cultural, 94 University of Ghana http://ugspace.ug.edu.gh and ad-specific differences, which should be of utmost concern to researchers and practitioners alike. Altogether, the central role of attitude in online advertising effects cannot be ignored as it appears the most frequently measured construct, either as an outcome in most cases or mediating variable in very few cases, derived from the understanding that attitudes are antecedent to behaviour and serve as an intervening response, determining the degree of online advertising influence on consumers or viewers (Wang & Sun, 2010b; Goodrich, 2011). Not surprisingly, although attitude toward advertising (ATA) in general has long been regarded a vital factor in determining advertising effectiveness (Lutz et al., 1983), attitude toward online advertising (ATOA) is only as old as the phenomenon of online advertising (Zha et al., 2015). Additionally, the experimental nature of most of the reviewed studies causes a tendency of the researchers to examine attitude toward specific ads, rather than ATOA with the exception of a few (e.g. Wang & Sun, 2010a; Valaei et al., 2016; Souiden et al., 2017). Also worth noting, is how silent extant studies have been on the dual (explicit and implicit) perspective of attitude in spite of its potential in explaining more vividly consumer behaviours (Goodrich, 2011). Given the relatively brief history, and limited empirical evidence on consumers’ ATOA, understanding more about its dynamics and role in enhancing or explicating behavioural responses of consumers toward online display ads is crucial for theory advancement. Specifically, this thesis examines ATOA as an internal consumer response that transmits or facilitates the effects of ad characteristics on consumer behavioural responses to ODA. 4.6 THEORETICAL APPROACHES IN ODA RESEARCH This section examines and discusses the conceptual or theoretical routes taken in the ODA studies that were reviewed. This was done in an attempt to appreciate the theoretical 95 University of Ghana http://ugspace.ug.edu.gh perspectives underpinning these studies as well as explore the representation of the varied theoretical stances identified from the review. Using the classification scheme provided by Heeks and Bailur (2007) which depicts the degree of theorization along a spectrum from profound theory-based through shallow category-based approaches, the study differentiates the conceptual approaches employed in these studies. From this standpoint: o theory-based studies make use of identified theories, through application or testing; o framework-based studies employed frameworks explicitly derived from a stream of theoretical works; o model-based studies applied models without alluding to any deeper knowledge stream; o concept-based works used non-theoretically grounded but defined concepts and; o category-based studies used sets of predefined factors to conduct analysis. Table 4.3 maps the reviewed articles by theoretical approaches and geographical setting. The theoretical basis of the studies suggests that, 24 articles (approximately 38%) made recourse to theories or stated specific theories on which their works were grounded. These studies employed such theories as the elaboration likelihood model (Spilker-attig & Brettel, 2010; Wang et al., 2009), stimulus organism response model (Tang et al., 2014; Bleier & Eisenbess, 2015b), reactance theory (Ying et al., 2009), processing fluency theory (Wang et al., 2013), goal system theory (Jung et al., 2011), mere exposure effects theory (Goodrich, 2011; 2014; Yeu et al., 2013), reversal theory (Jung et al., 2014; Seyedghorban et al., 2016), and central capacity theory of attention (Simola et al., 2011) among others. As reflected in the review, most of the featured theories are drawn from the field of psychology. Overall, the elaboration likelihood model (ELM) by Petty et al. (1983), reactance theory by Brehm (1966), mere exposure effect theory by (Zajonc, 1968), Stimulus organism response model (SOR) by Mehrabian and Russell (1974), and reversal theory by Apter (1984) were the five theories that were used in more than a single study. Specifically, ELM was utilised in four 96 University of Ghana http://ugspace.ug.edu.gh studies, reactance and mere exposure effect theories were used in three studies each, and reversal theory, and SOR were used in two studies each. These theories (as shown in table 4.3) were either used by some studies in an explanatory capacity and/or integrated with other concepts or frameworks to examine online display advertising from various perspectives. Other studies in most cases, merely alluded to these theories in their discussions. The review’s consideration of a study as being theory driven was on the basis of these three instances. Aside the few theory-based studies, the remaining 39 papers, though are not based on identifiable theories, used concepts, constructs and frameworks within the broader advertising literature as well as those peculiar to online advertising to provide valuable insights on ODA and its outcomes in terms of consumer responses. The apparent low level of theory-based studies in the ODA literature resounds the outcome of Pitt et al’s. (2005) audit on theory use in advertising research from three advertising journals which found merely 17% of articles utilizing specific theories. Again, Kim et al. (2014) studied trends in advertising research and found a little less than half of their study sample to be theory- driven. Though this may represent a growing recognition of the role of theory in advertising research, this progress is minimally reflected in the ODA papers reviewed. For Faber (2015), as an applied and interdisciplinary field, there is a degree of latitude to borrow and adapt theories from other domains. Adding his voice to Faber’s view, Laczniak (2015) asserts that advertising research should be driven by theory and calls for empirical testing of these theories in diverse context in order to determine the degree of their generalisability. Keeping on course with the purpose of this review, the following section examines the methodological approaches employed in the body of literature reviewed. 97 University of Ghana http://ugspace.ug.edu.gh Table 4.3 Distribution of Articles by Geographical Setting and Theoretical Approaches Region/Country Article Theoretical Approach Americas o USA Auschaitraku & Mucherjee (2017) Framework-based Bright & Daugherty (2012) Theory-based (EVM) Bruce et al. (2016) Model-based Fulgoni & Morn (2009) Category-based Goodrich (2014) Theory-based (MET) Goodrich (2013) Concept-based Goodrich (2011) Theory-based (MET & DAM) Goodrich et al. (2015) Category-based Hoban & Bucklin (2015) Model-based Jung et al. (2011) Theory-based (GST) Kireyev et al. (2015) Model-based Kim (2018) Theory-based (ACT) Liu & Mattila (2017) Concept-based Pashkevich et al. (2012) Model-based Rosenkrans (2009) Theory-based (DT) Segev et al. (2014) Concept-based o Canada Souiden et al. (2017) Category-based o Ecuador Flore et al. (2014) Category-based Europe o Netherlands Tutaj & van Reijmersdal (2012) Framework-based Van Dorn & Hoekstra (2013) Category-based Van Reijmersdal et al. (2016) Category-based o Spain Belanche et al. (2017) Category-based Martin-Santana & Beerli-Palacio (2012) Category-based o Belgium Janssens et al. (2012) Framework-based o Germany Spilker-attig & Brettel (2010) Theory-based (ELM) o Greece Drossos et al. (2011) Category-based o Norway Ying et al. (2009) Theory-based (RT) o Finland Kuisma et al. (2010) Theory-based (IPT) o Germany Bleier & Eisenbess (2015b) Theory-based (SOR) Asia o China Li et al. (2009) Category-based Song et al. (2011) Category-based Wang et al. (2009) Category-based Nihel (2013) Category-based o South Korea Yeu et al. (2013) Theory-based (MET) Kim et al. (2011) Model-based o Taiwan Wang et al. (2009) Theory-based (ELM) Heish & Chen (2011) Theory-based (ELM & SAT) o Malaysia Li-Ming et al. (2013) Theory-based (TAM) Valaei et al. (2016) Theory (HCD) o India Eshghi et al. (2017) Category-based o Singapore Chan et al. (2010) Theory-based (MT, PFT & PKM) Middle East o Iran Seyedghorban et al. (2016) Theory-based (RET) o Turkey Nasir (2017) Category-based o Syria Mahmoud (2014) Category-based Australia 98 University of Ghana http://ugspace.ug.edu.gh o Hussain et al. (2018) Theory-based (BTT) Cross-Cultural o USA & China Sun & Wang (2010) Category-based Wang & Sun (2010a) Category-based o USA & Romania Wang & Sun (2010c) Category-based o China & Romania Wang & Sun (2010d) Category-based o USA, China & Romaina Wang & Sun (2010b) Category-based Silent on Region o Bleier & Eisenbeiss (2015a) Model-based o Chapelle et al. (2014) Model-based o Fridgeirsdottir & Najafi-Asadolahi (2018) Model-based o Jung et al. (2014) Theory-based (RET) o Lambrecht & Tucker (2013) Theory-based (CLT) o Miralles-Pechuan et al. (2017) Model-based o Simola et al. (2011) Theory-based (CCTA) o Tang et al. (2014) Theory-based (SORM & RT) o Kim (2018) Theory-based (ELM & RT) o Lee & Cho (2010) Theory-based (BTT) o Xu et al. (2014) Model-based o Wang et al. (2013) Theory-based (PFT) Note:DT=Distinctiveness Theory; RT=Reactance Theory; ELM=Elaboration Likelihood Model; MT=Mindset Theory; TAM=Technology Acceptance Model; HCD=Hofstede’s Culture Dimensions; SAT=Selective Attention Theory; RET=Reversal Theory; EVM=Expectancy Value Model; GST=Goal Systems Theory; IPT=Information Processing Theory DAT=Dual Attitude Model; MET=Mere Exposure Effect Theory; BTT=Berlyne’s Two-factor Theory; ACT=Advertising Contextual Theory CLT=Construal Level Theory; SORM=Stimulus Organism Response Model; CCTA=Central Capacity Theory of Attention; PKM=Persuasion Knowledge Model; PFT=Processing Fluency Theory 4.7 METHODOLOGICAL APPROACHES IN ODA RESEARCH Generally, all articles reviewed were empirical with a high proportion of quantitative studies representing 95 percent of the total number. The review records only one qualitative study (Tang et al., 2014), and two mixed methods studies (van Doorn & Hoekstra, 2013; Drossos et al., 2011). The quantitative studies largely employed experiments, which is the most common or favoured method in advertising and online advertising research (Chang, 2017; Ha, 2008). Nineteen of the thirty-seven (approximately 60%) experimental studies were controlled experiments while eighteen were field experiments. A few of these studies conducted surveys subsequent to their experimental designs as a means of measuring behavioural intentions such as recall, revisit and purchase intentions (e.g. Goodrich et al., 2015; Janssens et al., 2012). Moreover, fourteen articles (approximately 22%) used surveys, and eight articles (approximately 13%) used secondary market data. Most of the surveys 99 University of Ghana http://ugspace.ug.edu.gh were conducted online owing to time and cost efficiency considerations as well as ease of recruiting compared to offline surveys since the targeted respondents are generally web or internet users. Relevantly, the experimental studies and those that utilized secondary market data, were prevalent in the ‘antecedents to online advertising effectiveness’ and ‘measures of online advertising effectiveness’ themes as these studies observed user activities and measured actual cognitive and behavioural responses (such as clicks, shopping frequency, ad watching time) to ads (e.g. Belanche et al., 2017; Bleier & Eisenbess, 2015a; Pashkevich et al., 2012). The survey studies, however, were dominant under the online advertising attitudinal theme (See Appendix A1). The qualitative study (Tang et al., 2014) represented in this review, categorised consumer online behavioural responses into more distinct and polished categories. Through a content and thematic analysis of primary data gathered, Tang et al. (2014) conceptualised a two- dimensional framework that combined the behavioural direction and intensity of the behavioural effort and generated four types of behavioural responses - active approach, passive approach, active avoidance, and passive avoidance. The two mixed-method studies (Drossos et al., 2011; van Dorn & Hoekstra, 2013) included in this review, also provide insightful contribution to ODA research. In their study, van Dorn and Hoekstra (2013) examined how personalisation triggers perceptions of intrusiveness, and its effect on consumer purchase intentions. The study used a sequential mixed-methods approach in which a qualitative pre-study was conducted through 12 in-depth interviews with managers and focus group interviews with customers to assess their experiences and attitude toward personalised or customised ads. The next stage was an online experiment involving 233 participants on a consumer panel. The study highlighted that; highly personalised ads can be a two-edged sword heightening purchase intention but may also negatively affect purchase intentions through higher perceived intrusiveness. Conventionally, qualitative 100 University of Ghana http://ugspace.ug.edu.gh studies are suitable when the phenomenon or field of enquiry is embryonic (Yin, 2009), following which quantitative approaches are typically applied in an attempt to validate these qualitative findings and propositions (Malhotra, 2010). A plausible explication, therefore, for the bias against qualitative studies in ODA is that online advertising, as a field of scholarly enquiry, has a history of two and a half decades and has received considerable exploration and conceptualisation in the past years. Sampling methods show that respondents or participants recruited for surveys and experiments have largely, been college students - accounting for 42 percent of these studies. Although they do not form a representative populace of all internet users, students are considered as more conversant with the internet and seem more easily accessible (Bright & Daugherty, 2012). Also, regarding the methods of analysis employed, most of the papers applied first generation methods of statistical analysis, reporting frequencies, t-tests, ANOVA, or regression analysis (e.g. Tujat & Reijmersdal, 2012; Goodrich et al., 2015; Nasir, 2017), as well as more advanced data analysis methods such as economic, mathematical, and structural equation modelling (e.g. Valaei et al., 2016; Miralles-Pechuan et al., 2017; Fridgeirsdottir & Najafi-Asadolahi, 2018). Given their various advantages over first generational methods, advanced and more contemporary streams of analyses, could be used more by researchers to provide sturdy empirical support and validation to research findings. The Figure 4.3 shows the methodological approaches and research strategies employed in these papers. Studies classified under secondary market data are those that used data collected from weblogs and other online databases of firms, as well as advertising and data mining agencies without any prior influence from researchers. These studies are distinguished from field studies, which design or influence ad formats and campaigns for 101 University of Ghana http://ugspace.ug.edu.gh the purposes of the research in real life environments; and controlled experiment which are conducted in laboratory settings. In the subsequent section, discussions on the geographical spread or coverage of ODA literature under review are presented. Figure 4.3 Distribution of Articles by Methodology 20 18 16 14 12 Survey 10 Field Experiment 8 Controlled Experiment 6 Secondary Market Data 4 Content Analyasis 2 0 Quantitative Qualitative Mixed-Methods Methodology 4.8 GEOGRAPHICAL AND ECONOMIC DISPERSION OF ODA RESEARCH As indicated in Figure 4.4 the distribution of the reviewed articles shows greater attention on developed economies (approximately 56%) with a paucity of evidence from developing economy settings (17%). Specifically, there is a predominance of studies from North America, Europe and Asia; with less representation from the Middle East, South America, Australia and Oceania, and no apparent representation from Africa (see Table 4.3). The country focus of these studies fluctuates as well, with a substantial amount of North America studies domiciled in the USA. The seeming absence of studies from Africa may be attributed to the embryonic state of internet technology usage in most part of the region (Fuchs & Horak, 2008). Although the uptake of internet-related technologies by firms and the general populace is increasing; as indicated by the internet penetration rates, access to and internet 102 Number of Articles University of Ghana http://ugspace.ug.edu.gh usage in most African countries still lag behind developed economies and even other developing middle-income countries. There were a number of trans-regional studies which covered more than one country or region, as well as others which did not clearly state the context or geographical setting in which the studies were conducted and so, were labelled ‘silent on region’. The reviewed articles also, depict a high number of multiple authorships; 55 percent of the articles had more than two authors, 31 percent had two authors and only 14 percent were single- authored. Following the discussions so far, a number of research gaps have been identified and are discussed in the succeeding section. Figure 4.4 Distribution of Articles by Economic Setting 35 56% 30 25 20 15 17% 19% 10 8% 5 0 Developed Developing Cross-Region Silent on Region Economies Economies Economic Setting 4.9 RESEARCH GAPS AND FUTURE RESEARCH AVENUES Following from the discourses on the findings from the review, a number of gaps are apparent in the literature, which prompt several avenues for future research, some (not all) 103 Number of Articles University of Ghana http://ugspace.ug.edu.gh of which provide a springboard for this thesis. Despite being wide and varied (see Appendix A4), these are organised along three strands: gaps in issues and evidence; theoretical and contextual gaps; and gaps in methodological approaches. 4.9.1 Gaps in Issues and Evidences This section identifies six major areas which could benefit from further research. Foremost, the constant emergence of new advertising formats presents an opportunity for additional investigations to establish their usefulness (Belanche et al., 2017). Some relatively nascent ad formats and forms such as advergames, native advertising, in-stream videos and skippable videos have received scant attention (de Pelsmacker & Neijens, 2012; Goodrich et al., 2015). From one angle, besides Burns and Lutz’s (2008) work about a decade ago, it would seem no research has provided a comprehensive investigation on a diversity of ad formats in a single study. Importantly, because these formats may become obsolete or “fall out of favour”, a comprehensive study of them (as has been done with older formats) may point out the pertinent design elements or features that may serve as enduring avenues for future studies as well as guide the design of newer formats (de Pelsmacker & Neijens, 2012). Much more relevantly, several design elements and/or executional features are pointed out in the literature. The findings regarding their effects or influences in the literature are conflictive and less disciplined. It is argued that the fragmented and inconclusive results are attributable to the investigation of different ad stimuli, methods, measures, audiences and contexts (Kuisma et al., 2010), which make such studies hardly comparable. As such, there is a need for future studies to examine a multiplicity of these features within a single investigation to provide a broad base for comparison (Spilker-attig & Brettell, 2010) as well as more precise and concerted postulations of their effects. From the perspective of some researchers (Rosenkrans, 2009; Bleier & Eisenbeiss, 2015b; Hussain et al., 2018), explicit 104 University of Ghana http://ugspace.ug.edu.gh focus on issues of personalisation, exposure, placement, interactivity require more dedication from future research in order to clarify the varied perspectives and establish more resolutely their influences. Authors such as Brajnik and Gabrielli (2010) and Kim (2018) have also called for studies addressing issues concerning the interactions between these features as the way forward. Besides considerations of ad format and design elements or executional features, a few studies suggest that the presence of different types of ads on a webpage affect consumer responses (Lewis et al., 2011). Because the internet is a stimulus-dense environment, commercial webpages selling advertising space feature numerous ads on a single web page and run multiple ads during commercial interruptions. However, what occurs when a webpage hosts multiple interactive, personalised, and rich media ads is presently unclear (Brajnik & Gabrielli, 2010). Since the likelihood that the effects stimulated by some of these ads trickle to other ads as well as the hosting website (Rosengren et al., 2013), future research needs to examine consumer responses to a sequence of online ads rather than a single ad as well as competitiveness between ads on a website and across specific online platforms such as gaming and social networking platforms (e.g. YouTube, Facebook, Twitter etc.) (Liu & Mattila, 2017; Kim, 2018; Liu-Thompkins, 2019). Second, the consumer-related issues identified in the review as precursor to, and intervenors or confounders of the effects of online display advertising are skewed. More focus has been given to issues of intrusiveness, attitude, and involvement and privacy concerns among other things. The internet is a goal-directed medium, a vital feature besides interactivity that sets it apart from conventional media (Wang et al., 2009) and so, one would expect considerable amount of extant works to give attention to user motive/mode or goal orientation as a user-related or situational variable in examining online advertising 105 University of Ghana http://ugspace.ug.edu.gh influences. However, how user mode or goal orientation influence consumer perceptions, attitude and responses to online display advertising has seldom been examined (Kuisma et al., 2010; Seyedghorban et al., 2016). Future studies are thus needed to lend further insights into the role user modes or goal orientation plays in consumer responses to online display advertising (Bleier & Eisenbess, 2015b). Concerning the third gap, attitude toward online advertising has so far been studied from a unidimensional explicit perspective. A majority of extant online advertising studies that considered the attitude-behaviour relationship (e.g. Wang & Sun, 2010b; 2010c; Souiden et al., 2017) so often presented a composite view on the role of attitude as a mediator of the beliefs/perceptions-behaviour nexus. It has, nonetheless, been argued in a recent study (Goodrich 2011) that the existence of dual attitudes toward online advertising is possible; an implicit attitude formed without conscious awareness and control, and an explicit attitude based on intentionally generated evaluations. Because implicit attitudes are different from their explicit counterparts on the basis of their formation, storage, retrieval and operations, a common theme that runs through recent streams of internet-related studies (e.g. Serenko & Turel, 2018) and that of Goodrich (2011) is the need to validate implicit attitude as a major attitudinal dimension germane to behaviour or behavioural responses, and not explicit attitude alone. Thus, necessitating the need to examine both perspectives in future studies to help explain more vividly, the role of attitude in influencing consumer responses to online advertising. Others also call for a focus on attitude toward online advertising in general, rather than attitude toward specific ads or formats based on the scope of such studies (Souiden et al., 2017). Fourth, the focus of the measures of effectiveness of online advertising has been on short- term effects of ad campaigns. However, as Breuer and Brettel (2012) suggests, the carryover 106 University of Ghana http://ugspace.ug.edu.gh and long-term effects of advertising differ significantly from its short-term effects. With behavioural response and brand building as the two major objectives for advertising (Li & Leckenby, 2004), and the substantial use of online advertising by practitioners for brand- building purposes (Hollis 2005), future studies need to provide a much-enhanced understanding of the long run effect of online advertising campaigns on consumers. An essential question to provide answers to in this regard could be how the effect of an online ad exposure wanes/decays over time. Some extant research suggests that the degree of decay may vary by ad format (e.g. Breuer et al., 2011); driving the need for more substantive enquiries of the decay pattern (Liu-Thompkins, 2019). Fifth, there appears to be very few works focusing on service offerings (e.g. Kireyev et al., 2015; Seyedghorban et al., 2016; Liu & Mattilla, 2017) as most of the studies span different categories of physical products. Given the inherent peculiarity of services, it would be enlightening to know how these idiosyncrasies confound advertising of such in online environments to help improve research and advertising practice. Of particular value would be works that investigate the variations that exist in the ad outcomes for the different service processing categories: people (e.g. healthcare), possession (e.g. vehicle repair or maintenance), mental stimulus (e.g. education), and Information (e.g. banking) as has been done for physical products by majority of the reviewed studies (e.g. Flore et al., 2014; Eshghi et al., 2017). Also, various contrasts exist in the broader marketing literature, one of which is the product-service dichotomy. Springing on this knowledge, the review indicates that studies attempting to understand the differences and disparities that might exist in the advertising of service and product offerings in the online domain are absent. It would be instructive, therefore, for future research to examine this disparity (or lack thereof) in the ODA context. 107 University of Ghana http://ugspace.ug.edu.gh Finally, another area of neglect is practitioner or industry perspective. As mirrored by this review, internet users and consumers’ attitude and perceptions toward online advertising is prevalent in the literature. However, the views of practitioners are almost absent (Drossos et al., 2011). It has been pointed out that such practitioners as website owners, marketing managers, advertisers and advertising agencies, among others, share and hold varying viewpoints (Knoll, 2015). As such, future investigations that delve into this pertinent area particularly, in diverse countries on the basis of internet penetration and maturity rates, may offer direct managerial implications, as well as contribute to theory building (Drossos et al., 2011); since the practitioner perspectives may be juxtaposed with the scholarly viewpoints to find points of convergence and divergence (Ha, 2008). 4.9.2 Theoretical and Contextual Gaps As pointed out in an earlier section, besides a little over a third of the reviewed articles, which clearly stated the theoretical foundations of their work, the literature depicts a general deficiency of studies grounded in established theories. In lieu of explicitly averring their theoretical underpinning or developing testable hypotheses from such theories, most of the studies commonly stated research questions and focused on specific factors or constructs and their interrelationships using prior empirical findings. These problem-driven and phenomenon-based approaches raise pertinent concerns from two perspectives. First, in view of the rapid advancements in the domain of online media, such studies run the chance of becoming dated. In addition, outcomes of such studies are restricted by the particular stimuli, message type, the category of product, and consumers, and so their applicability and generalisability across diverse market settings besides the initial study contexts become tricky (Goodrich et al., 2015). In light of the above, researchers such as Knoll (2015) and Johar (2016) have called attention to the need for future studies to examine theories of how advertising works by developing research questions on more abstract levels and deriving 108 University of Ghana http://ugspace.ug.edu.gh hypotheses from established theories. Owing to the ‘variable’ nature of advertising and its reputation for borrowing theories from older fields (Faber, 2015), other theoretical perspectives could be examined in the online advertising display literature. Rather recently, Lackzniak (2015) re-echoed the need for universal theories to be contextualised to advertising studies, as depicted by the few theory-based studies identified in the review. Although most of these theories (e.g. ELM, MET, PKM) represented in the review have seen fair usage in both offline and online advertising contexts, some have not seen much usage and representation in the online advertising literature. For instance, from an SOR model perspective, there is a lack of theoretically driven understanding of the relationship between ODA characteristics and consumer attitude and behavioural responses. Because theories should provide explanations to why diverse individuals may respond to an ad in a particular or different way, and also be helpful in allowing advertisers to design ad messages that have anticipated influences on receivers from both a firm and societal perspective, Lackzniak (2015) avers that theories should be extended from other domains and applied to the field. For Faber (2015), where some theories appear deficient in providing explanations, they should be integrated with others to offer a more comprehensive view on issues. What is more, empirical testing of theories particularly, in diverse geographical contexts to determine the degree of their generalisability is an essential aspect of scholarly enquiry. Such tests and conclusions derived from them, bring perspective to theoretical expectations, and extend theoretical limits of fields (Lackzniak, 2015). In spite of the borderless nature of online advertising, it is still crucial to inquire about how applicable theoretical perspectives may be across diverse countries (Janssens et al., 2012). Interestingly, articles that did not disclose the geographical setting of their study, showed a high representation of theory- 109 University of Ghana http://ugspace.ug.edu.gh based works, followed by Asian, North American, and European studies. As it would seem, theory-based works were non-existent, in cross-cultural research works (see Table 4.3). For an applied field like advertising, the contextualisation in theory development is needed to provide researchers with the prospects to account for certain idiosyncrasies that may offer meaningful insights to scholars and practitioners alike (Faber, 2015; Lackzniak, 2015). Still speaking from a contextual standpoint, the reviewed studies focused more on developed market/country settings to the neglect of developing contexts (Lascu et al., 2016). Also, the developing context studies were mainly from Asia and the Middle East with seemingly no African representation (See Table 4.3). Developing countries or markets differ in several ways from developed markets. However, differences exist among emerging or developing markets as well because such markets in Eastern Europe for instance, may vary in culture, economic development and other institutional conditions from those in sub-Saharan Africa. For this reason, researchers like Seyedghorban et al. (2016) and Eshghi et al. (2017) have called for findings to be tested across various geographical settings in order to progress understanding of online advertising influences and outcomes. Consequently, there is room for contribution to be made to the academic literature from a developing/emerging sub- Saharan African country perspective in both theoretical and practical terms. Lastly, although some studies (e.g. Wang & Sun, 2010b; 2010c) have provided cross cultural comparisons of consumer perceptions and responses to online advertising, these studies are sparse, and the cultural perspectives appear limited. Since culture and advertising are so intertwined, future research need to broaden the cross-cultural examination of online advertising influences in order to advance present understanding of the impacts of global and local cultures on online advertising strategies (De Mooij & Hofstede, 2010; Bleier & Eisenbess, 2015b; Valaei et al., 2016). 110 University of Ghana http://ugspace.ug.edu.gh 4.9.3 Gaps in Methodological Approaches The methodological gaps in the reviewed studies are along three strands: the increasing dominance of experimental quantitative studies causing a dearth of qualitative studies; the prevalent use of student samples and; lack of longitudinal research. Online advertising research is dominated by quantitative approaches, particularly controlled experimental designs. While experiments provide the leeway to collect data unobtrusively and are free of self-reporting predisposition in real media atmospheres (Chang, 2017), the behaviour of respondents in laboratory experiments are considered falsified, given their knowledge of the research activity. Again, it may be argued that experimental designs allow the influence of extraneous factors to be controlled (Hussain et al., 2018), but it is also relevant to point out that cognitive and behavioural activities of people are stifled or restrained in controlled laboratory environments. In “normal” daily situations, people’s behaviours have a greater degree of variations, and such behaviours can be examined through observations or self- report data (Tang et al., 2014). Even more so, researchers like Brajnik and Gabrielli (2010) have called for studies that use actual ads, particularly, most commonly used or seen ones within the study area in lieu of temporary ones developed for the purposes of a specific study. In doing this, future studies could heighten the ecological validity of their findings and provide a more reliable depiction of users’ experience with, perceptions of, and responses to online advertising. In view of this, more field experiments, tracking studies (or use of secondary data) and surveys are, encouraged as they offer more uncontrived objectivity (Kim, 2018; Lewis & Rao, 2015). The general lack of qualitative studies may be attributable to the neglect of certain areas of enquiry and geographical context. For instance, practitioner perspectives on ODA received little attention and were represented by only two studies from a developed, and a middle- 111 University of Ghana http://ugspace.ug.edu.gh income economy (Drossos et al., 2011; Li et al., 2009). What is more, these studies did not provide much avenue for propositions to be made as they were mainly descriptive. In essence, more exploratory studies from developing contexts, where internet advertising is relatively nascent are required to serve as a springboard for further quantitative studies. The second methodological gap concerns sampling. The essence of sampling in any research is generalisability, and this is even more so for the advertising field because of its closely- knit link with practice. Sampling has been an essential methodological problem ailing online advertising research (Ha, 2008; Knoll, 2015) as is once more reflected in the reliance on college/university students by the majority of studies in this review. Given the torrent of evidence that variations exist in consumer attitude toward advertising in terms of several factors (e.g. age, education, outlook on technologies, usage rates, and familiarity with the internet etc.), it is essential that studies use diverse and broader respondent populations as possible, especially when the explanatory power of predictors are interacted with respondents’ characteristics (Gao et al., 2009). Albeit youthful adults with ‘high educational levels’ are a key segment for advertising, other segments exist that have not been considerably studied so far (Brajnik & Gabrielli, 2010). As the representativeness of student samples has been questioned and the applicability and external validity of results using such are limited (Chang, 2017), there is a need for future studies to depend less on student samples and extend the scope of internet user categories employed in ODA studies (Ying et al., 2009). Drawing from the general advertising literature, it has been observed that real-life practices are constantly metamorphosing into much complicated states than existing methods can test. This makes the homogeneousness in method and prevalence of cross-sectional studies worrying (Chang, 2017). Following the recommendation of Ha (2008) over a decade ago, 112 University of Ghana http://ugspace.ug.edu.gh attempts have been made by researchers to employ tracking and longitudinal studies. However, these studies are still extremely minimal relative to the standard approach of addressing online advertising effects from a cross-sectional standpoint. The online advertising landscape keeps evolving, and the dearth of longitudinal studies suggests that scant focus has been given to the dynamic changes in the expectations of consumers and advertisers. For instance, as Ha (2008) stated, studying the changes in the proportion of unsolicited and solicited ad exposure among internet users over time could help advertisers in creating suitable advertising strategies. As a means of reverberating the author’s decade- long call, future studies may employ more longitudinal studies to provide a trajectory of changes in perceptions, attitude and behaviours of internet users and practitioners toward online advertising among other developments. 4.10 CHAPTER SUMMARY This chapter conducted a review of online advertising literature with a focus on ODA research over the past decade by probing the issues and evidences, the theoretical and methodological approaches, and avenues for future research to be explored. Through a synthesis of relevant studies, the review demonstrates that online display advertising has progressed over the past ten years as several empirical papers have advanced our knowledge and understanding of the field. Much contributions have examined changes in online display advertising formats, several design elements or executional features as well as consumer- related factors pertaining to their efficacy. Generally, however, subjects such as examination of some new and emerging ad formats; explicit and collaborative focus on relevant design elements or executional features; practitioner perspectives on online display advertising; the dearth of theory-driven studies in online advertising research, the lack of qualitative approaches to enquiry, and the prevalent use of student samples are areas requiring future research attention. 113 University of Ghana http://ugspace.ug.edu.gh Purposely, the interrelationships among some pertinent ODA features (exposure conditions, personalisation, interactivity, informativeness, and placement), attitudes toward online advertising, user mode or goal orientation, as well as the nature of the advertised brand (product or service), and the effect they may have on consumer behavioural responses have been given scant research attention and is the area in which this thesis seeks to contribute. The thesis also aims to address these influences from the viewpoint of the SOR model and reversal theory which complement each other in explaining how these ad-related and consumer-related factors may influence behavioural responses. Particularly considered a budding phenomenon in most developing countries, the ODA literature lacks empirical findings from emerging markets or developing country contexts. This thesis, therefore, explores these interrelationships, and also responds to calls for empirical views from developing countries in order to elucidate extant literature. 114 University of Ghana http://ugspace.ug.edu.gh CHAPTER FIVE CONCEPTUAL FRAMEWORK AND HYPOTHESES DEVELOPMENT 5.1 CHAPTER OVERVIEW The first chapter provided a background to the entire thesis, the second chapter discussed the context of the study and the third chapter discussed the two theories (the SOR model and reversal theory) underpinning the study. Chapter four reviewed extant research on online display advertising to show how far academic enquiries on the subject matter have progressed as well as point out gaps in existing research that require future scholarly attention; some of which provided the motivation for this thesis. Specifically, the review revealed that though research focus on features that characterise ODA are increasing, the literature depicts a paucity of explicit focus on some pertinent characteristics, and their influence on behavioural responses. Springing on the three preceding chapters, this chapter develops a research framework to guide the current study. The framework provides a graphical view of this study highlighting and integrating the ODA characteristic (interactivity, placement, informativeness, personalisation, and exposure conditions), consumer variables - attitude toward online advertising, and user mode as well as behavioural response (ad avoidance and ad acceptance) and their interrelationships as they pertain to this study. The chapter is organised in four sections with the overview as the opening section. The second section introduces, presents the graphical view of the conceptual framework, and outlines the assumptions underlying the framework, the third section formulates and presents the hypotheses that guide the empirical investigation, and the fourth section summarises discussions in the chapter. 115 University of Ghana http://ugspace.ug.edu.gh 5.2 INTRODUCTION AND ASSUMPTIONS UNDERLYING THE FRAMEWORK A conceptual framework is considered an essential constituent of any research, and an analytical device that guides a particular study, specifying the central ideas, concepts and variables drawn from various fields of enquiry relevant to the study (Saunders et al., 2009). The framework outlines the constructs to be examined, and the proposed interrelationships among them to be verified and validated. It serves the purpose of aiding a researcher to develop understanding of the phenomenon under investigation by organising the examination procedure of the study data, and also furthering the presentation of results from the research (Grant & Osanloo, 2014). As such, a research framework that is well-illustrated helps the researcher make meaningful deduction from subsequent findings (Smyth, 2004). Guided by the research objectives, theoretical background, and the ODA literature reviewed, a conceptual framework is proposed to guide the empirical part of this thesis. According to Crossan et al. (1999), a framework can only appropriately guide a study if it identifies the subject of enquiry; clearly describes the interrelationships among the components that make up the framework; and clearly states the main assumptions behind the framework. As such, this study examines consumer behavioural responses to online display advertising with a focus on how ad-related and consumer-related variables influence these responses. The framework is graphically portrayed by Figure 5.1 which captures the various factors and their interconnections pertinent to understanding consumer behavioural responses to ODA in Ghana. Using the SOR model as a theoretical lens, the framework assumes that online display advertising (ODA) is characterised by certain stimuli/features conceptualised in this thesis to include interactivity, placement, and informativeness, personalisation, and exposure 116 University of Ghana http://ugspace.ug.edu.gh condition. These ad characteristics are postulated to have a direct influence on consumer behavioural responses (ad avoidance and ad acceptance) to display ads. The framework also suggests that these influences are enhanced/facilitated (mediated) by consumer attitude toward online advertising. Underpinned by the reversal theory, the framework further purports that the effects of the ODA characteristics on attitude toward online advertising, and behavioural responses are moderated by internet user mode. To that end, the ODA characteristics are modelled as the independent variables (stimulus) while behavioural responses are modelled as the dependent or outcome variables. Attitude toward online advertising (ATOA) and user mode are modelled as mediating and moderating variables respectively. The subsequent sections discuss the variables and their interrelationships with corresponding hypotheses formulated to guide the remainder of the thesis. F igure 5.1 Conceptual Framework Stimulus (S) Response (R) Organism (O) ODA Characteristics Interactivity Placement H H Behavioural Responses 12a - 12e Informativeness H1- Attitude toward H11a- Ad Acceptance H Online Advertising Ad Avoidance Personalisation H6a-b - H10a-b Exposure Condition H14a-e H13a-H13e Control Variables Age, Internet usage User Mode rate, Familiarity with online advertising, Gender 117 University of Ghana http://ugspace.ug.edu.gh 5.3 HYPOTHESES DERIVATION “This section examines the proposed interrelationships among the variables in the research framework and formulates 14 major testable hypotheses. The first set of hypotheses, H1-H5 test the direct relationships between the online display advertising characteristics (interactivity, placement, informativeness, personalisation and exposure condition) and attitude toward online advertising (ATOA), while hypotheses H6 a-b -H10 a-b assess the direct relationship between the ODA characteristics and behavioural responses (ad acceptance and ad avoidance). Hypothesis H11a-b tests the direct relationship between attitude toward online advertising and behavioural responses. Also, hypothesis H12a-H12e tests the mediating role of attitude toward online advertising in the relationship between the ODA characteristics and behavioural responses, and hypothesis H13a- H13e and H14a-e test the moderating role of user mode in the relationship between the ODA characteristics and ATOA as well as behavioural responses. Additionally, age, gender, internet usage rate, and user familiarity with online advertising were used as control variables in the study framework. Figure 5.1 visualises these hypothesised relationships, but before these relationships are discussed and their corresponding propositions stated, attention is first turned to how behavioural responses was operationalised in this study in order to put the hypotheses derivation in the right perspective.” 5.3.1 Conceptualising Behavioural Responses Behavioural responses are exhibited actions that follow changes in cognitive and affective states elicited by advertising stimuli (Brajnik & Gabrielli, 2010). From the perspective of Rodgers & Thorson’s (2000) Interactive Advertising Model, they are outcomes of advertising which may include activities and decisions such as ignoring/forgetting the ad, attending to the ad, clicking on the ad, e-mailing the advertiser, searching about the product, purchasing the product, and so forth. While these behavioural responses may take several 118 University of Ghana http://ugspace.ug.edu.gh forms, according to the tenets of the SOR model, the primal behavioural response of consumers or internet users following exposure to stimuli, would be toward (approach) or away from (avoidance) the stimulus environment (the ad) from which other forms of responses may derive (Eroglu et al., 2003). Approach behaviour suggests that the individual moves towards and remains “within” the stimulus environment, whereas avoidance behaviour specifies that the individual moves away or “escapes” from the stimulus environment (Chang et al., 2011). And so, consumers will first have to approach or avoid an ad, before further behavioural actions are taken. What is more, in consumer behaviour research, approach and avoidance are commonly used to depict two broad directions of behavioural responses (e.g. Clark et al., 2009). However, it has been observed by previous research (e.g. Pagani et al., 2011) that examining behavioural responses to stimuli by focusing on only the behavioural direction, does not provide adequate understanding about the behavioural efforts. It is reasoned that during exposure to an advertising stimulus, individuals may not always devote an equal degree of efforts in responding to the ad; they may engage in approach or avoidance behaviours with differing levels of intensity (actively or passively). As such, assessing both the direction (approach or avoidance), and intensity (active or passive) is essential to understanding online behaviours of consumers at a more filtered level (Tang et al., 2014). In light of the above, this thesis considers both the direction (ad acceptance and ad avoidance) and intensity (active and passive) of behavioural responses. The study conceptualises ‘ad acceptance’ as actions taken by consumers to remain with, attend to or engage with an ad. In this vein, active ad acceptance describes effortful actions in this direction and may include, clicking on ads or links provided in the ads, bookmarking online ads, while passive ad acceptances may include such minimal behavioural efforts as paying 119 University of Ghana http://ugspace.ug.edu.gh attention to the ad, reading or watching the ad etc. (Tang et al., 2014). Ad avoidance on the other hand, within the parameters of this thesis denotes actions consumers take to escape from or get rid of online ads (Rejón-Guardia & Martínez-López, 2014). In an active form, avoidance behaviours may include clicking away from online ads, leaving webpages displaying online ads, skipping or closing online ads, and using ad blockers on computers, whereas passive ad avoidance behaviours may include ignoring the ad, looking away from the ad or waiting for the ads to go away (Cho & Cheon, 2004; Seyedghorban et al., 2016). By focusing on the two-dimensional view of responses, this thesis contributes to a clearer understanding of how consumers respond to online display advertising, driven by the various ODA characteristics, as well as their attitude toward online advertising, and their user mode (goal-directedness). 5.3.2 Online Display Advertising (ODA) Characteristics as Stimuli The postulations in the research framework suggests that ODA characteristics (interactivity, placement, informativeness, personalisation and exposure condition) function as stimuli that influence consumer perceptions and actions. To the extent that these characteristics serve as mechanisms by which awareness is raised and product quality is communicated, they convey both informative and persuasive intent; thus, may elicit feedbacks in the form of attitude toward online advertising, ad acceptance and ad avoidance. Next, we present the relationship between the various variables from diverse empirical viewpoints toward the derivation of relevant hypotheses. 5.3.2.1 Interactivity Interactivity describes the extent to which a person can act on and react to (that is, affect and be affected by) a specific stimulus (Florenthal & Shoham, 2010). In the domain of online advertising, interactivity has been defined as “a characteristic of computer mediated 120 University of Ghana http://ugspace.ug.edu.gh communication in the marketplace that increases with the bidirectionality, timeliness, mutual controllability and responsiveness of communication as perceived by consumers and firms” (Yadav & Varadarajan, 2005; p.585). Interactivity is an essential element of the online environment, and in light of this and other definitions considered a three-dimensional construct comprising active control, two-way communication and synchronicity. In the online advertising setting, interactivity allow consumers to actively partake in the persuasion process by controlling ad messages, through selection of content, timing, and order of presentation according to their preferences (Song & Zinkhan, 2008). With advances in internet-related technologies, emerging display ad formats are imbued with interactive features that transcend hyperlinks, to include those that give consumers the opportunity to engage with the advertisement by selecting items or topics of interest in the ad (Baron et al., 2014). Some interactive ads provide automated responses to consumer actions with the ad thus, reducing the spatial separation between advertiser and consumers (Rosenkrans, 2010). Owing to these dimensions, an online display ad can have varying levels of interactivity. The extant literature suggest that highly interactive ads give consumers a significant amount of control and choice that help shape their attitude toward online advertising (Ching et al., 2013). Previous studies on interactive advertising also show affirmative findings in the sense that consumers develop positive attitude toward such ads because they perceive them as pleasant and enjoyable (Jung et al., 2011). Rosenkrans (2009) in his study also indicates that interactive ads influence behavioural outcomes such as click-throughs and mouse rollovers as they tend to engage consumers in the process. Similarly, Pashkevich et al. (2012) established the importance of the interactive and controllable mechanisms of skippable video ads as a crucial feature influencing consumer attitude and reducing negative advertising effects or outcomes. Through its innovative features, interactivity is argued to encourage users to pay closer attention to ads as well as induce cognitive involvement in 121 University of Ghana http://ugspace.ug.edu.gh processing such ads (Jung et al., 2014). The level of interactivity of online display ads is therefore, expected to affect consumers attitude as well as their behavioural responses. Following this rationale, it is hypothesised that: H1: Interactivity positively and significantly influences attitude toward online advertising H6a: Interactivity positively and significantly influences ad acceptance H6b: Interactivity negatively and significantly influences ad avoidance 5.3.2.2 Placement As stated in the previous chapter, although placement issues are diverse, these study focuses on ad-context fit or congruency and so operationalises placement as such. Congruency in advertising research, refers to “the significance of the degree of similarity between the program content and the advertisement content” (Furnham et al., 2002, p.526). “Alternately stated, congruency depicts the extent to which advertising material or content is thematically alike to the editorial content of the media channel or platform (Zanjani et al., 2011). Ad- context congruency is considered a vital factor that affects consumers’ attention to and elaboration of advertising messages and ultimately advertising effectiveness (Kononova & Yuan, 2015). As such, it is seen an essential characteristic in online advertising execution and has received increasing research focus in the past decade in the online display advertising context. Nonetheless, extant research findings provide conflictive results regarding the effects of ad- context congruency explained by two mechanisms (Zanjani et al., 2011). For instance, some empirical evidence suggests that in a congruent ad–context, viewers become more vulnerable to the advertising message which causes intensive information processing and stimulates them to evaluate both the message and the advertisement more favourably (Segev 122 University of Ghana http://ugspace.ug.edu.gh et al., 2014; Belanche et al., 2017). These findings have been explained within the framework of contextual priming. Extant studies associated with this viewpoint argue that, when an ad is placed in an incongruent context (e.g. a mobile phone ad on a website about food), viewers find the contradiction, difficult to process or resolve resulting in less favourable evaluations, attitude and reactions toward the ad (Kononova &Yuan, 2015). On the other hand, the cognitive interference perspective contends that ad placement in an incongruent environment enhances advertising effectiveness. Studies aligning with this view assert that, on congruent webpages, viewers find it challenging to differentiate the advertisement’s stimuli from those of the context (Janssens et al., 2012). This is said to result in a merging of the components of the ad and the context which confuses viewers, and generates negative attitudes toward the ad, advertised brand, and hinders recall (Duff & Faber, 2011). Other studies (e.g. Teng et al., 2014) also contend that because webpage contents are unchanging (i.e. lacking in variety), an incongruent ad may be seen as an exciting, refreshing cue that encourages a more concentrated processing of the advertising message, and may generate favourable attitudes and responses. These contradictory viewpoints have spurred the need to examine the effect of ad-context congruency in this study. Yet, the current thesis finds compelling the contextual priming perspective, according to which context guides the choice of attributes used in processing an ad and the advertised offer by inducing top-of-mind awareness of these attributes, which then generates favourable ad evaluations and responses. Following this line of argument, the next hypotheses state:” H2: Congruent placement positively and significantly influences attitude toward online advertising H7a: Congruent placement positively and significantly influences ad acceptance H7b: Congruent placement negatively and significantly influences ad avoidance 123 University of Ghana http://ugspace.ug.edu.gh 5.3.2.3 Informativeness Informativeness is one of the content-driven features of advertising and describes the extent to which advertising provides valuable information to consumers. According to Wang and Sun (2010a) informativeness is the apprising role of advertising, which aids consumers in making better product and service decisions. It has long been argued that the ability of advertising to provide functional information, and present accurate depiction of products is what stimulates consumer’s perception of advertising value and is the foremost reason for their approval of it (Rotzoll et al., 1990). Today, these pre-millennial assertions even ring truer and may be brought to bear in this discussion because in a fast-evolving technological era, and an information seeking society, the internet offers informational gratification to consumers, and similar expectations are held of advertisements presented on the internet (Mahmoud, 2014). That is to say, consumers require beneficial information that is easily accessible and so, they expect online ads to be informative and useful to help them recognise product or brand differences, and make choices more effortlessly (Kim et al., 2010). Prior studies have identified informativeness as a determinant of consumer attitude toward online advertising as well as several behavioural responses (e.g. Wang & Sun, 2010a; 2010c; Taylor et al., 2011; Yang et al., 2017). Zha et al. (2015) for instance, assert that informative ads attract consumers’ attention and get them to interact with such online ads. This assertion finds support in other studies (e.g. Li-Ming et al., 2013, Mahmoud, 2014) who found informativeness as an online advertising feature that moves consumers to attend to ads. Goodrich et al. (2015) also found that among other features, the inclusion of useful information in an ad, reduces perceptions of intrusiveness, which is a studied cause of negative advertising outcomes such as ad abandonment and avoidance. Essentially, consumers perceived more favourably, online ads that provide beneficial information, and in view of this, the study advances the argument that informativeness or consumers’ 124 University of Ghana http://ugspace.ug.edu.gh perception of display ads as being informative plays a critical role in their attitude toward online advertising as well as their behavioural responses as encapsulated in the following hypotheses: H3: Informativeness positively and significantly influences attitude toward online advertising H8a: Informativeness positively and significantly influences ad acceptance H8b: Informativeness negatively and significantly influences ad avoidance 5.3.2.4 Personalisation Personalisation signifies the extent to which advertising messages are tailored to reflect the preferences, lifestyle and specific cultural and geographical characteristics of individual consumers (Leppäniemi & Karjaluoto, 2008). It is a strategy in persuasive communication that involves integrating elements in a message that refer to individual recipients on the basis of their personal characteristics (e.g. name, gender, residence or location, occupation), previous behaviours and so on (Maslowska et al., 2016; Liu & Matilla, 2017). According to Lambrecht and Tucker (2013), personalisation starts with a particular consumer, and efforts are made to design individualised advertisements that best match the individual’s preferences and personal interests in order to heighten the benefits the consumer may gain. Montgomery and smith (2009) opine that personalisation requires very little effort from the consumer because they depend on the marketer to recognise and fulfil their needs. By this, the strength of a personalised advertisement lies in the consumer’s perception of it as being personally useful (Aguirre et al., 2015). Past studies have reported that personalised ad messages allow marketers to reach their prospective consumers in a bespoke manner causing these consumers to be more responsive toward such advertisements (Bright & Daugherty, 2012). Recent evidence also suggests that 125 University of Ghana http://ugspace.ug.edu.gh the degree of personalisation plays a crucial part in consumers’ perceptions of online advertising by heightening personal relevance, lessening ad skepticism, and encouraging more focused processing of ads (Maslowska et al., 2016; Sahni et al., 2018). An earlier study by Campbell and Wright (2008) also showed that participants’ favourable attitude toward an ad increased when they were exposed to personally relevant messages, and they considered non-personally relevant messages as disruptive. Although, personalised advertising is growing in popularity, its downsides have been confirmed by some studies (Baek & Morimoto 2012; Tucker, 2014; Bleier & Eisenbeiss, 2015a). For instance, van Doorn & Hoekstra (2013) in their study found that personalisation increases purchase intention but heightens perceptions of intrusiveness which in turn negatively affects purchase intentions, thus calling it a “double-edged sword”. Similarly, Aguirre et al. (2015) found that click-through rates plummeted when consumers became aware that their personal data were traced and analysed without their permission. While both positive and negative effects of personalisation have been reported, this study puts forth the argument that to the extent that the personalised ads are relevant to consumers they are more likely to interact with the ad and form positive perceptions about it. In this light, the study hypothesises the following: H4: Personalisation positively and significantly influences attitude toward online advertising H9a: Personalisation positively and significantly influences ad acceptance H9b: Personalisation negatively and significantly influences ad avoidance 5.3.2.5 Exposure Conditions Exposure conditions as discussed in the previous chapter, border on issues of forced exposure, exposure repetition, as well as exposure duration. Exposure conditions in this 126 University of Ghana http://ugspace.ug.edu.gh study is therefore operationalised to comprise variables as how long an ad is, the frequency with which it is repeated, and the degree to which consumers are forced to view the ad. Forced exposure describes the degree to which internet users are coerced to view an ad if they wish to visit a webpage or continue in their online task (Edwards et al., 2002). The nature of the exposure is one criterion that distinguishes ODA formats because while some formats allow viewers to close an ad window, others are imposed and do not provide this opportunity (Li & Meeds, 2007). A pop-up ad for instance, is illustrative of forced exposure, since the ad is run exclusive of any user action and disappears mechanically after a given time passes. Given the ‘goal-orientedness’ of the internet, any form of interference causes internet users to respond cognitively, affectively, and behaviourally, however, the nature of the interference will determine the direction (approach or avoidance) of the response (Campbell et al., 2017). Researchers have noted that a forced exposure condition produces higher perceptions of ad intrusiveness, and more negative attitude toward the ad compared to unforced exposures which then affects the behavioural responses (Hegner et al., 2015). Instances of forced exposure have been pointed out by other studies to lead to ad avoidance (Baek & Morimoto, 2012). Concerning exposure frequency, studies such as Nottorf (2014) for instance, found that repeated exposure to display ads reduces the likelihood of ad clicking by consumers. According to McCoy et al. (2017), repeated exposure induces feelings of irritation resulting in unfavourable perceptions about online display ads. This provides support for Rejòn- Guardia and Martínez-Lopéz (2014) who submitted that the connection between advertising repetition and attitudes is driven by an internet user’s perceptions of intrusiveness and irritation. Yaverogly and Donthu (2008) also opined that when internet users are online to perform certain tasks, frequent exposure to the same ad produces lags in their internet usage experiences and causes them to react negatively toward such ads. 127 University of Ghana http://ugspace.ug.edu.gh Online advertising researchers have also discussed the effects of exposure duration. In this regard, some researchers (e.g. Goodrich et al., 2015; Li & Lo, 2015) reported from their study, that lengthy ads have more positive effects; as they generated more ad recall, recognition and reduced perceptions of intrusiveness while Pashkevich et al. (2012) in an earlier study in the context of skippable videos, confirm that a longer video ad reduces ad acceptance. From the perspective of Rejón-Guardia and Martínez-López (2014), ad messages that stay on the screen for long hold internet users’ “captives” causing them to abandon their activities or employ avoidance behaviours like leaving the webpage. Providing further clarity on exposure duration, Wang et al. (2013) established that longer ads may not in all cases lead to positive influences and reactions. The authors explained that the effect of ad length is conditioned by the ease with which consumers can process the ad. Since in examining the effects of advertising exposure on consumer perception, attitudes and behaviours, most studies consider the effect of repeated exposure to an ad (e.g. Lee & Cho, 2010); repeated exposure and forced exposure (e.g. Kim, 2018), as well as repeated exposure and exposure duration (e.g. Wang et al., 2013) rather than the three perspectives, this thesis examines online advertising exposure effects from these perspectives. This study contends that when internet users are exposed to forced formats of ODA, they are restricted from indulging in their purpose for being online. For this reason, when forced ads are perceived as lengthy, and are shown multiple times during an online session or website visit, the perceptions of interruptions are pronounced causing unfavourable attitudes and reactions toward the ad and the opposite may hold true. The study, therefore, proposes the following testable hypotheses in examining the influence of exposure conditions on attitude toward online advertising and behavioural responses: H5: Exposure conditions positively and significantly influence attitude toward online 128 University of Ghana http://ugspace.ug.edu.gh advertising H10a: Exposure conditions positively and significantly influence ad acceptance H10b: Exposure conditions negatively and significantly influence ad avoidance 5.3.3 Attitude toward Online Advertising (ATOA) Per the SOR model, consumers react to stimuli in two stages, the first are internalised responses indicated as organism response, which drive the external response (Chen & Yao, 2018). “Organism” is the consumers affective state which show the feelings and emotions of the consumer, following exposure as well as cognitive states which involve all that goes on in the consumer’s mind regarding acquiring, processing, retaining and retrieving information (Eroglu et al., 2001; Kamboj et al., 2018). According to Jacoby (2002), beliefs and attitudes are part of the organism component. In this thesis attitude is considered as the organism element of the study, predicated on early scholarly submissions (Mitchell & Olson, 1981; Petty et al., 1983) that attitude formation is an internal process that ultimately directs behaviour. Within the borders of attitudinal models, attitude generally describes the extent to which an individual has a favourable or unfavourable evaluation of an object of interest following the belief (perceived probability) that the object of interest possesses particular attributes (Palmgreen, 1984; Ajzen, 2001). The concept of attitude toward advertising was first conceptualised by Lutz (1985) as “a learned predisposition to respond in a consistently favorable or unfavorable manner to advertising in general” (p.53). Attitude toward online advertising is thus, a learned predisposition that individuals develop as they perceive the benefits and drawbacks that online advertising offers them and/or to others (Fransen et al., 2015). Essentially, attitude formation is preceded or conditioned by several forms of ad 129 University of Ghana http://ugspace.ug.edu.gh evaluation through self-perceptions based on direct exposure to online ads (Souiden et al., 2017; Wang et al., 2010b). Attitude is regarded as one of the major determinants of advertising efficiency, and the relationship between attitude and behaviour has been an area of key focus in advertising research (Mehta, 2000; Sung & Cho, 2012). Although research on attitude toward online advertising is relatively nascent and sparse, extant empirical evidence point to its role in affecting consumers’ subsequent behaviour toward online ads and even the advertised brand (Shaouf et al., 2016; Souiden et al., 2017). For instance, Wang and Sun (2010a; 2010c) found that attitude toward online advertising was a strong driver of ad clicking and shopping behaviours among consumers. Moreover, attitude toward online advertising encapsulates how essential consumers consider online advertising to be; whether they like online advertising, consider it fun to see as well as hold favourable opinions about online advertising (Woolin et al., 2002; Wang &Sun, 2010b), all of which have been found to have direct effect on consumer’ behaviour and reactions toward online ads (Sun & Wang, 2010; Zha et al., 2015). In spite of this, Wang et al. (2009) state that the associative relationship between consumers attitude toward online advertising and their behavioural responses is not fully grounded in the literature. In view of this, the present study, examines the formation of attitude toward online advertising as well as its effect on consumers’ behavioural response toward display advertising. Conceptually, we link attitude toward online advertising (as an internal response) to ad acceptance and ad avoidance (as consumers’ final external behavioural response). Hence the study hypothesises that: H11: Attitude toward online advertising directly influences behavioural responses such that: H11a: Favourable attitudes toward online advertising positively and significantly influence ad acceptance 130 University of Ghana http://ugspace.ug.edu.gh H11b: Favourable attitudes toward online advertising negatively and significantly influence ad avoidance 5.3.4 The Mediating Role of Attitude toward Online Advertising (ATOA) The SOR model draws the links between three interconnected levels of variables by stating that the relationship between stimuli and responsive behaviour is facilitated by the internal responses of the organism exposed to the stimuli (Islam & Rahman, 2017). Following from earlier discussions, attitude toward online advertising is considered a mechanism that may explain the effects of ODA characteristics on behavioural responses. The attitude of consumers toward online advertising basically stems from perceptions they hold about online advertising (Réjon-Guardia & Martinez-Lòpez, 2014), and previous research has provided footing for the assertion that attitudes toward online advertising affect consumers’ responses toward online advertising (Saadeghvaziri et al., 2013). A more positive attitude toward online advertising has been associated with favourable evaluations of specific ads as being informative, interactive, and entertaining among other characteristics (Wang & Sun, 2010a; Ching et al., 2013), and is said to lead to higher recalls and more positive behavioural responses toward ads and even purchase intentions (Goodrich, 2011; Goodrich et al., 2015). From this perspective, past studies have demonstrated that the influence of online advertising attributes as perceived by consumers on behavioural responses is theoretically and emperically mediated by attitudes toward online advertising (Wang et al., 2009). For instance, Wang and Sun (2010a) found that attitude toward online advertising mediates the relationship between consumers belief about online advertising in general, and their online behaviours of ad clicking and shopping frequency in China and USA. This study was replicated across several cultural settings (Wang & Sun, 2010b; 2010c; 2010d) and found attitude toward online advertising to be a mediator of the relationship between consumers’ 131 University of Ghana http://ugspace.ug.edu.gh online advertising beliefs (entertainment, informative, credibility, economy, value corruption) and behaviours. However, the mediating effect of attitude toward online advertising in the context of online display advertising is yet to be fully investigated. Based on the assumptions of the SOR model, we expect attitude toward online advertising to serve as an intervening response, transmitting or facilitating the effects of online display advertising characteristics on behavioural responses of consumers. Accordingly, the study puts forth the following hypotheses:” H12: The effects of ODA characteristics on consumer behavioural responses are mediated by attitude toward online advertising (ATOA) such that; H12a: ATOA mediates the relationship between interactivity and behavioural responses H12b: ATOA mediates the relationship between placement and behavioural responses H12c: ATOA mediates the relationship between informativeness and behavioural responses H12d: ATOA mediates the relationship between personalisation and behavioural responses H12e: ATOA mediates the relationship between exposure conditions and behavioural responses 5.3.5 The Moderating Role of Internet User Mode In the view of Aguinis et al. (2016) moderation represents the notion that the degree of the effect of predictors on a response variable depends on contingency factors, and as such describes the conditions under which such effects may vary in size. This current study along this line of reasoning, argues that internet user mode moderates the relationship between ODA characteristics and ATOA as well as behavioural responses. In the context of media usage, people have various motivation for consuming media content (Bleier & Eisenbess, 2015a). Particularly, in comparison with conventional media, the Internet is considered a more goal and task-oriented medium (Cho & Cheon 2004, Wang et al., 2013). This is because consumers go online to undertake certain tasks to achieve their goals, and according to the internet motivation inventory classification framework, usage motives come in four categories – researching, shopping, communication, and surfing (Rodgers et al., 2007). Per 132 University of Ghana http://ugspace.ug.edu.gh Jung et al. (2014), an internet usage motive “is an inner drive to carry out any online activity” (p.1309). Consistent with the reversal theory, the four internet usage motives are further classified into telic and paratelic modes (Rodgers et al., 2007), and it is presumed that internet users who are most often researching and shopping online, are more goal- directed (telic) while users that are most often surfing and communicating are less goal- directed (paratelic) (Apter, 1984). Extant literature suggests that goal-directed/telic users give more select attention to internet- to-goal-related information (Stanaland & Tan, 2010) and avoid disrupting information on the site (Duff & Faber, 2011) whereas experiential/paratelic users engage in screening activities, and easily shift attention from focal to unrelated materials (Janssens et al., 2012). Past studies have demonstrated that telic and paratelic users, or consumers in telic and paratelic modes respond differently to online advertisements (Jung et al., 2014; Seyedghorban et al., 2016). According to Jung et al. (2014), Internet users in a serious- minded (telic) state form highly positive attitudes toward online ads with low levels of interactivity whereas those in a playful-minded (paratelic) state develop positive attitudes toward ads with low interactivity levels. Also, Seyedghorban et al. (2016) in their study showed that the effects of predictors such as perceived goal impediment, and prior negative experience on advertising avoidance differed among telic and paratelic users. Further Bleier and Eisenbess (2015a) in an experimental study, found variations in consumers’ perception of ad informativeness and intrusiveness and its effect on click-through on the basis of goal- directed browsing. For Simola et al. (2011), the goal/user mode of an internet user can exert significant influence on attention allocation such that ads attract more user attention when users are casually browsing the internet, than when they are involved in for instance, a reading task. Essentially, although attitude toward as well as behavioural responses to display ads may be influenced by the characteristics of these ads, the direction and intensity 133 University of Ghana http://ugspace.ug.edu.gh of these responses may vary across user modes. In other words, according to the reversal theory, different user modes may generate different forms and levels of ATOA and behavioural responses. Considering these arguments, it seems reasonable to hypothesise that: H13: User mode moderates the relationship between ODA characteristics and behavioural responses such that; H13a: User mode moderates the relationship between interactivity and behavioural responses H13b: User mode moderates the relationship between placement and behavioural responses H13c: User mode moderates the relationship between informativeness and behavioural responses H13d: User mode moderates the relationship between personalisation and behavioural responses H13e: User mode moderates the relationship between exposure conditions and behavioural responses H14: User mode moderates the relationship between ODA characteristics and attitude toward online advertising (ATOA) such that; H14a: User mode moderates the relationship between interactivity and ATOA H14b: User mode moderates the relationship between placement and ATOA H14c: User mode moderates the relationship between informativeness and ATOA H14d: User mode moderates the relationship between personalisation and ATOA H14e: User mode moderates the relationship between exposure conditions and ATOA “In mapping the hypotheses to the research objectives, it should be noted that hypotheses 6 to 10 which propose a positive and significant influence of the ODA characteristic on consumer behavioural response, are stated to help achieve the first objective of the study. The second objective of the study is addressed by seven hypotheses; hypotheses 1 to 5 which propose a positive and significant influence of the ODA characteristics on attitude toward online advertising; hypothesis 11, which postulates a positive and significant relationship between attitude toward online advertising and behavioural responses, and hypothesis 12 which suggests the mediating influence of attitude toward online advertising in the relationship between ODA characteristics and behavioural responses. Lastly hypothesis 13 134 University of Ghana http://ugspace.ug.edu.gh and 14 propose a moderating influence of user mode on the relationship between the ODA characteristics and behavioural responses as well as attitude toward online advertising and address the third objective of the study.” 5.3.6 Control Variables “Scholars recommend that in order to eliminate possible non-hypothesised effects in a given study, researchers should determine the effect of control variables (van Reijmersdal et al., 2016). In a particular study, variables that may have a potential influence on the outcome/dependent variable but are not observed variables of interest are considered control variables. In most fields of enquiry, demographic variables serve as control variables, and evidence exists in the online advertising literature that consumers with diverse demographic characteristics respond differently to advertisements (Ketelaar et al., 2015; Shaouf et al., 2016). Concerning gender for instance, Goodrich (2014) identified in their study that males form more favourable attitudes toward online advertising with minimal or no advertising attention compared to women. Their study also showed gender differences in attitude formation based on ad placement in that, more favourable attitudes were generated for ads on the left of a webpage for males, and on the right for females. Regarding age, Goodrich (2013) found that older adults pay higher attention to online banner ads than younger adults, and mere exposure effects differ for the two age groups. We, therefore, control for the effect of demographic factors such as gender and age. Furthermore, internet users spend varying amounts of time online and show varying degrees of understanding with online advertising. In view of this, the framework as well integrates usage rate and familiarity with online advertising as control variables since these have been shown to influence consumer 135 University of Ghana http://ugspace.ug.edu.gh judgements, perceptions and responses to online advertising (Sun & Wang, 2010; Hanafizadeh et al., 2012). 5.4 CHAPTER SUMMARY This chapter presented the research framework that guides the empirical part of the thesis. The two theories (SOR model and reversal theory) discussed in chapter two provide an appropriate theoretical underpinning for the conceptual framework which integrates the pertinent ODA characteristics, and consumer-related variables that emerged from the review of ODA literature. The chapter discussed the interrelationships among the various ODA characteristics, attitude toward online advertising, user mode, and behavioural responses, from which 14 major testable hypotheses were derived (as captured in Figure 5.1) to guide data collection and analysis. It is projected that the testing of hypotheses postulated in the research framework will produce insightful findings that may guide online display advertising designs and activities of practitioners (e.g. advertisers, publishers) in settings with similar dynamics to those in the study setting. This chapter concludes the first part of the thesis which provided the theoretical directions for addressing the research problem, and as such sets the tone for the empirical part which begins with the next chapter on the research methodology. 136 University of Ghana http://ugspace.ug.edu.gh CHAPTER SIX RESEARCH METHODOLOGY 6.1 CHAPTER OVERVIEW As stated in the preceding chapters, this thesis uses the stimulus organism response model and the reversal theory as foundational theoretical perspectives in developing a conceptual framework geared toward understanding the interrelationships among ODA characteristics, consumer attitude toward online advertising, internet user mode and behavioural responses. To begin the empirical part of the thesis, this chapter outlines the methodology used for the research. The chapter provides descriptions on the methods employed in the study as well as discussions and justifications for the approaches used to achieve the objectives and research questions delineated in chapter one. Besides the overview section, this chapter consists of nine remaining sections (6.2-6.10); section two introduces the chapter, and section three discusses research paradigms, and the choice of positivism as the paradigmatic stance for this thesis. Section four outlines the research purpose; and section five and six present the research approach and strategy respectively. Section seven discusses the data collection methods, followed by discussions on the mode of data analysis (section eight). The penultimate section of the chapter discusses reliability and validity issues regarding the research instrument and the final section summarises the entire chapter. 6.2 INTRODUCTION Methodologies are considered systems of explicit rules on which research is grounded and claims regarding knowledge are evaluated (Frankfort-Nachmais & Nacmais, 1996). In view of this, any research work should be guided by a clear research methodology founded on scientific principles. Some scholars (e.g. Ying et al., 2009; Saunders et al., 2011) have pointed out that a researcher’s choice of methodology or research design is determined by 137 University of Ghana http://ugspace.ug.edu.gh the purpose and objectives of a given study. This suggests that a specific methodology may not be right or wrong but can be more or less suitable given the focus of the study for which it is employed (Silverman, 2001). From the perspective of Eldabi et al. (2000), methodological issues to be considered in any research work should include, paradigms/philosophical perspectives, research purpose, research approach, research strategy as well as data collection methods, and modes of data analysis. Following this suggestion, the structure of the methodology for this research is pictured in Figure 6.1, and the various issues pointed out by Eldabi et al. (2000) are subsequently discussed as they pertain to this thesis. Figure 6.1 Structure of the Methodology 1.Research Paradigm 2. Research Purpose 3.Research Approach 4.Research Strategy 5. Data Collection Methods 6. Data Analysis Methods 7. Reliability and Validity 6.3 RESEARCH PARADIGM AND PHILOSOPHICAL VIEWPOINTS “Generally, all academic research works are grounded on a selected paradigm or philosophical perspectives (Blaikie, 2010). According to Myers (2013), paradigms constitute the fundamental philosophical assumptions which describe what a ‘valid’ research is, and the suitable methods that the research may apply. An early definition of research paradigm is provided by Kuhn (1970) who refers to it as a set of beliefs, values and 138 University of Ghana http://ugspace.ug.edu.gh techniques that guide and dictate the sorts of problems that members within a discipline should address and the kinds of explanations that are acceptable to them. Largely, no specific paradigm can provide adequate grounds for understanding issues in any field of study; as such, academic fields can only be advanced if the diversities in the philosophical assumptions are allowed to aid complementary scholarly enquiries within disciplines (Bryman & Bell, 2015). Several paradigms exist and have clear distinctions among them based on their ontological, epistemological as well as methodological assumptions, and these assumptions act as a guiding structure which explicates and separates them from one another (Creswell, 2014). Husey and Husey (1997) therefore, highlight the need for researchers to recognize and understand their philosophical orientations in the context of the paradigm adopted for a specific research work. Ontological assumptions concern the reality studied by researchers and explains a researcher’s philosophical belief system about the nature of social reality (Healy & Perry, 2000). Two questions in the literature that confirm differences in research orientations explained in ontological assumptions border on whether ontology describes a phenomenon that is really ongoing (objective reality) or what the researcher beliefs is ongoing (subjective reality) (Hatch & Cunliffe 2006). Objectivism represents the position that social entities exist in reality, external to and independent of the social actor while subjectivism holds that social phenomena are created from the social actor’s perceptions and resultant actions (Bryman 2001). In academic spheres, particularly in the context of social sciences, scholars admonish that stating one’s ontological stance as a researcher is critical as this forms the foundation on which the researcher view reality (Crotty 2003). Epistemological assumptions look at how we come to know the world around us and so, focus on the relationship between the researcher and the reality and/or how this reality is 139 University of Ghana http://ugspace.ug.edu.gh known or captured (Neuman, 2014). Epistemology considers opinions about the most suitable ways of enquiring into the nature of reality (Easterby-Smith et al., 2012) and so, looks at what counts as accepted truth by specifying the criteria for deciding when knowledge is both adequate and legitimate (Blakie, 2010). Methodology involves the strategy; action plan and the diverse processes designed to inform the choice and use of specific research methods for a desired outcome (Crotty 2003). According to Wahyuni (2012), methodology refers to the outline used to conduct research, within the framework of a certain paradigm. Ponterotto (2005) opines that methodology comprises the procedures and tools employed in learning about reality and derives from a researcher’s ontological and epistemological position. It is worth noting that methodology can be contrasted from a research method which describes the set of specific techniques and tools used to gather and analyse the data specified by the research methodology (Blaikie 2010). In simple terms, ontology looks at the nature of reality, whereas epistemology concerns how we come to know this reality, that is, what counts as knowledge. And methodology pinpoints the techniques we employ in knowing and learning about the reality (Krauss, 2005). In light of these philosophical viewpoints, and as early on hinted, diverse typologies of paradigms have been espoused by scholars. However, the most central paradigms that mirror the major theoretical directions in social science research are positivism, interpretivism/constructivism, realism, critical realism and pragmatism (Blaikie, 2010; Neuman, 2014). For every paradigm, there is a distinct logical relationship between its ontological, epistemological, and methodological assumptions, and Table 6.1 provides a glimpse of these major paradigms and succinctly explains their philosophical perspectives.” After this, the study chooses and justifies positivism as its paradigmatic framework. 140 University of Ghana http://ugspace.ug.edu.gh Table 6.1 Major Philosophical Paradigms in Social Science Research “Paradigm Ontology Epistemology Methodology What is the nature of What is the nature of How is Knowledge reality knowledge generated? created? Positivism Singular, objective and Distance and impartiality. Researchers apply deductive tangible reality-researchers Knowledge generated is reasoning. That is, research reject or fail to reject objective, free of time influences, questions are formulated, and hypothesis. and is independent of context hypotheses derived, and tested (e.g., researchers objectively under controlled collect data on instruments) circumstances. Interpretivism Reality is socially Closeness. Knowledge is Researchers apply inductive constructed; thus, multiple composed via interactions reasoning. Knowledge is realities exist - researchers between the researcher and generated by identifying provide quotes to illustrate participants or objects of various constructions of different perspectives. investigation. Knowledge reality through the views of generated is therefore, subjective, participants to build up time-bound and dependent on patterns, theories, and context. generalizations. Realism Reality is real but only Value-cognizant. Findings are Social phenomenon is Probabilistically and probably true, but researcher understood via hypotheses imperfectly apprehensible; needs to triangulate any which are tested to establish so, to learn about it, perception patterns of associations and triangulation from many collected. hence the most possible sources is needed. explanation. Critical Realism Two worlds - transitive and Transitive world is value laden Researchers apply retroductive intransitive. The former is and changing continually. reasoning; seek to deconstruct what we observe and learn Intransitive world has underlying and understand the structures with our mind – perception structures and mechanisms that & mechanism underlying the of reality. The latter are ‘relatively enduring’ – that is subjective existent realities. represents the reality which what we want to study Triangulation from many is independent of what the sources is required to try to mind thinks. know it. Pragmatism Singular and multiple Practicality - researchers collect Combination of logical realities - researchers test data by what works to address stances - researchers collect hypothesis and provide research question. both quantitative and multiple perspectives. qualitative data and mixing them. ” Source: Creswell and Clark (2010); Neuman (2014) 6.3.1 Positivism as the Paradigmatic Stance of this Study “The central belief of positivism is the view that the social world exists as an external environment where definite structures affect people in similar ways and vice versa (Proctor, 2005) and therefore, its properties should be measured through objective methods, rather than be inferred subjectively through sensational reflection or intuition (Easterby-Smith et al., 2012). Positivism seeks unbiased findings through value-free approach and ensures that 141 University of Ghana http://ugspace.ug.edu.gh the researcher is independent from the respondent (Malhotra & Birks, 2007). A positivist approach entails the need to reduce a research problem into coherent sub-units, operationalise concepts of interest in order to make measurements, select appreciable large samples to increase validity, and develop hypotheses to demonstrate and test their authenticity (Easterby-Smith et al., 2012). This current study, (1) formulates research objectives geared toward answering the overaching research question, (2) provides definitional and measurement parameters within which the key constructs in the study are used, (3) estimated and used a considerably large sample based on scholarly suggestions and precedence from earlier works and, (4) developes 14 testable hypotheses to guide the empirical part of the study. In so doing, this study exhibits features that typify positivism and so, passes as a positivist research. More specifically, by adopting a positivist paradigmatic position, the study assumes an objective ontology. According to Denscombe (2008), positivist ontology is an approach that pursues the natural science model of research to inquiries of social phenomena and explanations of the social world. This philosophical school of thought believes that a researcher is an entirely objective, neutral observer of an existing social reality (Easterby- Smith et al., 2012). This study examines features that characterize online display advertising and assesses the behavioural responses of consumers as a result of their perceptions of these features, attitude toward online advertising and their internet usage motive or goal orientation (user mode). The nature of the research problem, and the research objectives indicate that the phenomenon (online display advertising) being enquired into exists external to the researcher whose views do not influence the subject under study. From an epistemological perspective, positivists attempt to discover the truth about the social world, through observable and measurable facts in order to generalize fundamental 142 University of Ghana http://ugspace.ug.edu.gh laws about universal social realities (Easterby-Smith et al., 2012; Saunders et al., 2009). Generally speaking, studies conducted from a positivist perspective, attempt to test theory, with the intention to surge the predictive understanding of a phenomena (Myers, 2013). This substantiates the primary aim of the current study which is to measure pertinent ODA characteristics, examine their influence on consumers attitude toward online advertising and their behavioural responses, as well as assess the variation that may exist in the direction and intensity of these behavioural responses on the basis of consumer user mode and the nature of the advertised brand. Relevantly, ‘interactivity, personalisation etc. as ODA characteristics are considered ‘real’, can be operationalized, and are objectively measurable without the influence of the researcher’s biases which may provide the most satisfactory scientific evidence (Saunders et al., 2009). Methodologically, Gill and Johnson (2010) argue that positivist research highlights extremely structured methodology so as to enable replication and quantifiable observations culminating in statistical analysis. In addition, according to Saunders et al. (2009), the positivist epistemology that objective facts provide most suitable scientific evidence most probably leads to the choice of quantitative research methods. Consistent with this view and the purpose of this study, a quantitative approach is employed through a cross-sectional survey to gather quantifiable responses, which were subjected to statistical analysis in order to test and confirm or disconfirm the hypotheses formulated from existing literature. This approach is more closed to bias and is thus, more objective reflecting the essence of positivism which also highlights the use of existing theory to test hypotheses (Blaikie 2010; Saunders et al., 2009). Advertising is no new concept in marketing, however, following from the arguments laid out for the essence of an advertising study in the online domain, only justifies the researcher’s attempt to generate deductively new knowledge from existing ones by testing the constructs of empirical data as practiced by positivists. Having explicitly 143 University of Ghana http://ugspace.ug.edu.gh stated the research paradigm for this study, the next section explains the basis on which the study is purposed.” 6.4 RESEARCH PURPOSE/DESIGN A research purpose offers the fundamental direction for conducting the research, and scholars (e.g. Neuman, 2014) suggest that a study may aim to explore, describe, explain, predict and change among other things. According to Marshall and Rossman (2014), dependent on the type of research questions a study seeks to answer, there can be a singular or multiple purpose to a research. From a social science research perspective however, there are three categories of research purpose or design namely; exploratory, descriptive and explanatory (Saunders et al., 2011). Descriptive Research Descriptive research systematically describes a phenomenon, situation or problem and usually asks the ‘what’ question and is often employed when a problem is well-organised. Studies conducted with a descriptive purpose or design, enable profound enquiries into a research phenomenon. According to Babbie (2004), in descriptive studies, the researcher observes and then describes what was observed, and these descriptive reports can be conveyed in words or numbers and may comprise the development of sets of classifications (Blaikie, 2010). Exploratory Research With exploratory research, the main focus is on the discovery of ideas and insights and so, is mostly used when a researcher examines a new interest or when the subject of study itself is relatively new (Blaikie, 2010). Because the phenomenon of interest in such studies is considerably new and unfamiliar to the researcher, more information is needed to clarify the 144 University of Ghana http://ugspace.ug.edu.gh concept and scope of the study, and to put the phenomenon into the right perspective and paint an accurate picture of the research issue (Saunders et al., 2011). A research purposed on an exploratory basis could be conducted through a number of techniques including interviews, focus group, case study and so forth (Cooper & Schindler, 2006) which are used to better understand the research problem. 6.4.1 Explanatory Research as the Chosen Design Explanatory research focuses on studying and understanding specific situations or problems in order to explain the relationships among variables (Saunders et al., 2011). According to Yin (2012), explanatory research is also known as the causal research design as it addresses cause and effect relationships. This type of research aims to develop defined schemas that can be used to clarify a phenomenon, resulting in a generalization from the research (Green, 2008). Studies conducted using explanatory designs allow researchers to ascertain the effect a variable(s) has on other variables within a specified framework. In order to examine the interrelationships among ODA characteristics, attitude toward online advertising, as well as their effect on consumer behavioural responses to ODA, this study is purposed on an explanatory basis with the intent of explicating the nature of relationship among these variables as illustrated in the conceptual framework. It must be noted that, explanatory research is generally quantitative and mostly tests previous hypotheses by measuring the connections between variables (Maxwell & Mittapalli, 2008). In this view, the explanatory design also aids the current study to provide clarifications on the contingency(moderating) effect of Internet user mode on these focal relationships. Following its explanatory design, the ensuing section presents the research approach used to conduct the study. 145 University of Ghana http://ugspace.ug.edu.gh 6.5 RESEARCH APPROACH Research approach is the procedure for research and entails various stages from data gathering through analysis to interpretation (Creswell, 2014). Creswell and Clark (2007) advance three key non-discrete approaches to conducting research – qualitative, quantitative and mixed-methods; all of which are used in social science research. Quantitative and qualitative are the general approaches representing two ends of a spectrum, and a mixed- methods falls in the middle, incorporating elements of the two main approaches (Creswell, 2014). According to Cooper and Schindler (2006), where there is inadequate understanding of the phenomenon under investigation qualitative research is considered more fitting since it provides the researcher more descriptive space and does not restrict a study to rigid formulated processes. Alternatively, quantitative research is mainly concerned with figures and representativeness and has extremely controlled methods for data collection (Hair et al., 2010). The dichotomy between these two approaches lies in the number of research respondents used, and the processes of data collection and analysis (Creswell, 2014). In order to make a cogent case for quantitative approach as the suitable style for this study, the next subsections discuss the three approaches. 6.5.1 Quantitative Research The quantitative research approach involves testing objective theories in order to determine the degree of association among variables (Creswell, 2014). The approach emphasises the assessment, measurement and analysis of causal links between variables, and is aimed at heightening objectivity, replicability and generalisability (Ponterotto, 2005; Creswell & Clark, 2007). In other words, results gotten from quantitative studies are more generalisable, and their findings can be reproduced (Creswell, 2014). Quantitative studies are structured because they may begin with specific hypotheses or research questions developed from existing theories and prior studies and use objective instruments to collect data from a 146 University of Ghana http://ugspace.ug.edu.gh carefully chosen sample (Hair et al., 2010). Studies employing this approach, generate findings primarily via statistical analysis, adopt deductive logical positions, and are designed to isolate and minimize bias, as well as control for alternate explanations (Podsakoff et al., 2012; Shaughnessy et al., 2012). Despite its merits, quantitative research has often been criticized for lacking the capability to provide in-depth insights or understanding into a given phenomenon unlike qualitative studies. 6.5.2 Qualitative Research Qualitative research involves an array of empirical procedures devised to illustrate and explain the experience of research participants in specific contexts (Creswell, 2014). Studies that adopt a qualitative approach to enquiry, investigate the social world generally from the viewpoint of participants involved in the study (Yin, 2012), and focus on unearthing the experiences of these participants using verbal summaries with no or minimal statistical analysis (Shaughnessy et al., 2012). That is to say emphasis is placed on words to create in- depth understanding of the studies relative to what a quantitative approach provides. Because the qualitative approach is designed to help researchers understand people and their environment by gathering data from participant’s in their own setting (Creswell, 2014), it entails extremely close interactions with small purposive samples over longer time periods (Yin, 2012). Qualitative research is mostly inductive, progressing from defragmented facts to a more wholesome view of a situation (Wynn & William, 2012) by employing open- ended questions that gives participants the latitude to articulate their views and offer deep insights into the research problem (Creswell, 2014). Although qualitative research may be an end in itself, it is also often relied upon for the subsequent development of hypotheses and the identification of variables that could be used in quantitative studies (Malhotra, 2010). A major downside to this approach, however, is the lack of replication since 147 University of Ghana http://ugspace.ug.edu.gh qualitative research assumes that data analysis is restricted to the researcher, as well as the conditions and time of analysis (Yin, 2009). 6.5.3 Mixed-Methods Research Mixed-methods refers to the gathering and analysis of data using a combination of both qualitative and quantitative approaches in a single investigation (Creswell & Clark, 2007). The approach emphasises the view that the reliance on a combination of both quantitative and qualitative approaches, enhances the quality of research and provides a comprehensive and detailed understanding of the research issue compared to when a single approach is used (Creswell, 2014). Three types of mixed methods have been pointed out in social science research namely; concurrent, sequential and embedded (Johnson et al., 2007). Concurrent mixed methods (triangulation) uses both qualitative and quantitative approaches simultaneously to provide a complete analysis of the research issue. In embedded mixed methods, one approach supplements the other (Blaikie, 2010). Sequential mixed-methods builds on the results of one approach with another approach and may first start with a qualitative interview in order to explore the issue and is then followed by a quantitative survey to generalize results (i.e. exploratory sequence). The study may as well start with a quantitative method to verify a theory and follow up with a qualitative study for in-depth exploration with a few cases (explanatory sequence) (Creswell & Clark, 2010). Albeit mixed methods may be time intensive and does not provide researchers the proficiency in a particular approach, it advances the exploitation of both approaches in a complementary manner (Neuman, 2014). To this end, scholars (e.g. Creswell, 2014) have admonished that researchers consider it as an essential part of knowledge creation. As hinted earlier, this study employs a quantitative approach to tackle the research objective and questions. The positivist paradigm adopted by this thesis, favours the quantitative 148 University of Ghana http://ugspace.ug.edu.gh approach which embraces both first (e.g. ANOVA, regression, correlation) and second (structural equation modelling) generational statistical techniques to conduct analysis (Hair et al., 2010). Aligned with the aim of the thesis which is to establish appropriate explanations for the interrelationships between ODA characteristics, consumer attitude toward online advertising and behavioural responses, the quantitative approach enables the researcher test and verify the associations among these constructs by developing clear hypotheses and producing or using numerical data. The approach as well provides the means to examine the direct relationship among the ODA characteristics and behavioural responses; the indirect relationship through the intervening role of attitude toward online advertising simultaneously; and also, the contingencies and alternative effects of user mode, nature of the advertised brand, and demographics (among other variables) respectively. The approach is also considered more suitable given that the study uses extant explicit theories to develop a conceptual framework in order to learn whether or not the theories match the observation. This represents a deductive stance to reasoning, which is an integral logical position in positivist research. The next section presents the research strategy adopted in this study.” 6.6 RESEARCH STRATEGY Research strategy refers to the procedures, tools and techniques used in collecting data for a given study in order to fulfil the purpose of the research (Saunders et al., 2011). According to Saunders et al. (2009, p. 600), it is the “general plan of how the researcher will go about answering the research question(s)”. Several techniques are at a researcher’s disposal, given the paradigm the researcher endorses or subscribes to, and the corresponding research approach they decide to use (Aliyu et al., 2014). Saunders et al. (2009) point out seven research strategies, namely: action research, case study, archival research, ethnography, experiment, grounded theory, and survey. The authors, nevertheless, submit that there are 149 University of Ghana http://ugspace.ug.edu.gh considerable overlays among these strategies, so it is vital that researchers choose the one that may be most useful for a given study. The selection of a research strategy is usually dependent on the nature of the research questions and objectives, the amount of existing literature on the phenomenon under investigation, and the availability of resources and time (Saunders et al., 2011). Within the domain of business and management research, Creswell (2014) identifies four strategies that have been predominantly used by researchers. These include, case study, experiment, archival research and survey. Case study in the words of Robson (2002) “involves an empirical investigation of a particular contemporary phenomenon within its real-life context using multiple sources of evidence” (p.178). Case study research strategy is suitable in situations where the issues to be examined are very complex, and highly embedded within an organisation in order to provide deep understanding of the context of the subject of enquiry as well as the processes involved (Yin, 2009). Archival strategies focus on trends by examining past issues, and how they change over time using administrative records and manuscripts as the major source of data (Saunders et al., 2009). The strategy, however, is mostly limited by the nature and quality of records accessible to the research. In experimental research, two groups (experimental and control) that share similar characteristics are used. The researcher usually subjects the experimental group to some form of manipulation (using the independent variable) and subsequently re-measures the outcome variables for both groups before and after the manipulation to establish causalities (Neuman, 2014). Although experimental strategies are minimally used in management research, they are the most favoured method in online advertising research given their edge over others in terms of internal validity (Liu-Thompkin, 2019). Surveys on the other hand, use questionnaires to collect information about the perceptions, evaluations and 150 University of Ghana http://ugspace.ug.edu.gh characteristics of a large group of people (Malhotra & Birks, 2007). Following from the objectives of this thesis as well as the chosen research approach, this study employs survey as a suitable corresponding strategy, and in the next subsection, the survey strategy is discussed in detail, and justifications for its selection are presented. 6.6.1 Survey as the Chosen Strategy for this Thesis “A survey is a method of gathering information about the opinions and attitudes of a large group of people (Creswell, 2014). According to Shaughnessy et al. (2012), surveys are commonly used in management research and are devised to directly enquire into the thoughts, feelings and opinions of respondents regarding a phenomenon. Surveys are highly favoured by studies conducted within the positivist paradigm and geared toward achieving systematic observation through structured research questions to achieve standardisation and consistency (Bryman & Bell, 2015). They are appropriate for studies concerned with collecting primary data about a population that may be too large to directly observe. A survey strategy requires a researcher to select a representative sample with characteristics that mirror those of the broader population, as well as carefully design standardised questionnaires to generate responses in a similar way from all participants (Babbie, 2004). Although, surveys commonly use questionnaires for data gathering, structured observations and interviews are sometimes used too (Creswell, 2014). Surveys allow researchers to gather quantitative data that can be analysed using various statistical techniques. They are mostly effective in situations of low researcher control over behavioural events; and provide the researcher the merit of examining several variables, as well as generalising research results gained from a sample to the entire population (Yin, 2009; Wimmer & Dominick, 2011). According to Robson (2002), surveys are also suitable for cross-sectional studies. Cross- sectional research are studies that examine a phenomenon using a cross-section of a 151 University of Ghana http://ugspace.ug.edu.gh population at a particular point in time, whereas its opposite, longitudinal studies, examine phenomena over an extended time period (Creswell, 2014). Because for the most part, academic research activities are time-bound, cross-sectional studies are the most common of the two, and commonly employ surveys (Saunders et al., 2011). Following from the above discourse, this study adopts the survey strategy since it aims to obtain direct responses from internet users and online consumers regarding their perceptions of ODA characteristics by employing structured questionnaires. As argued in the preceding chapters, the purpose of this study is to enhance theoretical and practical understanding of consumer behavioural responses to ODA in Ghana by examining the direct and indirect relationships among the ODA characteristics, consumer attitude toward online advertising, user mode and their behavioural responses. Given the large population of internet users, it is deemed necessary to gather data from a large number of respondents if the study is to produce reliable results. This necessitates the choice of a survey. It must also be highlighted that; surveys typically follow the deductive method of enquiry which is consistent with the methodological stance of positivism as adopted in this thesis. Since the objective of this study is to generate further insight about specific ad-related and consumer-related factors and the interrelationships among them using multivariate data analysis, the survey strategy seems most appropriate for this study. What is more, the choice of survey is based on the fact that the study is cross-sectional in nature, and surveys have been pointed out as suitable in such situations (Fink, 2009; Easterby-Smith et al., 2012). Malhotra and Birks (2007) outline various activities to be conducted by a researcher employing the survey strategy. Among these activities the authors highlight the need to design a sampling strategy by defining the study population, as well as constructing a survey instrument which operationalises the major constructs within the 152 University of Ghana http://ugspace.ug.edu.gh study. The next section is therefore dedicated to data collection methods with a focus on sampling methods, survey instrument design and administration.” 6.7 DATA COLLECTION METHODS To achieve the objectives of a given study, the researcher must make decisions regarding the sources from which data will be collected. As pointed out in the literature, researchers may choose from primary or secondary sources (Saunders et al., 2011). Primary data are collected for a particular study, and are direct reports of observations, while secondary data are collected for purposes other than the current problem commonly sourced from trade publications, personal records and so on (Malhotra, 2010). This study uses a cross-sectional survey to gather data, and because the study aims to examine the interrelationships among ODA characteristics, consumer attitude toward online advertising, and their behavioural responses as conditioned by user mode (goal orientation), it was vital to obtain responses directly from internet users and online consumers who are exposed to online display advertisements, and so, are considered primary sources. Considerations made in specifying the target primary sources are discussed next. 6.7.1 Population, Sampling Technique and Sample Size Determination For any study, the target population refers to all persons that hold the information the researcher seeks, and from whom the study can make deductions (Churchill & Iacobucci, 2009). A target population is considered the general collection of constituents from which a study sample is drawn and to which the study may generalise its findings. According to Malhotra and Birks (2007), as a way of reducing the difficulties involved in sample selection for research, it is required that researchers not only define their study population but also, do so as precisely as possible. When the target population of a particular study is appropriately defined, the researcher is afforded the prospect of directing the study 153 University of Ghana http://ugspace.ug.edu.gh recommendations and generalising findings to the appropriate audiences (Malhotra & Birks, 2007). The authors further assert that target population definitions should cover elements, sampling units, extent and time. The elements refer to the respondents in a survey research, and the sampling unit is said to contain the element. The extent describes the geographic confines of the survey, and time denotes a particular time-period under consideration (Malhotra & Birks, 2007). In light of these suggestions, this study considers all Internet users as its target population (sampling unit), and Internet users exposed to online display advertising as its element. Additionally, the study focuses on Internet users in Ghana, and limits exposures to ODA to a period of up to above six months (preferably, 12 months). While every researcher would wish to gather data from all elements of a target population (an approach referred to as census), this is mostly only feasible when the target population of a given study is very small. As such, studies involving large target populations, as is the case with this thesis, rely on sampling techniques to choose representative samples from the population of interest (Malhotra, 2010). Malhotra and Birks (2007) define sample as “a subgroup of the elements of the population selected for participation in the study” (p.405). Sampling is therefore, considered the selection of an adequate number of elements from a bigger population expecting that the data gathered from them will lead to making precise judgements and inferences about the entire population (Hair et al., 2010). The significance of sampling lies in its ability to expedite data gathering, guarantee more accurate results are produced, and make accessibility to potential respondents easier (Saunders et al., 2011). 6.7.1.1 Sampling Technique for the Study “The sampling techniques identified in literature are in two broad categories namely: probability and non-probability sampling techniques (Saunders et al., 2011). Probability sampling is commonly associated with surveys and experimental research where each 154 University of Ghana http://ugspace.ug.edu.gh element in the population has an equally known likelihood of being selected while the opposite is true for non-probability sampling which mostly relies on personal judgements (Saunders et al., 2011). This makes it difficult to make valid inferences about the population of interest in situations where the latter is employed. In spite of this, Saunders et al. (2009) point out that studies employing non-probability sampling could still make generalisations about the target population on theoretical, but not on statistical grounds. There are various types of probability and non-probability sampling techniques. However, this study employs purposive and snowball sampling – non-probability sampling techniques. Purposive sampling involves selecting participants based on the researcher’s discretion and judgment concerning who possesses the needed information to help achieve the research objectives (Saunders et al., 2011). This technique is considered appropriate for this thesis on the grounds that, though the target population or sampling unit (Internet users) is large, the researcher chose respondents that were highly informative. Hence for the purpose of this thesis, internet users were selected on the basis of their exposure and familiarity with online advertising. These were mainly initial contacts on several online platforms (social media, blogs and commercial sites). Using the snowball technique, these initial contacts were then asked to forward the link to their networks and acquaintances online. Because the study used predominantly an online questionnaire, sampled respondents were spread across several regions in the country. However, the offline questionnaire which complemented the online survey was administered in Accra at business centres, shopping malls, university campuses, offices etc. Although the study elicited responses from internet users, to make sure that potential respondents possess the information sought by the researcher, the questionnaire was structured to ensure that only internet users who report frequent exposure to online display advertising during their online sessions were included in the survey.” To do this, any participant who responded “never” to the preliminary question 155 University of Ghana http://ugspace.ug.edu.gh “how often are you exposed to online display ads while using the internet?” was not allowed to proceed to the next phase, and a pop-up information was displayed to said respondents. In the case of the offline survey, respondents were asked the qualifying question to be sure they fell within the sample population before a questionnaire was given to them. 6.7.1.2 Sample Size for the Study After the target population and sampling technique have been specified, the next logical step is to determine the sample size to be used for this study. A sample size refers to the overall number of elements, cases or participants to be included in a study (Malhotra, 2010). Various considerations that impact sample size determination include the research approach, nature of analysis, sample sizes used in prior similar studies as well as resource limitations (Malhotra & Birks, 2007). From a quantitative standpoint, it is reasoned that large sample sizes are prudent because they heighten the likelihood that the statistics derived from data analysis will mirror the precise estimates of the population, and minimise the possibilities of errors as the sample size increases (Malhotra, 2007; Hair et al., 2010). Although this appears to be agreed upon, scholarly perspectives regarding what qualifies as a large sample size is divergent as Tabachnick and Fidel (2007) aver that a sample size of 200 is reasonable and 300 is suitable; and Hair et al. (2010) also states that any number above 100 is fitting for statistical analysis. However, to achieve a suitable sample size determined on the basis of issues related to the current study, and the method of analysis to be employed, the researcher finds convincing two arguments. First is the assertion by Tabachnick and Fidell. (2007) that there should be a minimum of five-times more elements as there are items to be analysed. Second, is Nunnally’s (1978) early submission that researchers apply a 10 to1 ratio of respondents to scale items. Since this study comprises 55 scale items (see Table 6.2), applying these two 156 University of Ghana http://ugspace.ug.edu.gh perspectives provide an upper boundary sample size of 550 and lower boundary sample size of 275. Based on these two recommendations, as well as cues from prior survey-based online advertising studies (e.g. Wang & Sun, 2010c; Seyedghorban et al., 2016; Nasir, 2017), the study estimated 550 as a suitable sample size from whom data could be gathered. Owing to the combination of online and offline administration of the survey instrument, as well as the blend of purposive and snowball sampling employed, the total number of completed surveys returned were 886, far exceeding the estimated sample size. These responses were coded and inputted into SPSS, after which the data set was screened and cleaned for wrongly inputted scores, outliers and missing responses. During the screening process, 7 cases had missing responses in a number of sections, and 12 cases responded “never” regarding their frequency of exposure to online display ads when using the internet. These 19 cases were discarded resulting in a final sample of 867 cases or respondents. After this, based on some scholarly precedents (e.g Souiden et al., 2017) the data set was split into two; with 275 cases used for the EFA and 592 cases used for the CFA and the other major analyses. The data collection process was facilitated by utilizing structured questionnaires designed after an extensive review of the extant literature was conducted. The survey instrument design and administration are detailed out next. 6.7.2 Survey Instrument Development and Administration The survey instrument was designed based on the study’s objectives and research questions following a thorough review of literature from which the components of the research framework were drawn. The instrument was a four-page questionnaire developed to consists of four main sections (see Appendix B). The first section captured information on respondents’ internet usage activities, motives and familiarity with online advertising. The second section examined issues relating to respondents’ perceptions of ODA characteristics; 157 University of Ghana http://ugspace.ug.edu.gh and the third section obtained information regarding their attitude toward online advertising, and behavioural responses. The fourth and final section comprised internet users’ demographic information that profiled the respondents in terms of gender, age, educational level, employment status, nationality, and region of residence. The major construct or variables were operationalised by adapting measurement items from extant online advertising research, as well as culling others from the ODA literature reviewed. These measurement/scale items were calibrated on a seven-point Likert scale. Specifically, respondents were asked to rate their perceptions on scales ranging from 1 (strongly disagree) to 7 (strongly agree). Table 6.2 provides a summary of the major constructs in the study framework as measured in the questionnaire, with their corresponding number of measurement items and sources. Table 6.2 Summary of Study Constructs and Sources Study Construct Number of Items Literature Source Interactivity 8 scale items Ko et al. (2005), Campbell and Wright (2008), Gao et al. (2009) Placement 7 scale items Segev et al. (2014). Informativeness 8 scale items Wang and Sun (2010a), Taylor et al. (2011) Personalisation 7 scale items Van Dorn and Hoekstra (2013), Bleier and Eisenbess (2015a; 2015b) Exposure Condition 8 scale items Chan et al. (2010), Wang et al. (2013), Kim (2018) Attitude toward online advertising 5 scale items Wang and Sun (2010a; 2010b) Wolin et al. (2002) Ad Acceptance: Passive 3 scale items Tang et al. (2014), Goodrich et al. (2015), Belanche et al. (2017) Active 3 scale items Tang et al. (2014), Goodrich et al. (2015), Belanche et al. (2017) Ad Avoidance: Passive 3 scale items Tang et al. (2014) Active 3 scale items Cho and Cheon (2004), Tang et al. (2014) 158 University of Ghana http://ugspace.ug.edu.gh As indicated earlier, the choice of a questionnaire was based on the research purpose and objectives, which is to examine the interconnections among variables using a quantitative approach. Its cost effectiveness was also considered because the population and the sample for the study is large, and moreover, its standardised form offers some level of reliability and eases the process of data analysis as well as makes it more appropriate for a survey (Smith & Albaum, 2005; Malhotra, 2010). In spite of its advantages, it must be mentioned that written questionnaires come with certain challenges. For instance, statements may be improperly constructed which creates the possibility of wrongful interpretation on the on the part of the respondents. Additionally, it has been argued that questionnaires restrict respondents to options selected by the researcher which may not be very representative of the exact situation (Malhotra & Birks, 2007). To curb these challenges, the initial questionnaire draft was pre-tested as suggested by scholars (e.g. DeVellis, 2003; Saunders et al., 2011). 6.7.2.1 Pre-testing, Questionnaire Modification, and Administration The essence of pre-testing the questionnaire was to ascertain its suitability as well as assess wording errors and any ambiguity that may be found in the questionnaire before it is administered. It is advised that respondents for pre-testing should bear similarities with the actual population of a given study in order to correctly screen measurement items for their appropriateness (Cooper et al., 2006). In this regard, after the research supervisor evaluated the questionnaire, and following his suggestions, some statements were re-worded for clarity, six (6) final year PhD students in the Department of Marketing and Entrepreneurship of the University of Ghana Business School who are internet users and show familiarity with online advertising, as well as two (2) employees from digital advertising firms, and twelve (12) other internet users totalling twenty (20) pre-test sample were used for a pilot test of the initial draft. These individuals were provided with links via WhatsApp, emails, 159 University of Ghana http://ugspace.ug.edu.gh and other social media platforms to fill an online version of the questionnaire. Following their recommendations, further minor revisions were made to rectify clerical and grammatical errors found. The pre-testing process also ensured that the scale items were relevant to, and reflective of the constructs they are designed to measure. The final version of the questionnaire administered to the target sample is provided in Appendix B. Concerning the process of administering the questionnaire, an online survey was the primary mode of administration complemented by a “paper and pencil” or offline version of the questionnaire. Online surveys are considered adequate means of reaching respondents who are familiar with the Internet and online activities (Souiden et al., 2017), and are predominantly used in survey-based online advertising research (e.g. Valaei et al., 2016; Nasir, 2017). The online questionnaire was designed using google forms (a survey administration application), and the link to the questionnaire was sent to respondents’ emails, shared on social media pages and platforms (e.g. Facebook, Twitter, WhatsApp and LinkedIn). The survey link was also displayed on some blogs. To be sure, responses were based on online display advertising, which is the focus of this thesis, at the opening part of the questionnaire, display ads were explained. Additionally, to guarantee that the required sample that truly reflects the population is acquired, the online survey was supplemented with an offline survey, in which respondents were intercepted at business centres, shopping malls, university campuses and workplaces after permission and approval was sought from the appropriate authorities (management and administrators). In such instances, with the help of research assistants, respondents were approached, and their consent was sought to help fill the questionnaire by explaining to them the purpose of the study. The data collection process lasted for four weeks, that is between September and October 2019. 160 University of Ghana http://ugspace.ug.edu.gh 6.8 METHODS OF DATA ANALYSIS This section discusses the data analysis processes or techniques used in this study. Following from the discussions on the positivist stance, explanatory purpose, quantitative and deductive approach, and survey as the research design guiding this study, the thesis correspondingly applies quantitative methods of data analysis. The study employs specifically, descriptive analysis, and multivariate data analyses such as factor analysis (EFA and CFA), Structural Equation Modelling (SEM), Analysis of Variance (ANOVA), and Logistic Regression through the use of two analytical software; Statistical Package for Social Sciences (SPSS) version 22.0 and its add-on Analysis of Moment Structures (AMOS) version 22.0. These two are common software used in social science research and more so by online advertising researchers (e.g. Goodrich 2011; van Reijmersdal et al., 2016; Ham, 2017) making them appropriate for this study. For initial coding, inputting and cleaning of the study data, SPSS was used. The data cleaning or screening process helped identify missing values, and outlying responses, as well as assess the normality of the data set. Since data sets subjected to factor analysis and SEM (the major analytical techniques used in this study) are required to be normally distributed, the screened data set was subsequently tested for normality using skewness and kurtosis. The results from the test produced values that were within the recommended level and parameter – closer to zero (Pallant, 2011), and by this the univariate normality of the data set was established. After these requirements were met, preliminary analysis (such as descriptive statistics and EFA) were conducted still using SPSS. The final data set was then transferred to AMOS for further substantive analysis. 6.8.1 Factor Analysis After the relevant initial analysis were conducted, factor analysis was used to define the structure of the study constructs. Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) are two major factor analysis approaches (Field, 2013). Although both 161 University of Ghana http://ugspace.ug.edu.gh approaches are used in this study, this section focuses on EFA, and CFA is discussed in the subsequent section. EFA is a statistical technique used in the initial stages of multivariate data analysis to examine the interrelationships among large numbers of variables by ensuring their propensity to be grouped together (Bryman & Bell, 2011). It is a technique that helps reduce large data sets into lesser more manageable sets of factors in order to eliminate issues of multicollinearity and improve their measurement quality among other things (Tabachnick & Fidell, 2013). Particularly, since the scale items were not adopted (i.e. some were adapted and others were drawn from the literature), an EFA was therefore, carried out in this study to refine and streamline the structure of the constructs/factors before further analysis (CFA and SEM) was conducted. The data set was, explored in line with the stages advanced by Pallant (2011) which are: assessing the suitability of the data for factor analysis, extracting the factors, and rotating and interpreting the factors. These phases are detailed next. 6.8.1.1 Assessing the Appropriateness of the Data Set for Factor Analysis “To ensure the suitability of a data set for analysis, two key issues are given consideration – the sample size and the strength of the association among the scale items (Pallant, 2011). As stated in an earlier section, scholars (e.g. Nunnally, 1978; Hair et al., 2006; Tabachnick & Fidell, 2007). recommend a 5 to 1 or 10 to 1 ratio, that is ten or five cases for each item to be explored. Since these suggestions were applied in determining the sample size for study, a reasonably large data set of 867 (317 cases more than the estimated sample size was retrieved and found usable) cases was used in carrying out the factor analysis. Specifically, for the EFA, the 5 to 1 ratio (Tabachnick & Fidell, 2007) was applied resulting in 275 cases being used for exploration, and the remaining 592 cases were used for the CFA and other more substansive analyses. 162 University of Ghana http://ugspace.ug.edu.gh The second issue that addresses the factorability of the data set is the strength of association among the scale items, and three statistical measures are used to verify this: the correlation coefficients, the Bartlett’s test of sphericity, and the Kaiser-Meyer-Oklin (KMO) measure of sampling adequacy. First, it is recommended that the correlation among the items should produce a much higher number of coefficients that are greater than 0.3 than those below (Tabachnick & Fidell, 2007). Second, when generated, using SPSS, the Bartlett’s test of sphericity should be significant at p < 0.05 (Bartlett, 1954; Hinton et al., 2014), and third, the KMO index although ranges from 0 to 1, a minimum value of 0.6 (p ³ 0.6) depicts a suitable data set that can be factor analysed (Tabachnick & Fidell, 2007). These three criteria were satisfied by the study data set.” 6.8.1.2 Factor Extraction, Rotation and Interpretation “Factor extraction comprises determining the fewest number of factors that best depict the interconnections about the items (Pallant, 2011). This study extracted an ideal number of factors using the maximum likelihood method assisted by Kaiser’s criterion (eigen value rule). By virtue of this rule, all factors with eigen values of 1.0 and above were maintained for further analysis. To aid in the interpretation of the resultant factors, the Promax rotation method with Kaiser normalisation was used to assess the number of strong loadings as well as specific items that loaded considerably onto the various factors/components (Pallant, 2013). The Promax method is an Oblique rotation technique that is ideal for larger data sets, and affords the researcher the ability to derive a simple structure by ensuring that each item loads strongly on only one component, and each component is represented by a number of high loading items (Tabachnick & FIdell, 2013). The loadings, therefore, simply describe the correlation between the items and their factors.” 163 University of Ghana http://ugspace.ug.edu.gh 6.8.2 Structural Equation Modelling (SEM) Following the preliminary data analysis procedures and EFA, a two-stage SEM was consecutively conducted based on the study’s objectives. SEM appears to be a principal multivariate technique used in statistical analysis, particularly, in management and marketing studies that intend to assess cause and effect relationships among several latent constructs (Hair et al., 2011). Structural Equation Modelling (SEM) describes “a system of equations that establishes the structure of relationships among observed and unobserved (latent) quantitative variables” (McQuitty & Wolf, 2013, p.59). According to McIntosh (2007) SEM is a family of statistical methods designed to test a conceptual or theoretical model. The technique involves measuring a model that defines the latent variables using one or more observable variables and a structural regression model that links the latent variables together (Hair et al., 2010). Though some researchers (e.g. Bagozzi & Yi, 2012) caution that SEM does not assess or “prove” causality in the actual sense of its usage as in the natural sciences, others (including Byrne, 2013) also contend strongly that the technique tests and evaluates the influence one construct may have on another within a fully specified model. “The choice of SEM as the major data analysis technique for this study is founded on scholarly arguments that it helps deal with the challenges of first-generation statistical methods. For instance, SEM, combines elements of path analysis, factor analysis and multiple regression which permits the assessment of complex interrelated dependent relations among variables (Schumacker & Lomax, 2010; McQuitty & Wolf, 2013,). The technique also simultaneously incorporates the influence of measurement errors within the relationship into its structural coefficients, and measures the unidimensionality, reliability and validity of each construct in the model (Kline, 2005). According to Bagozzi and Yi (2012), SEM provides an integrative function and (1) helps researchers to be more precise in specifying hypotheses, and operationalising constructs, (2) takes into consideration 164 University of Ghana http://ugspace.ug.edu.gh reliability of measures in tests of hypotheses in ways that transcend the averaging of multi- measures of constructs, (3) guides exploratory and confirmatory research in a manner combining self-insight and modelling skills with theory. SEM is therefore, suitable in aiding this study specify, estimate, and test the research model by means of a causal path diagram that depicts the hypothesised interrelationships among the study variables as well as allow the modification and deletion of any causal paths that do not fit with the principal model (Kline, 2011).” 6.8.2.1 Two-Stage SEM “There are two widely used approaches in performing SEM: one-stage and two-stage. The one-stage approach processes the analysis of both the measurement and structural models simultaneously (Kline, 2005; Schumacker & Lomax, 2010) while the two-stage approach, separates the measurement model and structural model estimation (Hair et al., 2010). On the basis that the two-stage approach avoids interaction that is needless between constructs during testing of the structural model (Anderson & Gerbing, 1988), this study uses it to test the research model as done by prior online advertising studies (e.g. Wang & Sun, 2010a; Souiden et al., 2017).” 1st Phase - Measurement Model/Confirmatory Factor Analysis (CFA): At the measurement stage, the measurement model was specified indicating the relationship between the various constructs and their measures. This was done by conducting a CFA. As the first stage or the basis of SEM, CFA is “a version of factor analysis in which specific hypotheses about structure and relations between the latent variables that underlie the data are tested” (Field, 2013, p. 872). CFA is used to propose relationships between the observed measures and a-priori theoretical pattern of factors and then assess the hypothesised model statistically (Byrne, 2013). The CFA, therefore, shows whether the model matches the actual 165 University of Ghana http://ugspace.ug.edu.gh data that was gathered by examining reliability and validity (convergent and discriminant) (ibid.). Convergent validity of the constituents of the research model is ensured by ascertaining that the measurement items of the various constructs are correlated, and they are also correlated with the constructs they measure (Neuman, 2014). Discriminant validity on the other hand assesses the existence of high correlation between the various constructs (i.e. above 0.90), as well as empirical differences or co-variation among them (Kline, 2011). Specifically, Average Variance Extracted (AVE) was computed, and the squared correlations between the constructs were compared and used as indicator of convergent validity and discriminant validity of the research model. Internal consistency of the research model was confirmed using Composite Reliability (CR), AVE, and Cronbach’s alpha. These indicators are discussed in detail in a later section on ‘reliability and validity of the research instrument’. 2nd“ Phase – Structural Model: At this phase, the structural model was specified, depicting how the various constructs are interrelated (Tabachnick & Fidell, 2007). During the structural model specification, the non-structural covariances among the unobserved factors are substituted with the hypothesised structure, and the data is reanalysed. Alternately stated, this stage allowed for the testing of the hypothesised relationships among the various constructs in the study – ODA characteristics, attitude toward online advertising (ATOA), user mode and behavioural responses (ad acceptance and ad avoidance). The appropriateness of the measurement and structural model were confirmed using the goodness-of-fit measures, and the significance of the various paths among the constructs and their measures were as well examined using the coefficient parameter estimates (Hair et al., 2010).” 166 University of Ghana http://ugspace.ug.edu.gh 6.8.2.2 Evaluating the Fitness of the Model “Model fitness evaluation involves the interpretation of how suitably the research model fits the empirical data and results. Scholars (such as Hair et al., 2010; Iacobucci, 2010) have suggested various measures researchers can use to evaluate the general acceptability of both the measurement and structural models in a given research. This procedure is essentially comparative in nature since it involves choosing between various fit indices that subjectively show whether the data matches the theoretically postulated model (Bagozzi & Yi, 2012). The various goodness-of-fit indices with their cut-off criteria (conventionally acceptable values) proposed by scholars are categorized into three clusters namely; (1) absolute fit indices, (2) comparative fit indices and (3) parsimonious fit indices. The absolute fit indices also known as the predictive fit indices provide a fundamental assessment of how well the data gathered matches or is close enough to the hypothesized model (Hair et al., 2010). Fit indices commonly used to check for absolute fit include the Chi-square (χ²) statistics, Goodness-of-Fit Index (GFI), Adjusted Goodness-of-Fit Index (AGFI), Standardised Root-Mean-Square Residual (SRMR), and Root-Mean-Square-Error of Approximation (RMSEA). These criteria are based on differences between the observed and model-implied correlation or covariance matrix (Byrne, 2013). The comparative (incremental) fit indices compare the fit of the given research model with an estimated baseline model (null model) which operates under the assumption that all the observed variables are uncorrelated (Kelloway, 1998; Lombardi & Pastore, 2012). The fit indices under this category include the Comparative Fit Index (CFI), the Incremental Fit Index (IFI), the Relative Non-centrality Index (RNI), and the Tucker-Lewis Index (TLI). The parsimonious fit indices, however, permit the researcher to determine which model out of a set of competing models is the most suitable in terms of complexity (Hair et al., 2010). 167 University of Ghana http://ugspace.ug.edu.gh Measures of parsimony fit include the Parsimony-adjusted Normed Fit Index (NFI) and the Parsimony -adjusted Comparative Fit Index (CFI). In this study, the baseline fit indices used to ascertain the acceptability of the construct measures or how best the measurement and structural model fit the study data include the Root Mean Square Error of Approximation (RMSEA £ 0.08), and the Chi-square/degrees of freedom (χ2/df £ 2 or 3). The choice of the Normed Chi-square (χ2/df) stemmed from scholarly contentions that albeit the χ2 is a principal measure of absolute fit, it is a function of the sample size, and the difference between the observed and estimated covariance matrices (Hair et al., 2010). As such the value of the χ2 increases with increase in sample size causing challenges in achieving model fit. On the basis of this argument, the (χ2/df) ratio has been suggested as a suitable insulation against sample size influences (Byrne, 2010). The study also employed the Goodness-of-Fit Index (GFI ³ 0.90), the Comparative Fit Index (CFI ³ 0.90), Incremental Fit Index (IFI ³ 0.90), Normed Fit Index (NFI ³ 0.90), and the Tucker-Lewis Index (TLI ³ 0.90). These indexes were chosen on the basis that they are the universally acceptable criteria in social science research (Byrne, 2013).” 6.8.2.3 Mediation Analysis According to Aguinis et al. (2016, p.2) mediation refers to “the underlying mechanisms and processes that connect antecedents and outcomes”. Mediation occurs when an intervening variable conveys the effect of a precursor variable on an outcome partly or fully (Ndofor et al., 2011). It, therefore, points to the presence of a third variable “which represents the generative mechanism through which the focal independent variable is able to influence the dependent variable of interest” (Baron & Kenny, 1986, p.1173). Testing of mediation effects has a long tradition in social science research, and researchers have espoused and 168 University of Ghana http://ugspace.ug.edu.gh adopted several approaches to conducting mediation analysis (Mathieu et al., 2008). Four approaches that have been promulgated in the literature in addressing mediation effects include the causal steps approach (Baron & Kenny, 1986), the product of coefficients/ Sobel test (Sobel, 1986), the empirical M-test/ distribution of products approach (Holber & Stephenson, 2003), and the bootstrapping method (Preacher & Hayes, 2008). “The bootstrapping method is employed in this study to examine the mediating influence of attitude toward online advertising (ATOA) in the relationship between the ODA characteristics and behavioural responses. Bootstrapping involves generating an empirical representation of the sampling distribution of the indirect effect by dealing with the given sample as a microcosm of the broader population. The given sample is then used to produce another sample via an iteration process of replacement making it possible to put back cases initially drawn so they could be drawn again (Hayes, 2009). To state this concisely, bootstrapping involves repetitively sampling from the data and estimating the indirect influence in each resampled data. This results in a given number of estimates of the indirect effect, which functions as an empirical approximation of the sampling distribution used to establish confidence intervals of the indirect effect (Preacher & Hayes, 2008). The bootstrapping method is considered one of the suitable approaches for addressing mediating effects (Williams & MacKinnon, 2008) and was chosen for this study because as a nonparametric re-sampling method, it edges the Sobel test and the causal steps approach, since it does not impose assumptions of normality on the sample distribution. What is more, this method of testing for intervening variable effects, besides enabling the researcher to control for the incorrect rejection of a valid null hypothesis (Type 1 error), also helps in making inferences about the indirect influences in a given model irrespective of model complexity and/or how many the direct paths are (Hayes, 2009). It has also been argued by 169 University of Ghana http://ugspace.ug.edu.gh researchers (e.g. MacKinnon & Fairchild, 2009; Zhao et al., 2010) that the presence of a direct relationship between the predictor and outcome variables required by the causal steps approach for mediation to be determined is quite irrelevant in bootstrapping. Also, bootstrapping in comparison to the empirical M-test is less cumbersome in terms of computational burden since the method is incorporated in most SEM software (e.g. AMOS which is the software of choice for this study). In view of this, a bootstrap sample of 2000 as recommended by previous studies (Montoya & Hayes, 2017) is used to examine the indirect influence of ODA characteristics on behavioural responses through attitude toward online advertising (ATOA).” 6.8.2.4 Moderation Analysis Moderation refers to the function of a third variable “which partitions a focal independent variable into subgroups that establish its domain of maximal effectiveness in regard to a given dependent variable” (Baron & Kenny, 1986, p.1173). Moderation depicts the notion that the extent or level of the effect of an antecedent on an outcome depends on contingency factors; and thus, describes the circumstances under which an effect may vary in size (Aguinis et al., 2016). In essence, it suggests an interaction effect which can (1) increase the influence of the independent variable on the dependent variable; (2) reduce the influence of the independent on the dependent variable; or (3) reverse the influence of the independent on the dependent variable. According to Baron and Kenny (1986), any qualitative or quantitative variable that affects the direction and/or strength of the relationship between a predictor variable and an outcome variable can function as a moderator. Two principal approaches to conducting moderation analysis in SEM present in the literature are interactions and multi-group moderations (Boyd et al., 2012). Interactions are used when the moderator variable is continuous, and in such instances, researchers examine 170 University of Ghana http://ugspace.ug.edu.gh the relationship between a predictor (X) and an outcome (Y); a second predictor (Z) hypothesised to be a moderator; and then a product term between X and Z. The estimates or coefficients for the XZ product term then provide information on the presence and magnitude of the moderating effect (Little et al., 2007; Aguinis et al., 2016). In multi-group moderations however, the moderator variables are mostly categorical which calls for subgrouping analysis, and a comparison of the resultant estimates across the various subgroups or categories (Boyd et al., 2012). Baron and Kenny (1986) also suggests that in cases where both the independent variable (X) and moderator variable (Z) are continuous, and the researcher believes that the moderator changes the X-Y relationship in a stepwise function, then the moderator variable (Z) can be dichotomised for a multi-group moderation. As has been reiterated in the previous chapters, the third objective of this study is to assess the moderating effect of user mode on the relationships between ODA characteristics, and attitude toward online advertising as well as behavioural responses. This objective stems from the telic-paratelic viewpoint of the reversal theory, which proposes that internet users in telic (serious or goal directed) and paratelic (playful or experiential) modes form different attitudes toward, and respond differently to online advertising (Jung et al., 2014). For this reason, this study employs the multi-group approach to analysing the moderation effect of user mode on the focal relationships. To achieve this, because the respondents’ internet usage motive was measured as a categorical variable, respondents with shopping and (re)searching motives were coded and classified as telic users, and those with communication and surfing motives were classified as paratelic users as done in prior studies (Seyedghorban et al., 2016). This was done to ensure that the right categorisations are derived. 171 University of Ghana http://ugspace.ug.edu.gh 6.8.3 Analysis of Variance (ANOVA) “Analysis of variance (ANOVA) is another analytical tool employed in this study. ANOVA is used when a study seeks to compare the mean scores of two or more groups on a continuous variable (Pallant, 2011). Following the fourth objective of the study which addressed the differences in behavioural responses based on the nature (product vs. service) of the advertised brand, the study employs specifically, a one-way ANOVA. One-way ANOVA is suitable when a study examines the effect of a single predictor on an outcome and since the objective was to assess whether the behavioural responses of the study sample differed on the basis of the nature of the brand (product vs. service) featured in the ODA they were exposed to, for which they provided responses, one-way ANOVA was deemed appropriate for tackling this aspect of the study. It is essential to state that there are two main types of one-way ANOVAs namely; repeated measures ANOVA and between-groups ANOVA (Tabachnick & Fidell, 2013). The former collects data from the same respondents on multiple (more than two) occasions while the latter also known as independent sample ANOVA, gathers data in a single instance from two or more different groups of respondents (Pallant, 2013). Given the cross-sectional nature of this study and the stated objective for which one-way ANOVA was employed, the study used the independent sample/ between- groups ANOVA.” 6.8.4 Logistic Regression Following results from the ANOVA test, the study further assesses the intensity of ODA- specific responses among respondents on the bases of the nature of the advertised brand, and logisitic regression was employed to address this aspect of the study. Logistic regression is used to test the predictive power of a set of variables and to assess the relative contribution of each individual variable, particularly when the outcome or dependent variable is categorical (Pallant, 2013). Since the study sought to address the effect of the resultant ODA 172 University of Ghana http://ugspace.ug.edu.gh characteristics that passed through the structural equation modelling phase on the two- dimensional view of behavioural responses between product-featured and service-featured ODAs, the study used specifically, binary logistic regression. To do this, the two behavioural response variables (ad acceptance and ad avoidance) were categorized/dichotomised into active-passive segments, and a multigroup binary logistic regression was used to examine the impact of the ODA characteristics on the likelihood of respondents reporting active (since the passive segment was used as a dummy) behavioural responses for product-featured ODAs, and service-featured ODAs. 6.8.5 Unit of Analysis The unit of analysis refers to the type of element a researcher utilizes in measuring a given research phenomenon (Neuman, 2014). It is described as the level of data aggregation during analysis or the extent to which the level of enquiry or data gathered, focuses on specific object(s) (Zikmund, 2003). The object(s) could be a whole organisation, departments, groups, or individuals which are standard units of analysis in research in the social sciences. Among these, the individual is by far the commonest unit of analysis in surveys (Babbie, 2004). This study focuses on Internet users exposed to online display advertising. Since the study examines ODA characteristics that shape consumer attitude toward online advertising, and their behavioural responses, how their user mode may cause variations in these outcomes, as well as the likely difference that may occur in their responses based on the advertised brand (product or service), it is only logical to establish these interrelationships from the Internet user perspective. The individual (Internet user) is therefore a suitable unit of analysis for this study. The next section discusses the reliability and validity issues considered in this study. 173 University of Ghana http://ugspace.ug.edu.gh 6.9 RELIABILITY AND VALIDITY It is important that the procedures by which researchers assess the credibility and accuracy of their research findings are communicated (Creswell, 2014). This has been argued from the viewpoint that adopting and adapting items from extant literature or prior studies may potentially affect the originality of instruments used for a particular study. According to Creswell, this necessitates the need to assess the validity and reliability of such instruments when conducting data analysis. Reliability and validity though closely related, are independent concepts on the basis that, an instrument that is reliable may not necessarily be valid (Zikmund et al., 2009). On this premise, both reliability and validity were successively assessed to ensure that accurate conclusions can be drawn, and precise generalisations can be made from the findings of this study. The specific approaches to assessing the reliability and validity of the research instrument are explained in the succeeding sub-sections. 6.9.1 Reliability of the Research Instrument “Reliability explains the assessment of the magnitude of consistency or regularity among the items measuring a construct, which lends credibility to the findings of a study such that these findings could be reproduced (Burns & Burns, 2008; Hair et al., 2010). To put this differently, it is the extent to which the research instrument and analysis processes generate coherent replicable findings. From other scholarly viewpoints, reliability depicts the level to which measurement scales are devoid of arbitrary error so that the higher the reliability value, the lower the amount of error (Zikmund et al., 2009; Pallant, 2013). Research in the social sciences have mostly assessed reliability using two dominant approaches namely: internal consistency and test-retest reliability.” Internal consistency “is the degree to which the items that make up the scale are all measuring the same underlying attribute (i.e. the extent to which the items ‘hang together)” (Pallant, 2011, p.6). Test-retest on other hand, measures the correlation between scores obtained after an instrument is administered to the 174 University of Ghana http://ugspace.ug.edu.gh same group of respondents at two different times in order to test for stability or the degree to which the instrument produces similar outcomes in both successive instances (Hair et al., 2010). “For the purpose of this thesis, the internal consistency approach to assessing reliability is utilised. As mentioned in an earlier section, specifically, Cronbach’s Alpha (a) and Composite Reliability (CR) were employed to determine the reliability of the research instrument. Cronbach’s Alpha which measures the inter-correlation between various items representing a construct, is the predominantly used estimate of the reliability of a multiple item scale (Pallant, 2011). The literature suggests that Cronbach’s Alpha values/estimates closer to 1 show that the instrument and data gathered are very reliable, while coefficients closer to 0 show that the data is not reliable (Pallant, 2013). There are differing suggestions on the acceptable threshold for Cronbach’s Alpha estimates. According to Malhotra (2010), values between 0.7 and 0.9 are acceptable limits or criteria for reliability while Hair et al. (2006) argues coefficients of 0.6 and 0.7 to be the lower boundaries for acceptability. Although Cronbach’s alpha is the most widely used measure of reliability, it has been contended that its value or estimate is affected by the number of items measuring a construct (Streiner, 2013). It is therefore, criticised for not being a sufficient measure of overall reliability of a research instrument since it focuses on individual constructs (Botha & Van de Waldt, 2011).” Owing to the deficiencies of Cronbach’s alpha, at the confirmatory stage of data analysis (CFA), Composite Reliability (CR) tests and examination were conducted to complement the Cronbach’s Alpha estimates in order to determine the total reliability of the research instrument. Composite reliability measures the general reliability of a collection of diverse but similar items. According to Hair et al. (2014), a composite reliability score that is lower 175 University of Ghana http://ugspace.ug.edu.gh than 0.6 is an indicator of weak internal consistency while indicator values of 0.6 and above are considered acceptable. Evidence of the reliability measures are presented at both the EFA and CFA stages in the next chapter on data analysis. 6.9.2 Validity of the Research Instrument “Validity describes the match or fit between a construct and its measurement items (Burns & Burns, 2008), and assesses the quality of data and its corresponding results (Creswell and Clark, 2007). Essentially, it represents the accuracy of measurement items for a particular purpose. The main categories of validity commonly used in social science research are face validity, content validity, criterion validity, and construct validity (Pallant, 2011; Neuman, 2014). Face validity concerns how well items add up as a measure of a construct based on the judgements of others with expertise in the area. It is considered the subjective consensus among researchers and practitioners in the particular area of study that, the contents of a given scale soundly seem to measure what they intend to measure (Huck, 2012). Content validity measures the degree to which scale items correctly measure a particular construct by adequately and representatively capturing all the facets of the conceptual definition of said construct (Neuman, 2014). Criterion validity relies on outside verification by comparing other measures of the same construct in which the researcher has confidence. Construct validity, however, depicts how well the items measuring each individual construct converge, and how well the items measuring the various separate constructs diverge (Neuman, 2014). By virtue of this explanation, there are two types of construct validity namely; convergent and discriminant validity (Hair et al., 2010). Discriminant validity shows the degree to which a construct is different from others by ensuring that the items that measure it are unique indicators of the construct (Streiner, 2013). It is indicated by the low correlation between the measure of 176 University of Ghana http://ugspace.ug.edu.gh interest and the measures of other constructs (Malhotra, 2010). Convergent validity ensures that the constructs identified are truly reflected by their measures, and these measures share a high amount of common variance (Hair et al., 2010). In this study, the research instrument was validated using face validity, content validity and construct validity. Adhering to the suggestions of researchers such as Ghauri and Gronhaug, (2005) and Hair et al. (2010), that a simple test for face and content validity is to obtain the viewpoints of others knowledgeable in the study area as well as pre-test the research instrument, content and face validity was guaranteed by allowing experts (both academics and practitioners) to review the scales used in the study after which the questionnaire was pilot-tested. Construct validity was also used to detect how well the results obtained using the research instrument fit theoretical expectations (Neuman, 2014). Construct validity was established by analysing the convergent and discriminant validity (Hair et al., 2010) through the AVE values as well as the comparisons between the square-root of the AVEs and the inter-construct correlations estimated. The suitable limit for AVE score is 0.5 and above (Hair et al., 2010), and results for validity measures are also presented at the CFA stage in next chapter. 6.10 ETHICAL CONSIDERATIONS Ethical issues describe the appropriateness of a researcher’s conduct concerning the rights of a study’s participants and are essential considerations in any research. (Saunders et al., 2009). A key ethical issue considered in this study was to obtain informed consent or the willful agreement of individuals to participate in the research based on their understanding of the purpose and nature of the study. To do this, an application was submitted to the Ethics Committee for Humanities (ECH) at the University of Ghana, where ethical clearance (see Appendix C) was given for the conduct of the study. 177 University of Ghana http://ugspace.ug.edu.gh On the basis that informed consent has four elements; disclosure, competence, comprehension and voluntariness (Jabreen 2012), the protocol consent form issued by the ethics committee, had the researcher detail out the purpose of the study. Besides, the study purpose was also provided on the questionnaire in order to steer clear off misunderstanding and lack of confidence. Also, the study ensured that participants possess some knowledge of the phenomenon being studied, had the ability to provide the information sought by the study, and understood the particular reason they were to partake in the study. In addition, no specific details related to the identity of the participants were taken, and because the data was analysed quantitatively, matching responses to a specific respondent was impossible. Finally, the study respondents were made aware that participation was voluntary, and they may decline and withdraw from the study at any time without any penalty. 6.11 CHAPTER SUMMARY This chapter described the various research methodological approaches used in the study. Arguments for the choice of positivism and its various philosophical assumptions have been presented. Grounded on the arguments made for a positivist paradigmatic position coupled with the objectives of the study, a quantitative and deductive approach was considered the most suitable design. The quantitative research method or approach is presented requiring the discussion of specific methodological issues. In view of this, the chapter established a cross-sectional survey as the strategy employed in the study. The chapter then clearly described the procedures used in the questionnaire design, respondents’ selection, sample size determination, and data collection. The chapter concludes by describing the main analytical methods employed in the study and provides support for their appropriateness in this study. In the next chapter, detailed discussions are presented on data analysis, interpretations and results with a focus on sample characteristics, descriptive statistics, Exploratory Factor Analysis (EFA), two-stage Structural Equation Modelling (SEM), Analysis of Variance (ANOVA), and logistic regression. 178 University of Ghana http://ugspace.ug.edu.gh CHAPTER SEVEN DATA ANALYSIS AND PRESENTATION OF EMPIRICAL RESULTS 7.1 CHAPTER OVERVIEW This chapter presents results from the data analysis and includes two preliminary sections and five major analytical sections as discussed in the methodology. The chapter starts off with discussions on sample characteristics. After the sample characteristics are discussed, the chapter presents results of the preliminary data analysis which involved descriptive statistics and exploratory factor analysis (EFA). Assessment of common method variance which was conducted during the data exploration phase is subsequently discussed. Next the chapter presents results of the confirmatory factor analysis (CFA) for the latent variables. At the confirmatory stage, various reliability and validity tests on the scales used in the study are conducted to validate and substantiate the final model obtained. The third analytical section focuses on the results from the structural model, which tested the study hypotheses depicted in the conceptual framework thus, examining the effect of ODA characteristic on attitude toward online advertising (ATOA) and behavioural responses. This section includes the test for the mediation and moderation effects of ATOA and internet user mode respectively. The fourth and fifth key section present results from the one-way ANOVA analysis as well as from the logistic regression respectively. This is followed by a summary of the chapter. 7.2 SAMPLE CHARACTERISTICS The sampled respondents of the study were profiled according to nationality, gender, age, educational level and employment status. The characteristics of the respondents also addressed issues regarding their internet usage. This was done to summarize and provide a cursory view of the profile of the sample or study respondents. In Table 7.1 results of the 179 University of Ghana http://ugspace.ug.edu.gh demographic characteristics of the respondents are presented. The sample comprised 96.5% Ghanaians and 3.5% non-Ghanaians who were 325 (54.9%) female and 267 (45.1%) male. Majority of the respondents fell within the age bracket of 18 and 40 years constituting approximately 93% of the sample. Expressly, respondent within ages between 18 and 30 composed 53.7%, and those between 31 and 40 years constituted 39.2%. Only 6.1% and 1.0% of the respondents were between 41 and 50 years and above 50 years respectively. Table 7.1 Demographic Profile of Respondents Sample Composition Measures Frequency Percent Age 18-30 years 318 53.7 31-40 years 232 39.2 41-50 years 36 6.1 Above 50 years 6 1.0 Gender Male 267 45.1 Female 325 54.9 Educational Level Secondary/Vocational/Technical 13 2.2 HND/Diploma/Undergraduate Degree 265 44.8 Postgraduate Degree 277 46.8 Professional Degree 37 6.3 Employment Status Unemployed 64 10.8 Self-employed 86 14.5 Salaried worker (Full-time) 277 46.8 Salaried worker (Part-time) 45 7.6 College/Tertiary Student 120 20.3 Nationality Ghanaian 571 96.5 Other 21 3.5 N=592 Concerning the educational qualification of the respondents, 2.2% have second-cycle training (secondary, vocational and technical), and the remaining respondents have tertiary education comprising HND/Diploma/Undergraduate degree (44.8%) postgraduate degree (46.8%), and professional degree (6.3%). This provides an indication of the respondent’s ability to comprehend the key issues being addressed in the study as well as provide accurate 180 University of Ghana http://ugspace.ug.edu.gh responses to questions. Most (approximately 70%) of the respondents were employed, while 20.3% were college/tertiary students and 10/8% were unemployed. Of the working respondents 54.4% were salaried workers (both full-time and part-time), and 14.5% were self-employed. Table 7.2 Internet Usage Profile of Respondents Measures Frequency Percent Frequency of Internet Usage Several times a week 35 5.9 Every or almost every day 187 31.6 Several times a day 370 62.5 Internet Usage Motive I go online to make purchases and buy things - Shopping 123 20.8 I use the internet to search for any kind of information – (Re)searching 208 35.1 When online, I’m mostly browsing for fun and exploring new sites - Surfing 102 17.2 I go online to chat with friends, send and check emails - Communication 159 26.9 Familiarity with Online Advertising Somewhat familiar 21 3.5 Familiar 126 21.3 Very familiar 445 75.2 Frequency of Exposure to ODA Rarely 6 1.0 Sometimes 54 9.1 Always 532 89.9 How long ago you viewed and/or made a purchase following an online display ad Below 1 month 257 43.4 1-2 months 95 16.0 3-4 months 65 11.0 5-6 months 30 5.1 Above 6 months 145 24.5 Nature of the website on which you saw the ad Informational website (e.g. blogs, news site etc.) 70 11.8 Social media site (e.g. Facebook, Twitter, Instagram, YouTube etc.) 366 61.8 Commercial website (e.g. Tonaton, Jumia, OLX, Kaymu Amazon etc.) 111 18.8 Search engine (e.g. Google Adwords) 45 7.6 Nature of the Advertised Brand Product 455 76.9 Service 137 23.1 N=592 181 University of Ghana http://ugspace.ug.edu.gh As mentioned earlier, the respondents were also profiled on the basis of their internet usage. Results of this as depicted in Table 7.2 reveal that approximately 94% of the respondents use the internet daily and several times in a day. Of all these respondents, 35.1% use the Internet mainly for (re)search purpose, 20.8% stated shopping as their major reason for using the Internet, those who use the Internet for surfing were 17.2%, and 26.9% go online with the key aim to communicate. 75.2% of the respondent also show high familiarity with online advertising, and approximately 90% indicated that they are always exposed to ODA while 9.0% and 1.0 % stated that they sometimes and rarely get exposed to ODA respectively. This is further confirmed by responses regarding how long-ago respondents viewed and/or made a purchase following an online display ad as majority (43.4%) of the responses were below a month. Also, most of the online display ads respondents are exposed to are advertised products (76.9%) relative to services (23.1%) and are mostly seen on social media sites (61.8%), commercial websites (18.8%), informational sites (11.8%) and a few on search engines (7.6%). 7.3 DESCRIPTIVE STATISTICS Scholars (e.g. Pallant, 2011; Hair et al., 2010) “in social science research, speak to the usefulness of first conducting descriptive analysis on research data so as to control the violation of assumptions necessary for major statistical tests. Descriptive statistics are numerical and graphical methods used to summarise data. Pallant (2011) and Zikmund et al. (2009) point to measures of central tendency such as mean, median and mode, and measures of dispersions such as standard deviation, skewness and kurtosis as relevant numerical descriptive statistics used to summarise research data. Table 7.3 and Table 7.4 presents the descriptive statistics of the various measurement/scale items of the study constructs from the survey instrument that are used for data analysis. 182 University of Ghana http://ugspace.ug.edu.gh Table 7.3 Descriptive Statistics – ODA Characteristics Item S.E Std. Scale Items Code Mean Mean Dev. The online display ad I saw … had interactive features IT1 4.94 0.06 1.49 had links I could click for further information IT2 5.54 0.06 1.37 provided opportunity for me to give my feedback IT3 4.65 0.07 1.70 gave instantaneous information when I requested IT4 4.46 0.07 1.64 made me feel the advertiser wants to listen to customers IT5 4.31 0.07 1.65 allowed me a lot of control over my viewing experience IT6 4.46 0.07 1.66 allowed me to choose the timing of the ad IT7 3.20 0.08 1.97 provided two-way communication IT8 4.08 0.07 1.78 was a good fit/match for the webpage/site on which it appeared PL1 4.30 0.07 1.64 was consistent with the webpage/site on which it was featured PL2 4.72 0.06 1.55 and the webpage/site on which it appeared belong together PL3 4.08 0.07 1.66 had a lot in common with the webpage/site on which I saw it PL4 4.20 0.07 1.68 matched the content of the webpage/website PL5 4.20 0.07 1.68 was made more credible by the webpage on which it was featured PL6 4.68 0.06 1.51 was similar to other ads on the webpage/site PL7 4.62 0.06 1.42 was a good source of product/service information INF1 5.07 0.06 1.41 was a convenient source of product/service information INF2 5.06 0.06 1.45 supplied relevant information which was of value to me INF3 4.83 0.06 1.53 had information worth paying attention to INF4 4.92 0.06 1.51 gave me new ideas about the product/service INF5 5.05 0.06 1.49 helped keep me up to date with the product/service category INF6 4.98 0.06 1.52 made product information readily accessible INF7 5.16 0.06 1.44 supplied complete product/service information INF8 4.77 0.06 1.48 was tailored to my shopping situation at the time PS1 4.42 0.07 1.61 made recommendations that matched my needs at the time PS2 4.56 0.07 1.67 made me feel unique as an internet user PS3 4.19 0.07 1.76 was related to my search history at the time PS4 4.38 0.07 1.80 was useful and meaningful to me PS5 4.89 0.06 1.58 provided information based on my real-time location PS6 4.53 0.07 1.74 used my personal information (e.g. name & gender) PS7 4.23 0.08 1.83 permitted me to choose freely what I wanted to see EXC1 4.55 0.08 1.85 did not interfere with my online activity at the time EXC2 4.05 0.08 1.96 did not intrude on the content I was accessing EXC3 4.02 0.08 2.02 was not forced upon me EXC4 4.47 0.08 2.07 was not repeated while I was on the webpage EXC5 3.96 0.08 1.97 was not shown more than once during my activities on the webpage EXC6 3.74 0.08 1.94 did not stay on the screen for long EXC7 3.88 0.07 1.79 was not on the screen for up to 30 seconds EXC8 3.75 0.08 1.90 183 University of Ghana http://ugspace.ug.edu.gh The tables capture specifically the mean and standard deviation values which indicate the extent of respondents’ perception regarding the ODA characteristics (see Table 7.3) as well as evaluations of their attitude toward online advertising and behavioural responses (see Table 7.4). Generaly the mean values depict the extent of respondents’ general agreement or disagreement with the statements in the survey instrument, and the standard deviation values show how spread out responses are from the mean value.” The mean scores in both Table 7.3 and Table 7.4 represent how each of the 55 statements or items performed from the perspective of the respondents depicting mean scores ranging from 2.83 to 5.54 given the seven-point scale used. The statement “had links I could click for further information” recorded the highest mean score (5.54), followed by the item “I consider online advertising very essential” (5.46); and the stament “I bookmark online ads when I’m using the internet” had the lowest mean score of 2.83. The scale on which these items were measured had a midpoint of 4, and most of the mean values are above the midpoint which shows a good fit to the data set. The 38 items captured in Table 7.3 represent the components of the five ODA characteristics - interactivity, placement, informativeness, personalisation and exposure condition; and the 17 items shown in Table 7.4 are measures of attitude toward online advertising (ATOA) and behavioural responses (Ad avoidance and Ad acceptance) as visualised in the study framework 184 University of Ghana http://ugspace.ug.edu.gh Table 7.4 Descriptive Statistics – Attitude and Behavioural Responses Item S.E Std. Scale Items Code Mean Mean Dev. My general opinion of online advertising is favourable ATOA1 5.15 0.06 1.38 I consider online advertising very essential ATOA2 5.46 0.05 1.30 Online advertising is interesting and fun to see ATOA3 4.88 0.06 1.56 I appreciate seeing advertising messages on the Internet ATOA4 4.74 0.06 1.58 Overall, I like online advertising ATOA5 4.87 0.07 1.61 I intentionally ignore online ads when using the internet PADV1 4.64 0.07 1.82 I look away from online ads when I’m using the internet PADV2 4.30 0.07 1.69 I wait for online ads to go away, then I continue with what I’m doing PADV3 4.53 0.08 1.96 I scroll away from or leave webpages displaying online ads AADV1 4.68 0.07 1.71 I skip/close online ads that appear on my screen while I’m online AADV2 5.14 0.07 1.63 I use ad blockers on my devices AADV3 3.47 0.08 1.99 I pay attention to online advertisements PADC1 4.15 0.07 1.63 I carefully read the content of online advertisements PADC2 4.11 0.07 1.75 I watch online/read online advertisements to the end PADC3 3.41 0.07 1.77 I click on online advertisements or links provided in online ads AADC1 3.69 0.07 1.73 I bookmark online ads when I’m using the internet AADC2 2.83 0.07 1.76 I sign up or give feedback if the ad provides the option AADC3 3.08 0.07 1.82 7.4 EXPLORATORY FACTOR ANALYSIS (EFA) Following the descriptive analysis, an exploratory factor analysis was conducted on the 55 scale items with 275 cases from the data set using the maximum likelihood extraction method in SPSS version 23. Before the extraction, the data set was assessed to ensure its appropriateness for factor analysis. First, the correlation matrix generated, showed several satisfactory coefficients of 0.3 and above between the scale items; second, the Kaiser- Meyer-Oklin (KMO) measure of sampling adequacy recorded a value of .907 which exceeds the recommended threshold of 0.6 (Kaiser, 1970; Pallant, 2011); and third, the Bartlett’s Test of Sphericity produced an approximate Chi-square value of 111.744 which was significant (df. 1485, sig.000). The results of these three tests as displayed below in Table 7.5 established the appropriateness of the data set for factor analysis. 185 University of Ghana http://ugspace.ug.edu.gh Table 7.5 KMO and Bartlett's Test Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .907 Bartlett's Test of Sphericity Approx. Chi-Square 11132.744 df 1485 Sig. .000 7.4.1 Extraction, Rotation, Reliablity and Re-specification of the EFA After the suitability assessment, the factor extraction was conducted using the eigen value rule which unearthed factors/components with eigen values from 1.0 and above (Pallant, 2011). The extratction produced 8 factors, and these factors explained a total variance of 68.10%. To support and assist the interpretation of these 8 factors, the 55 items measuring them were rotated using an Oblique method, specifically Promax rotation in order to determine which items loaded considerably onto the various factors as well as those with strong loadings. To be retained for further analysis item loadings should be 0.7 and more although some scholars suggest a minimum value of 0.5 as acceptable (Hair et al., 2014). On the basis of this rule, 20 scale items which did not meet the recommended minimum 0.5 value were dropped from further analysis. Following the factor extraction and rotation, an assessment of the reliabilities of the remaining 35 scale items measuring the 8 factors was conducted as suggested by Pallant (2011) in order to ensure that the items are internally consistent or cohesive. The most widely used gauge for internal consistency, Cronbach’s alpha was employed to assess the reliability of the scales in the study. Scholars (e.g. Malhotra, 2010) suggest Cronbach’s alpha coefficients between 0.7 and 0.9 as ideal, and the assessment produced alpha values ranginging from 0.786 to 0.938 for all 8 factors. Also, as part of the reliability test, an inspection of the item-total correlation (ITC) was done to ensure the significance of the items that loaded onto the factors and all items had acceptable values of above 0.3 except 186 University of Ghana http://ugspace.ug.edu.gh two items; “was not repeated while I was on the page” and “was not shown more than ones during my activities on the page”. These items had ITCs of 0.207 and 0.284 respectively and so, were dropped from further analysis. Table 7.6 Rotated Component Matrix and Scale Reliability Maximum Likelihood Loadings Internal Consistencies Promax Variance Item-total Cronbach’s Alpha if items is (Oblique) Explained Correlation Alpha (a) deleted Factors Items Factor 1 IT3 .624 70.06 .611 .786 .725 IT4 .602 .613 .723 IT5 .776 .652 .681 Factor 2 PL4 .947 90.24 .805 .892 -- PL5 .955 .805 -- Factor 3 INF1 .750 72.95 .786 .938 .929 INF2 .879 .805 .928 INF3 .792 .824 .926 INF4 .688 .790 .929 INF5 .762 .789 .929 INF6 .898 .787 .929 INF7 .941 .799 .928 Factor 4 PS1 .789 65.31 .719 .866 .831 PS2 .606 .719 .830 PS3 .621 .704 .833 PS4 .737 .665 .844 PS6 .614 .637 .850 Factor 5 EXC1 .924 75.77 .549 .838 .911 EXC2 .886 .808 .667 EXC3 .672 .766 .710 Factor 6 ATOA1 .757 72.93 .752 .906 .888 ATOA2 .806 .743 .890 ATOA3 .718 .745 .889 ATOA4 .852 .793 .878 ATOA5 .726 .799 .877 Factor 7 PADV1 .669 62.13 .596 .796 .752 PADV2 .772 .604 .746 AADV1 .724 .605 .746 AADV2 .701 .625 .737 Factor 8 PADC3 .805 70.69 .692 .861 .830 AADC1 .825 .739 .811 AADC2 .703 672 .838 AADC3 .736 .730 .814 Note: Extraction Method: Maximum Likelihood. Rotation Method: Promax with Kaiser Normalisation 187 University of Ghana http://ugspace.ug.edu.gh The table 7.6 above summarises the factor loadings and the reliability (internal consistency) measures on the final variables retained for the 8 constructs or factors. The exploration reduced or purified an initial pool of 55 items to 33 as the 8 items measuring interactivity were reduced to 3; items measuring placement were reduced from 7 to 2; those measuring informativeness were reduced from 8 to 7; scale items for personalisation were reduced from 7 to 5; and exposure condition items were also reduced from 8 to 3. The 5 scale items measuring attitude toward online advertising all passed the exploration and reliability tests and were retained for further analysis. It is however important to mention that scale items for passive ad avoidance and active ad avoidance were each reduced from 3 to 2; and passive ad acceptance was reduced from 3 to 1, while active ad acceptance maintained all three items. More relevantly, the extraction and rotation process merged the active and passive dimension of both ad acceptance and ad avoidance respectively, and thus, were maintained as such (i.e. unidimensional) on the basis of conceptual fitness. 7.4.2 Examination of Common Method Variance/ Bias As the study data is based on self-reports, and as can be gleaned from Table 7.6 above, some factors (e.g. factor 2) accounted for higher levels of variance, thus necessitating the need to check and eliminate the presence of common method variance. Common method variance (CMV) is the false correlation between variables that result when a researcher employs a similar method to measure the predictor and outcome variables in a hypothesised relationship (Craighead et al., 2011). This causes a systematic measurement error that considerably exaggerates or reduces the measured relationships among theoretical constructs (Siemsen et al., 2010). Although the magnitude of CMV is said to vary across fields, it is considered “the most common and dangerous threat to correct interpretation of research results” (Pace, 2010, p.421). This calls for researchers to address issues of CMV whenever possible because its presence 188 University of Ghana http://ugspace.ug.edu.gh casts doubt on the adequacy of a scale’s reliability and convergent validity particularly in survey research (Richardson et al., 2010; Podsakoff et al., 2012). This study employed a post-hoc statitistical test specifically, Harman’s single-factor test which is the dominantly used approach in controlling for CMV in social science research (Tang et al., 2010). The key supposition of this test is that, during an EFA, if a single factor accounts for a majority of the variance in the variables or if a single factor is extracted from an unrotated factor solution, then CMV is purpoted to be present (Graighead et al., 2011). In conducting the EFA, none of these conditions surfaced. It has also been suggested that the test unearths bias if the researcher extracts only one factor and that factor accounts for more than 50% variance (Fuller et al., 2016). To test this assumption, during the EFA, one factor extraction was forced from the initial 55 scale items using unrotated maximum likelihood extraction. In that instance, the single factor solution accounted for 35.16% of variance, thus allaying concerns of CMV in this study, and instilling confidence in the outcome or findings of the research. 7.5 STRUCTURAL EQUATION MODELLING (SEM) In the methodological chapter, the two stage SEM approach was discussed as the analytical method used to test the hypothesised relationships in the research framework following suggestions by Anderson and Gerbig (1988). “In the first stage, which is the measurement phase, the causal relationships between the observed variables or measurement items and the underlying theoretical constructs were specified by conducting a confirmation factor analysis (CFA) using AMOS version 22. Through the CFA, a measurement model was generated. The second phase – structural phase subsequently assesses the causal relationships between the predictor and response variables as well as the the mediating and moderating effects using a structural model. Detailed discusions on the procedures followed 189 University of Ghana http://ugspace.ug.edu.gh in conducting this two-stage analysis and their results are presented in the next sub-sections (7.5.1 -7.5.2).” 7.5.1 Confirmatory Factor Analysis (CFA) / Measurement Model After reducing the measurement items/scales into a manageable set through the EFA, as a vital first step to SEM, a confirmatory factor analysis (CFA) was conducted using the larger data set (592 cases) to determine whether all the measures are accurate indicators of, or truly reflect the latent variables (i.e. Interactivity, Placement, Informativeness, Personalisation, Exposure condition, Attitude toward Online Advertinsing, Ad Avoidance, and Ad Acceptance) in the research framework. The essence of this is to ensure further purification, validation and modification (if needs be) of the measurement model before it is used to address the structural hypotheses (Byrne, 2010). The 8 resultant factors and their 33 corresponding measurement items from the EFA (see Table 7.6) were used for the CFA, and the initial output from AMOS indicated that the standardised loadings for some of the indicators were below the recommended threshold causing the fit indexes to be lower than the acceptable values. Adhering to scholarly suggestions (e.g. Hair et al., 2010; Koo et al., 2015), the initial measurement model was subjected to modification using the factor loadings, and correlation between measurement errors as adjustment criteria. During the modification process, a total of 6 scale items were dropped or removed in a step-by-step fashion so as to ensure that each item deletion or elimination was crucial to achieving a suitable model fit. Specifically, in the first modification phase (modified model I), one item each was deleted from Exposure Conditions (EXC1), and Ad Avoidance (ADV2), and two items from Personalisation (PS4 & PS6). In the second modification phase, two items were deleted – one from Attitude toward Online Advertising (ATOA3) and one from Ad Acceptance (ADC2). Additionally, the residual erros with high variances were eliminated or covaried. After the eliminations, the model was re- analysed resulting in an improvement of the goodness-of-fit indices which exceeded or 190 University of Ghana http://ugspace.ug.edu.gh approached the recommended standards (Bentler, 1990). For instance, the RMSEA was reported at 0.06, which is well beneath the suggested 0.08 cutoff point. The final measurement model with the acceptable fit indices was made up of 8 constructs with 27 corresponding indicators (measurement items). A summary of the modification process and the fit indices of the model are presented below in Table 7.7, and Figure 7.1 depicts a graphical view of the final measurement model. Table 7.7 Modification and Model Fit Summary of Measurement Model Modifications of Fit Indices Fit Indices Cut-off Original Modified Final Modified Criteria Model Model I Model GFI (Goodness of Fit Index) ³0.90 0.81 0.89 0.91 CFI (Comparative Fit Index) ³0.90 0.86 0.93 0.95 NFI (Normed Fit Index) ³0.90 0.83 0.91 0.92 TLI (Tucker-Lewis Index) ³0.90 0.84 0.92 0.93 IFI (Incremental Fit Index) ³0.90 0.86 0.93 0.95 RMSEA (Root Mean Square <0.08 0.08 0.07 0.06 Error Approximation χ2/df (Normed Chi-square) £ 2 or 3 5.23 3.47 3.12 Note: Model 1 (deleted: EXC1, ADV2, PS4 & PS6) Final Model (deleted ADC2 & ATOA3) 7.5.1.1 Reliability and Validity of the Final Measurement Model After the acceptable goodness-of-fit indices were achieved, the final measurement model as pictured in Figure 7.1 below, was further assessed for reliability and validity using Cronbach’s alpha coefficient, Composite Reliability (CR), the outer or factor loadings, and Average Vairance Extracted (AVE). First, because six items were deleted from the post-EFA pool during the CFA, alpha (a) values were again generated for the constructs in the model, and these values exceeded the satistactory limit of 0.7 – an indication of internal consistency. Also, the CR values which depict the degree to which the indicators are reflective of the latent constructs surpassed the acceptable minimum of 0.7 (Nunnally, 1978), with which the collective reliability of the final measurement model was confirmed. 191 University of Ghana http://ugspace.ug.edu.gh Figure 7.1 Final Measurement Model 192 University of Ghana http://ugspace.ug.edu.gh Table 7.8 CFA Results for Final Measurement Model Constructs Items Standardised t-Values R2 Value CR AVE Cronbach’s Loadings Alpha (a) Interactivity IT3 0.72*** 15.39 0.51 0.79 0.55 .77 IT4 0.73*** 15.57 0.53 IT5 0.78*** -- 0.61 Placement PL4 0.88*** 21.63 0.78 0.89 0.81 .89 PL5 0.91*** -- 0.83 Informativeness INF1 0.83*** 24.74 0.68 0.93 0.67 .94 INF2 0.82*** 24.60 0.67 INF3 0.88*** -- 0.77 INF4 0.85*** 28.50 0.73 INF5 0.81*** 20.99 0.66 INF6 0.77*** 22.19 0.59 INF7 0.78*** 22.75 0.60 Personalisation PS1 0.78*** 17.74 0.61 0.85 0.65 .85 PS2 0.85*** 21.53 0.72 PS3 0.79*** -- 0.62 Exposure EXC2 22.88 0.83 .91 Condition 0.91*** 0.91 0.84 EXC3 0.92*** -- 0.85 Attitude Toward 21.70 0.56 .89 ATOA1 0.77*** 0.90 0.65 OA ATOA2 0.75*** -- 0.57 ATOA4 0.84*** 20.32 0.70 ATOA5 0.88*** 21.31 0.78 Ad Avoidance ADV1 0.54*** 11.47 0.22 0.77 0.53 .75 ADV2 0.70*** 13.63 0.49 ADV3 0.91*** -- 0.82 Ad Acceptance ADC1 0.78*** 17.85 0.61 0.84 0.65 .84 ADC2 0.89*** 19.30 0.80 ADC3 0.72*** -- 0.53 Note: ***p<0.001 Fixed Values (--) Furthermore, convergent and discriminant validity were assessed to establish construct validity. In ascertaining the convergent validity of the measures in the model, the factor loadings for each item and the AVEs were examined. To be considered valid, Hair et al. (2010) recommend that factor loadings should be higher than 0.5. As can be gathered from Table 7.8 above, the factor loadings for all items are considerably high and within the range of 0.54 and 0.92; and were statistically significant at p<0.001. Also, the AVE values ranged from 0.53 to 0.84 which are all above the minimum required level of 0.50 suggested by Fornell and Lacker (1981). The AVE values show the amount of variance the items share 193 University of Ghana http://ugspace.ug.edu.gh with the construct they measure and so, describe the share of the variance that is explained by the indicators in relation to a particular construct. These two measures speak to the convergent validity of the constructs and their individual measurement items in the model. The essence of the discriminant validity assessment is to ensure that each construct is unique and captures dissimilar but theoretically linked variables (Duarte & Raposo, 2010). This is done by correlating pairs of latent variables or constructs. The rule of thumb is that, to determine the presence of discriminant validity, for every pair of construct, the square-root of the AVE (correlation between a construct and itself) must be higher than the corresponding inter-construct correlation (correlation between the construct and other constructs) (Lu et al., 2010). As can be surmised from Table 7.9, the square-root value of the AVEs are soundly higher than the correlations between each pair of constructs, thereby establishing the discriminant validity of the final measurement model. Table 7.9 Correlation Matrix for Discriminant Validity Constructs Mean Std. Dev In Pl Inf Ps Ex At Adv Adc Interactivity 4.478 1.663 0.743 Placement 4.199 1.679 0.401 0.897 Informativeness 5.009 1.477 0.532 0.496 0.819 Personalisation 4.390 1.682 0.570 0.613 0.767 0.806 Exposure Condition 4.037 1.991 0.323 0.483 0.500 0.551 0.915 Attitude Toward OA 5.055 1.468 0.486 0.376 0.626 0.603 0.445 0.807 Ad Avoidance 4.823 1.718 -0.095 -0.041 -0.160 -0.205 -0.148 -0.339 0.730 Ad Acceptance 3.394 1.774 0.402 0.260 0.458 0.485 0.275 0.656 -0.410 0.803 Note: Diagonal (bold) values are square roots of the AVE; off-diagonal values are the inter-construct correlations 7.5.2 Structural Model The next logical step after the measurement model is specified, confirmed to be reliable, and validated with acceptable fit, is to test the structural model. Considered the key stage of SEM 194 University of Ghana http://ugspace.ug.edu.gh analysis, this phase allowed for the estimation of the structural paths in order to test the hypothetical suppositions indicated in the research framework. Particularly, structural models help test multiple relationships concurrently in complex models while controlling for measurement errors (Byrne, 2013). Keeping on course with the aim of the study which is to examine the effect of ODA characteristics on consumer behavioural responses to online display advertising as mediated by attitude toward online advertising, as well as account for the moderating influence of internet user mode, the first sets of hypotheses suggest a direct influence of the ODA characteristics (interactivity, placement, informativeness, personalisation and exposure condition) on attitude towad online advertising (H1-H5), as well as on ad acceptance and ad avoidance (H6a,b – H10a,b). These direct proposed influences are first tested by creating single indicators for the latent variables as suggested by Ping (1995) to reduce the complexity of the model, and the results of their structural paths generated from AMOS are presented in figure 7.2 below. Figure 7.2 Structural Model Results for Direct Paths 7.5.2.1 Assessment of the Direct Structural Model By examining the structural model, the study establishes whether the data is consistent with the conceptualisation and resultant propositions put forth. To ensure this, foremost, the values of the relevant fit statistics were checked and found to meet the recommended parameters: 195 University of Ghana http://ugspace.ug.edu.gh normed Chi-square (χ2/df) = 2.71, Comparative Fit Index (CFI) = 0.98, Incremental Fit Index (IFI) = 0.99, Tucker Lewis Index (TLI) = 0.930, and Root Mean Square Error of Approximation (RMSEA) = 0.054. Second, the signs (negative or positive) of the various estimates (b-values) were checked to be sure the directions of the paths were as hypothesised. Third, the strength of the hypothesised paths was examined to guarantee they were significant. A vital condition for path significance is that the absolute figures of the critical ratio (t-values) should be higher than 1.96. Lastly the amount of variance in the endogenous variables explained by the exogenous variables depicted by the R2 values were checked for suitability. These statistics of interest are reported in Table 7.10 below. Table 7.10 Structural Model Results - Direct Paths Hypothesised Direct Structural Paths Estimates Result b t Sig. H1 Interactivity -----> Attitude toward Online Advertising 0.15 3.96 *** Supported H2 Placement -----> Attitude toward Online Advertising -0.01 -0.19 0.85 Unsupported H3 Informativeness -----> Attitude toward Online Advertising 0.31 6.60 *** Supported H4 Personalisation -----> Attitude toward Online Advertising 0.18 3.74 *** Supported H5 Exposure Condition -----> Attitude toward Online Advertising 0.13 3.35 *** Supported H6a Interactivity -----> Ad Acceptance 0.18 4.27 *** Supported H7a Placement -----> Ad Acceptance -0.04 -0.91 0.36 Unsupported H8a Informativeness -----> Ad Acceptance 0.18 3.43 *** Supported H9a Personalisation -----> Ad Acceptance 0.23 4.14 *** Supported H10a Exposure Condition -----> Ad Acceptance 0.04 1.02 0.31 Unsupported H6b Interactivity -----> Ad Avoidance -0.01 -0.12 0.91 Unsupported H7b Placement -----> Ad Avoidance 0.06 1.25 0.21 Unsupported H8b Informativeness -----> Ad Avoidance -0.06 -0.97 0.33 Unsupported H9b Personalisation -----> Ad Avoidance -0.13 -2.15 0.03* Supported H10b Exposure Condition -----> Ad Avoidance -0.07 -1.35 0.18 Unsupported Age -----> Ad Acceptance 0.02 0.64 0.52 Age -----> Ad Avoidance -0.04 -1.07 0.29 Gender -----> Ad Acceptance -0.02 -0.64 0.53 Gender -----> Ad Avoidance 0.01 0.22 0.83 Internet Familiarity -----> Ad Acceptance 0.01 0.28 0.78 Internet Familiarity -----> Ad Avoidance 0.09 2.35 0.02* Usage Frequency -----> Ad Acceptance -0.02 -0.55 0.58 Usage Frequency -----> Ad Avoidance 0.02 0.36 0.72 χ2/df=2.71 RMSEA=0.054 GFI=0.98 NFI=0.99 IFI=0.99. TLI=0.930. CFI=0.98 Note: ***p<.001 **p<.01 *p<.05 Attitude: R2=0.37 Ad Acceptance: R2=0.24 Ad Avoidance: R2 =0.05 196 University of Ghana http://ugspace.ug.edu.gh As shown in Table 7.10 above, hypotheses 1 to 5 proposed a significant positive influence of the ODA characteristics on attitude toward online advertising (ATOA), and there was statistical support for all these paths except the path from placement to ATOA (b = -0.01, p = 0.85); as such hypothesis 2 is not supported. Of the four supported paths, the strongest relationship exists between informativeness and ATOA (b =0.31, p < 0.001), followed by personalisation (b =0.18, p < 0.001), interactivity (b =0.15, p < 0.001), and exposure condition (b =0.13, p < 0.001). Because the study operationalised behavioural responses as ad avoidance and ad acceptance, hypotheses 6 to 10 had two sub-hypotheses each. Hypotheses 6a- 10a predicted a positive significant effect of the ODA characteristics on ad acceptance. There was support for H6a, H8a and H9a which represent the paths from interactivity, informativeness and personalisation respectively. However, results failed to support hypotheses H7a and H10a which showed non-significant results for the effects of placement (b=-0.04, p=0.36) and exposure condition (b=0.04, p=0.31) on ad acceptance. Of the supported ODA charateristics-ad acceptatnce relationships, the strongest direct predictor of ad acceptance was personalisation (b =0.23, p < 0.001), followed by interactivity (b =0.18, p < 0.001), and informativeness (b =0.18, p < 0.001). Additionally, hypotheses 6b to 10b posited significant negative influences of the ODA characteristics on ad advoidance. Nonetheless, results only supported Hypotheses 9b (b =-0.13, p<0.05) which addresses the effect of personalisation. Generally, H1, H3, H4, H5, H6a, H8a, H9a and H9b were confirmed while H2, H6b, H7a- b, H8b, and H10a-b were rejected in this study. In order to account for the other effects not hypothesised, the model also controlled for demographics such as age and gender, as well as frequency of internet usage and familiarity 197 University of Ghana http://ugspace.ug.edu.gh with online advertising (see Table 7.10). Regarding these control variables, age, gender, and internet usage frequency had no significant effect on consumer behavioural resoponses (ad acceptance or ad avoidance) in the model. Also, familiarity with online advertising had no effect on ad acceptance (b = 0.01, p = 0.78), but was found to be positively related to ad avoidance (b = 0.09, p =0.02). A chi-square difference test was then conducted to examinie whether user familiarity accounts for any differences in the model. Results from the test revealed that the different levels of user familiarity with online advertising did not explain any differences in the model (Δχ2 = 0.01, Δdf =1, p =0.95). In essence within the borders of this thesis, the relationships specified in the model are not confounded by the control variables. 7.5.2.2 Test for Mediating Effects The second objective of this study sought to establish the mediating role of attitude toward online advertising (ATOA) in the relationships between the ODA characteristics and behavioural responses (ad acceptance and ad avoidance). To this end, the study tested ten mediation paths from each of the five ODA characteristics to ad acceptance and ad avoidance through ATOA. Hypothesis 12ai - 12ei tested the path from interactivity, placement, informativeness, personalisation and exposure condition respectively to ad acceptance through ATOA; and H12aii - H12eii tested the path from interactivity, placement informativeness, personalisation, and exposure condition respectively to ad avoidance through ATOA. The goal is to determine if ATOA enhances the effect of the ODA characteristics on consumer behavioural responses to online display ads than the direct effects of their perceptions of the ODA characterisitics. “The bootstrapping approach to testing mediation effects was employed in the study using AMOS 22. Contrary to the prevalent causal steps approach espoused by Baron and Kenny 198 University of Ghana http://ugspace.ug.edu.gh (1986), bootstrapping assumes that a non-significant relationship between independent and dependent variables does not provide grounds on which to write off the presence of the intervening influence of a third variable (Preacher & Hayes, 2004; Hayes, 2015). In other words, the presence of a significant relationship between an independent and a dependent variable also provides no guarantee that a mediating effect may exist through an intervening variable (Hayes, 2009). In view of this arguments, the study used bootstrapping to address the mediation effect of ATOA in the relationships between all five ODA characteristics and behavioural responses in spite of the lack of a direct relationship between interactivity, placement, informativeness and exposure condition and ad avoidance as well as between placement, exposure condition and ad acceptance. To determine the significance of the mediated paths, the analysis relied on a bootstrapped sample of 2000 at a bias-corrected accelerated confidence interval level of 95%. Figure 6.3 below presents the structural paths of the mediation analysis. Figure 7.3 Structural Model Results for Mediated Paths Per the bootstrapping approach, partial mediation is established or confirmed when a significant direct path from an independent variable to a dependent variable occurs concurrently with a significant indirect path from said independent variable to the dependent 199 University of Ghana http://ugspace.ug.edu.gh variable through an intervening variable (Preacher & Hayes, 2008). On the other hand, full mediation is confirmed when a non-significant direct path from an independent variable to a dependent variable exists alongside a significant indirect path from the independent variable to the dependent variable through the intervening variable.” Illustratively, the results of the mediating test as shown in Table 7.11 below on the basis of the bootstrapping assumptions, indicate that no support was found for hypotheses 12bi and 12bii which represent the paths from placement to ad acceptance and ad avoidance respectively, as mediated by attitude toward online advertising. This is because, the study found no direct effect of placement on neither ad acceptance nor ad avoidance, and the indirect effect through ATOA was also not statistically significanct. Table 7.11 Structural Model Results - Mediated Paths Path Estimates Direct Direct Indirect Hypothesised Mediated Structual Paths without with Result Effect Mediator Mediator H12ai Interactivity -----> ATOA -----> Ad Acceptance 0.18*** 0.12*** 0.06*** Partial Mediation H12bi Placement -----> ATOA -----> Ad Acceptance -0.04 -0.04 -0.00 No Mediation H12ci Informativeness -----> ATOA -----> Ad Acceptance 0.18*** 0.05 0.13*** Full Mediation H12di Personalisation -----> ATOA -----> Ad Acceptance 0.23*** 0.15** 0.08*** Partial Mediation H12ei Exposure Condition -----> ATOA -----> Ad Acceptance 0.04 -0.01 0.05*** Full Mediation H12aii Interactivity -----> ATOA -----> Ad Avoidance -0.01 0.04 -0.05*** Full Mediation H12bii Placement -----> ATOA -----> Ad Avoidance 0.06 0.06 -0.00 No Mediation H12cii Informativeness -----> ATOA -----> Ad Avoidance -0.06 0.04 -0.10*** Full Mediation H12dii Personalisation -----> ATOA -----> Ad Avoidance -0.13* -0.07 -0.06*** Full Mediation H12eii Exposure Condition -----> ATOA -----> Ad Avoidance -0.07 -0.02 -0.04*** Full Mediation Note: ***p<0.001 **p<0.01 *p<0.05 Attitude: R2=0.37 Ad Acceptance: R2=0.35 Ad Avoidance: R2 =0.11 200 University of Ghana http://ugspace.ug.edu.gh In addtion, the results indicate that attitude toward online advertising (ATOA) partially mediates the relationship between interactivity and ad acceptance as well as the relationship between personalisation and ad acceptance with significant direct effects (interactivity; b = 0.12, p<0.001; personalisation; b = 0.15, p<0.01) and significant indirect effects (interactivity; b = 0.06, p<0.001; personalisation; b = 0.08, p<0.001) for both ODA characteristics. The paths from informativeness and exposure condition to ad acceptance show non- significant direct effects (informativeness; b = 0.05, p>0.05; exposure condition; b = -0.01, p>0.05) but significant indirect effects (informativeness; b = 0.12, p<0.001; exposure condition; b = 0.05, p<0.001) indicating a full mediating role of ATOA in these relationships. Pertinently, ATOA fully mediates the relationships between interactivity, informativeness, personalisation, exposure condition and ad avoidance as, the direct paths of these ODA characteristics to ad avoidance showed non-significant effects but significant indirect effects for all four paths at p<0.001 level. Results of the mediation analysis therefore provide support for eight of the ten Hypothesised mediated paths (i.e. H12ai, H12ci, H12di, H12ei; and H12aii, H12cii, H12dii, H12eii). 7.5.2.3 Test for Moderating Effects After support was found for some of the direct effects and mediating effects, the next step was to include the proposed moderator variable into the revised model (depicting significant paths) in order to gain further insights. Underpinned by the reversal theory, this phase examines the moderating effect of user mode on the supported relationships established between the ODA characterisitcs and attitude toward online advertising as well as behavioural responses. To test the hypotheses (H13ai, H13di, and H14a, H14c, H14d, H14e) addressing the resultant paths, a multigroup moderation analysis was conducted. This was 201 University of Ghana http://ugspace.ug.edu.gh done in two stages in AMOS. First, a structural invariance test was conducted to ensure differences in the hypothesised relationships across the subgroups (telic vs. paratelic users). To do this, the paths in the model for both groups were estimated concurrently allowing the estimates of the direct paths to vary accross the groups, and this represented the baseline model. Next the parameter estimates of the groups were constrained to be the same resulting in a fully constrained model. In both cases, relevant fit indices were used to assess each model to ensure they meet the required benchmark. A chi-square difference test was then conducted between the baseline and constrained model (Anderson & Gerbing, 1988), which showed significant differences (Δχ2 = 54.65, p<0.001) among the two user modes (telic versus paratelic) at the model level – an indication that the model is not structurally invariant. Results for the invariance test are shown in Table 7.12 below. Table 7.12 Invariance Test Results: Multigroup Analysis Overall Model χ2 df χ2/df NFI TLI GFI CFI RMSEA P-value Baseline Modela 42.36 20 2.118 0.97 0.92 0.99 0.99 0.04 Structural Weights Modelb 97.01 36 2.66 0.94 0.86 0.97 0.96 0.05 χ2 Differences Test (Δχ2, Δdf) 54.65 16 *** Note: aUnconstrained model bConstrained model ***p<0.001 After invariance is established, for more sound results, it is suggested that researchers further conduct a path-by-path analysis to ascertain the particular paths that differ significantly across the subgroups (Byrne, 2004; 2011). Hence as the second stage of the multigroup test, a pairwise critical ratio difference test was subsequently conducted to identify whether there were significant differences for the individual paths across groups. This was done using the baseline (unconstrained) model which displayed relatively more acceptable fit indices from the invariance test. Results of the moderation test as shown in Table 7.13 suggest that only two out of the six hypothesised paths in the revised model were supported. 202 University of Ghana http://ugspace.ug.edu.gh User mode was found to be a significant moderator of the relationship between informativeness and ATOA (hypothesis 14c: z-score=2.649; p<0.01), as well as between personalisation and ad acceptance (hypothesis 13di: z-score=1.924; p<0.10). To further probe the moderated paths, the standardized estimates from the results were examined, and this indicate that the effect of informativeness on ATOA is stronger for paratelic users (b=0.50; p<0.01) than for telic users (b=0.25; p<0.01). Also, personalisation had a significant positive effect on ad acceptance among telic users (b=0.21; p<0.01) but not among paratelic users b=-0.05; n.s.). However, there is no support for hypothesis 13ai which proposes that user mode moderates the relationship between interactivity and ad acceptance. The study also failed to find support for, H14a, H14d, H14e. For these unsupported hypotheses, the results show that the impact of interactivity, personalisation and exposure condition on attitude toward online advertising do not vary between telic and paratelic users. Table 7.13 Structural Model Results for Multigroup (User Mode) Moderation Hypothesised Moderated Paths Telic Paratelic z-score b Sig. b Sig. Interactivity -----> ATOA 0.14 0.00 0.12 0.04 -0.21 Informativeness -----> ATOA 0.25 0.00 0.50 0.00 2.65*** Personalisation -----> ATOA 0.20 0.00 0.08 0.20 n.s. -1.57 Exposure Condition -----> ATOA 0.10 0.00 0.03 0.44 n.s. -1.32 Interactivity ----->Ad Acceptance 0.11 0.01 0.18 0.04 0.67 Personalisation -----> Ad Acceptance 0.21 0.00 -0.05 0.52n.s. -2.83*** Note: ***p<0.01 n.s.= not significant 7.6 ONE-WAY ANALYSIS OF VARIANCE (ANOVA) Though not hypothesised, in order to gain further insights into other influences as well as examine the intensity of behavioural responses, the study addresses differences in behavioural responses on the basis of the nature of advertised brands featured in ODAs. This was encapsulated in the fourth objective of the study. Since the survey instrument 203 University of Ghana http://ugspace.ug.edu.gh elicited information on the nature of the advertised brand consumers purchased after exposure to a recent display ad, the study categorised responses into product-featured ODAs and service-featured ODAs. Pursuant to this objective, the study first performed an ANOVA test, and result as shown in Table 7.14 indicate significant differences in acceptance and avoidance responses among consumers exposed to ODAs that featured product brands and those exposed to ODAs that featured service brands. Precisely, avoidance was higher for service-featured ODAs relative to product-featured ODAs with mean values ranging from 4.96 to 5.54 for the former and from 4.55 to 5.01 for the latter. Additionally, acceptance was higher among consumers exposed to product-featured ODAs (3.20 – 3.80) compared to those exposed to service-fearured ODAs (2.70 - 3.30) Table 7.14 ANOVA Results Product- Service- F Sig. Behavioural Responses to ODA featured ODAs featured ODAs n=455 n=137 Ad acceptance I watch/read online advertisements to the end 3.57 2.90 15.80 .000 I click on online advertisements or links provided in 3.80 3.30 9.07 .003 online ads I sign up or give feedback if the ad provides the option 3.20 2.70 8.24 .004 Ad Avoidance I intentionally ignore online ads when using the internet 4.55 4.96 5.31 .022 I click/scroll away from or leave webpages displaying 4.55 5.12 12.04 .001 online ads I skip/close online ads that appear on my screen while 5.01 5.54 10.82 .001 I’m online 7.7 LOGISTIC REGRESSION To elucidate the ANOVA results further, the study assesses the intensity of the behavioural responses by categorising respondents into active and passive behavioural segment or levels. After the categorisation, which was done in SPSS, two sets of homogeneous groups were generated for both ad acceptance and ad avoidance. In examining the intensity of ODA-specific behavioural response among the study sample, a multigroup logistic 204 University of Ghana http://ugspace.ug.edu.gh regression was performed to assess the effect of the four ODA characteristics (Interactivity, Personalisation, Informativeness and Exposure Condition). that passed the major (SEM) analytical stage. Although not all four characteristics had direct influences on ad acceptance and ad avoidance in the structural model, the study employed logistic regression to assess the impact of their effect on the likelihood of respondents reporting active behavioural responses. Four models were generated, and the four ODA characteristics were the independent variables in all models. However, active-passive binary segments of ad acceptance were the dependent variable in model 1a and 1b; and active-passive segments of avoidance were the dependent variables in Model 2a and 2b. The nature of the brand (product versus service) featured in the display ads respondents viewed, were inputted as selection variables in order to determine ODA-specific behavioural response. All sorted models in the analysis were drawn from active behavioural response (active acceptance and active avoidance) categories since the analysis procedure treated the passive segments as dummies. Results from the regression output are presented in Table 7.15. The full unsorted model 1 (containing both product-featured and service-featured ODAs), was statistically significant, χ2 = 98.33, df = 4, p value < 0.001. The model explained between 15.3 percent (Cox & Snell R2) and 20.7 percent (Nagelkerke R2) of the variance in ad acceptance and an overall predictive accuracy of 69.1 percent of cases in the data - a sign that the model differentiated the two intensity levels of ad acceptance. The two sorted models; model 1a (χ2 = 95.54, df = 4, p value < 0.001) and model 1b (χ2 = 11.34, df = 4, p value < 0.05) were statistically significant., and both had predictive accuracies of 62 percent and 66.6 percent respectively. In Model 1a, three ODA characteristics except informativeness (Wald =0.05, p value = n.s.) made unique statistical contributions to the model. Controlling for all the other ODA characteristics, the odds ratio values reveal that consumers or internet users are 1.54 times more likely to actively accept personalised ODAs featuring product brands; 205 University of Ghana http://ugspace.ug.edu.gh 1.28 times more likely to actively accept interactive ODAs featuring product brands; and 1.15 times more likely to actively accept product-featured ODAs presented with unforced exposure. In model 1b, which was based on service-featured ODAs, the statistics show that only interactivity (Wald = 8.24, p value < 0.001) made statistically significant contribution to the model, and its corresponding odds ratio value indicates that internet users are 1.67 times more likely to actively accept interactive service-featured ODAs than non-interactive ones. Table 7.15 Logistic Regression with likelihood Ratio for Intensity of Behavioural Responses 95% C.I. for Odds Ratio Nature of Advertised ODA Odds Brand Characteristics B Wald Sig. Ratio Lower Upper Model 1a Product-featured Interactivity 0.25 7.33 0.01 1.28 1.07 1.54 ODA Informativeness 0.10 0.59 0.45n.s. 1.11 0.85 1.43 Personalisation 0.43 14.01 0.00 1.54 1.23 1.93 Exposure Condition 0.14 4.60 0.03 1.15 1.01 1.31 Model 1b Service-featured Interactivity 0.52 8.24 0.00 1 .67 1.18 2.38 ODA Informativeness -0.04 0.04 0.85 n.s. 0.96 0.62 1.47 Personalisation 0.13 0.56 0.45 n.s. 1.14 0.81 1.60 Exposure Condition -0.03 0.05 0.82 n.s. 0.97 0.78 1.22 Model 2a Product-featured Interactivity -0.06 0.35 0.55n.s. 0.94 0.93 2.55 ODA Informativeness -0.15 0.75 0.38 n.s. 0.87 0.63 3.06 Personalisation -0.35 6.24 0.01 0.70 0.66 1.76 Exposure Condition 0.08 1.08 0.30 n.s. 1.08 0.38 0.94 Model 2b Service-featured Interactivity 0.43 2.81 0.09 n.s. 1.54 0.76 1.16 ODA Informativeness 0.32 0.66 0.42 n.s. 1.39 0.62 1.20 Personalisation 0.08 0.09 0.77 n.s. 1.08 0.53 0.93 Exposure Condition -0.52 4.93 0.03 0.60 0.93 1.26 Dependent variables: Model 1abAd Acceptance Model 2abAd Avoidance n.s.= not significant The full unsorted model 2 (comprising both product-featured and service-featured ODAs) was also statistically significant, χ2 = 17.08, df = 4, p value = 0.001. The Hosmer and Lemeshow Test (H-L statistic) was not statistically significant (0.46), and the the model explained between 3.2% (Cox and Snell R2) and 5.2% (Nagelkerke R2) of the variance in 206 University of Ghana http://ugspace.ug.edu.gh ad avoidance, as well as correctly categorized 83.4% of cases which means the model is quite a good fit. As illustrated in Table 7.15, above in model 2a, only one ODA characteristic, personalisation made a statistically significant (B = -0.35, Wald = 6.24, p value = 0.01) contribution to the model reporting an odds ratio of 0.70. This shows that controlling for all other ODA characteristics, internets users are 0.70 less likely to actively avoid product-featured ODAs that are personalised than those that are not personalised. Model 2a also shows that, service-featured ODAs that are not presented with forced exposures, are 0.60 less likely to cause active avoidance among consumers or internet users with all other ODA characteristics controlled for; given that exposure condition was the only significant contributor (B = -0.52, Wald = 4.93, p value = 0.03) to the model. 7.8 CHAPTER SUMMARY This chapter presented the analysis of empirical data obtained from a sample of 592 internet users in order to test the postulated hypotheses geared toward achieving the research objectives. The methods, approaches and procedures spelt out in the methodological chapter were employed in analysing the research data. The demographic characteristics of the respondents and their internet usage information were presented in this chapter; this was followed by preliminary analysis particularly, descriptive statistics and exploratory factor analysis which helped reduced the scale items into a manageable set. The factors that emerged from the EFA and their corresponding measures were used in conducting a confirmatory factor analysis. At the CFA stage, the measurement model was specified, and its validity and reliability were assessed to ensure the required benchmarks were met, producing an acceptable and reliable model for the structural model analysis. The direct relationships as well as the mediation and moderation relationships posited were tested, and results indicate that four out of the five ODA characteristics examined in this study had significant influences on consumer attitude toward online advertising (ATOA) as well as 207 University of Ghana http://ugspace.ug.edu.gh behavioural responses with differing strengths of effects. Attitude toward online advertising was found to fully mediate the relationships between these four ODA characteristics (interactivity, personalisation, informativeness and exposure condition) and ad avoidance. For ad acceptance, attitude toward online advertising partially mediated its relationship with interactivity and personalisation; and fully mediated its relationship with informativeness and exposure condition. As the moderator variable in the study, user mode moderated the relationship between personalisation and ad acceptance as well as informativeness and attitude toward online advertising. Also, differences were found in consumers’ behavioural responses to ODA based on the nature of the advertised brands. These results are discussed in detail in the following chapter. 208 University of Ghana http://ugspace.ug.edu.gh CHAPTER EIGHT DISCUSSION OF RESEARCH FINDINGS 8.1 CHAPTER OVERVIEW This chapter discusses the findings of the study obtained from the data analysis conducted in the preceding chapter. Discussions were done in relation to extant literature as results are presented in the context of previous studies, the research framework, and the setting of the study, and were structured on the basis of the research questions as well as the formulated hypotheses. Specifically, the chapter is in three major sections including the opening section. The second section provides a brief overview of the study, and further discusses the research findings to address the stated research questions in the introductory chapter of the study. The section first discusses the stimuli effect of ODA characterisitcs on ATOA and behavioural response, this is followed by the mediating role of ATOA in the relationship between the ODA characteristics and behavioural responses, the moderating influence of user mode on the relationship between ODA characteristics and ATOA as well as behavioural responses are also presented, and the section wraps up with discussions on the variations in the direction and intensitiy of behavioural responses based on the nature of the advertised brand. A summary of the chapter is then presented in the third section. 8.2 INTRODUCTION AND DISCUSSION OF FINDINGS The overarching aim of this study is to examine the effect of ad characteristics, consumer attitude toward online advertising, and user mode on behavioural responses to online display advertising. Arguments put forth in earlier chapters of the study, pointed to ODA as an essential form of online advertising which currently accounts for 49.7% of the online advertising market, and is its fastest-growing and dominant category (eMarketer 2018), thus, instigating marketers or several firms to invest heavily in display advertising 209 University of Ghana http://ugspace.ug.edu.gh (Fridgeirsdottir & Najafi-Asadolahi, 2018). However, issues of advertising clutter, and consumers’ selective attention to display ads have left practitioners in need of vital indicators to generate desired responses to their ads. A thorough review of the ODA literature suggested that to provide vital insights and offer strategic ways to enhance positive online advertising effects and decrease negative outcomes (Brajnik & Gabrielle, 2010; Tang et al., 2014), a clear and explicit examination of design element or executional characteristics, as well as consumer-related factors are essential areas of focus. Springing on the stimulus-organism-response (SOR) model, and the reversal theory, this study contends that to better understand consumer behavioural responses to online display ads, an assessment of specific ODA features or characacteristics (interactivity, placement, informativeness, personalisation and exposure condition) that serve a stimuli role, in consort with pertinent consumer-related variables and situations which function as intervening (attitude toward online advertising) and contingency factors (user mode) may offer valuable insights on their integrated influence in driving positive or negative behavioural responses. The major results from the theoretical and empirical investigations carried out in pursuit of this aim are discussed in detail in the succeeding sections of this chapter. 8.2.1 The Stimuli Effect of ODA characterisitcs on ATOA and Behavioural Response The study first sought to establish the relationship between ODA characterisctics and attitude toward online advertising, and behavioural responses. Five ODA characterisitics were examined in this study namely interactivity, placement, informativeness, personalisation and exposure condition. Results from the data analysis revealed that placement does not exert any influence on consumers’ or internet users’ internal responses – attitude toward online advertising (ATOA), neither does it on their approach/avoidance behavioural responses. Within the tenets of the SOR model, placement of ODAs from the perspective of Ghanaian consumers or internet users does not serve as a useful stimulus that 210 University of Ghana http://ugspace.ug.edu.gh generates internal or any form of behavioural response. Personalisation, interactivity and informativeneness however, surfaced as notable ad characteristics that directly influence consumers’ approach behavioural responses to online display advertising (ad acceptance). Personalisation emerged as the only ODA characteristic that directly influences negative behavioural responses (ad avoidance). The study results also show informativeness, personalisation, interactivity and exposure condition as significant influencers of attitude toward online advertising (ATOA). Because a unique contribution of this study lies in its examination of the effects of this individual ODA characteristics, the discussions in this section are organised to focus on each ad characteristics and the stimuli function they play in influencing consumers’ internal response (ATOA) as well as approach-avoidance behavioural responses (ad acceptance and ad avoidance). 8.2.1.1 Personalisation Among the ODA characteristics revelaed to serve a stimuli function, personalisation emerged as the only predictor of consumers’ attitude toward online advertising as well as both ad acceptance and ad avoidance. As ealier explained, personalisation denotes the degree to which advertising messages are tailored to reflect preferences, lifestyle and specific cultural and geo-demographic characteristics of individual consumers (Leppäniemi & Karjaluoto, 2008). The effects of personalisation in the context of online display advertising has been of vital attention in prior recent studies with findings generally suggesting that personalised ads enhance more attentive processing of ads and enhance favourable attitude toward such advertisements (Campbell & Wright, 2008; Sahni et al., 2018). Studies such as van Doorn and Hoekstra (2013) also opine that personalisation increases negative perceptions and generate unfavourable responses among consumers resulting from feelings of intrusiveness. 211 University of Ghana http://ugspace.ug.edu.gh Contrary to the mixed findings in prior literature, fairly consistent results emerged in this study as personalisation positively influenced ad acceptance and consumers’ ATOA, and negatively influenced ad avoidance. Although its effects on all three response variables were significant, relative to ad acceptance and ATOA, the effect of personalisation on ad avoidance was less prominent. The negative influence of personalisation on ad avoidance, disputes the submission of researchers that tracking online activites and collection of personal data are at variance with consumers’ privacy concerns (Baek & Morimoto, 2012), which results in higher levels of intrusiveness and uneasiness especially in cases where ads reflect consumers’ exact preferences to a high degree (Bleier & Eisenbeiss, 2015a). In essence, in this study, consumers consider personalised display ads as less distracting, and the negative influence suggest that the more personalised an ad, the less consumers will avoid it. A conceivable explanation put forth as accounting for the findings in the current study is that, consumers are increasingly becoming cognizant of the tracking of their online activities as well as collection and analysing of their personal data and information (Liu & Matilla, 2017), and this knowledge could be a probable buffer for the negative outcomes of personalisation suggested by earlier studies. The significant positive effect of personalisation on ad acceptance and ATOA notheless, affirms perspectives and findings from studies such as Tucker (2014), Aguire et al. (2015) and Liu and Matilla (2017). The work of Aguirre et al. (2015) for instance, found that consumers’ perception of an ad as being personally relevant, substantively influences their favourable attitude and responses toward such ads. Tucker (2014) in their study as well found that personalised display ads are twice as effective as non-personalised ads since they match the preferences and tastes of consumers, and so attract more consumer attention, engagement and favourable evaluations. These positive influences are driven by the 212 University of Ghana http://ugspace.ug.edu.gh heightened benefits consumers perceive they may gain from such ads (Lambrecht & Tucker, 2013) since messages in personalised ads are assumed to be directed toward them. Overall, it can be gathered from the findings of this study that in the Ghanian context, display ads that are tailored to consumers’ shopping situation at the time of internet usage, browsing activities, or exposure are likely to engender more acceptance manifested in interactions and attention to the ad. Also, to get consumers to perceive display ads more favourably, advertisers and publishers need to demonstrate knowledge of their consumers or internet users, evoking in them a feeling of uniqueness by making product recommendation through ads that match their behavioural, preferential and locational needs (Liu & Matilla, 2017). Furthermore, with the volume of online display advertising on a constant increase, generic (unpersonalised) ads are gradually becoming less effective. Thus, in order to break through the clutter, and reduce negative consumer responses to display advertsising, ODAs should be personally relevant or meaningful to consumers among other things. This is particularly vital as personalisation emerged as the only ODA characteristic that directly, significantly and negatively influences ad avoidance in this study. 8.2.1.2 Interactivity As a defining feature of the internet, interactivity is also an essential characteristic of online advertising and has been posited to play a vital role in the effectiveness of online display advertisements (Jung et al., 2014) and was thus, examined in this study. Besides personalisation, interactivity was also found to have significant influence on ATOA (β = 0.15, t-value = 3.96, P<0.001) and ad acceptance (β = 0.18, t-valu e= 4.27, P<0.001) but not ad avoidance (β = -0.01, t-value = -0.12, P = n.s.). Interactivity is one of the commonest and a default feature of online ads that allows consumers to click through to brand pages in order to access more product or service information (Rosenkrans, 2009). Interactivity as examined 213 University of Ghana http://ugspace.ug.edu.gh in this study however, transcended this default aspect of ads to include their ability to allow consumers to provide feedbacks, provide instantaneous information in response to consumer request or enquiries, and make consumers feel that advertisers want to listen to them. The findings that interactivity did not exert any influence on ad avoidance failed to support the stated hypothesis (H6aii), which posited its negative effect on ad avoidance. This finding was unexpected since pashkevich et al. (2012) in their study asserted that interactivity reduces negative advertising effects or outcomes. This is on the basis that higher levels of interactivity offer users a certain degree of control over the ad content to which they are exposed, and the presence of such lessens the distractive and intrusive perceptions they hold of such ads (Johnshon, 2013). Conversely, McCoy et al. (2008) earlier contended that consumers find the mechanism of control as invasive since they will have to wilfully close, move or interact with the ad when their actual desire may be to not have the advertisement at all. Gleaning from the results in the CFA, two scale items of interactivity which depict control; did not emerge as valid measures of the construct in the Ghanaian context of online display advertising, hence providing a probable explanation for the lack of effect on ad avoidance. The positive relationship between interactivity and consumers’ acceptance of online display ads as well as their favourable perceptions of such ads are supported in this study. Interactive ads are known to allow consumers to participate in their ad viewing experience which encourages them to engage with the ads and so stimulates more positive behavioural response (Rosenkrans, 2010). Jung et al. (2011) affirm that consumers develop favourable perceptions about display ads they find entertaining, and the entertainment value of an ad is influenced by the level of ad interactivity. The findings also appear consistent with Pashkevich et al. (2012), who found that the interactive element of skippable video ads is a 214 University of Ghana http://ugspace.ug.edu.gh valuable feature influencing consumer attitude. By virtue of its inventive qualities, interactivity is also contended to urge users to pay closer attention to ads, and also induce cognitive involvement in processing such ads (Jung et al., 2014). The study findings in this regard, therefore, lend credence and further support to extant research which found interactivity as an essential determinant of online advertising effectiveness (Campbell et al., 2010; Bleier & Eisenbeiss, 2015b), and also from the viewpoint of the SOR model, establishes interactivity as an essential stimulus that generates positive internal responses and approach behaviours. 8.2.1.3 Informativeness Also of significance to consumers’ ATOA and ad acceptance is the informativeness of display ads. The provision of information has long been a basic role of advertising and has been shown to be a fundamental reason why consumers approve of advertisements in all media channels (Rotzoll, 1990). Informativeness is considered one of the content-driven characterisitics of advertising and it depicts the degree to which advertising offers valuable information to consumers. Kim et al. (2010) argue that because consumers require beneficial information that is easily understandable, they expect online ads to be informative and helpful to aid them detect product/service brand distinctions and make choices more readily. The study therefore, proposed that informativeness has a significant positive influence on attitude toward online advertising and ad acceptance, and these postulations were supported by the study findings which revealed significantly positive relationships between consumers’ perception of display ads as being good, convenient and relevant sources of valuable product/service information, and their ATOA as well as their acceptance of display ads. The reasoning here is that consumers will most possibly find display ads interesting, essential and have favourable opinions of them if the ads make complete product/service information easily accessible. In such instances, consumers may also most likely read/watch 215 University of Ghana http://ugspace.ug.edu.gh online display ads to the end and click on the ads and links provided in such ads. Thereby liking and interacting with the ads on the basis of the ad’s informativeness (Mahmoud, 2014). These findings find support in the work of Goodrich et al. (2015) who in the context of video ads assert that informativeness is a way to improving receptivity toward and interaction with online ads; and online ads with beneficial information are perceived more favourably by internet users. The findings also confirm the works of Li-Ming et al. (2013) and Mahmoud (2014) who point to informativeness as a gratifier of the informational needs of internet users thus, driving them to attend to ads that fulfil this need. Zha et al. (2015) therefore, admonish that in the online environment where advertisers intend to appeal to consumers’ and have them respond positively to their ads, it behoves them to make ads as informative as feasible. This is particularly helpful if their consumers actively use online ads for information gathering. As can be gathered from the internet usage profile of the study sample, a majority (approximately 35%) of them use the internet mainly for information (research) purposes, which provides further insight into the significant influence of informativeness as a vital stimulus that engenders positive attitude toward online advertising as well as approach behavioural response in the Ghanaian setting. Regarding the absence of a relationship between informativeness and ad avoidance, evidence in extant literature are contrary to the finding in this study. Some of these evidences suggests that when consumers find the information provided in display advertisements useful, their perceptions of intrusiveness and irritation are lessened (Goodrih et al., 2015). It is also argued by Rejón-Guardia and Martínez-López (2014) that perceptions of intrusion and irritation are mental states that drive avoidance of online ads and other negative behavioural outcomes that may be detrimental to a firm’s advertising goals. Since the study 216 University of Ghana http://ugspace.ug.edu.gh failed to prove that ODA informativeness lessens or reduces ad avoidance, it seems fairly reasonable to assume that the relationship between the two constructs may be better explained through some other mechanism which is discussed in a subsequent section on the mediating role of ATOA as postulated by the SOR model. 8.2.1.4 Exposure Condition Exposure condition also surfaced in this study as a major ad characteristic impacting consumers’ attitude toward online advertising. Within the boundaries of this study, exposure condition was conceptualised to initially comprise variables such as how long an ad is, the frequency with which it is repeated, and the degree to which consumers are forced to view the ad (nature of the exposure). After the exploratory and confirmatiory analysis however, the resultant indicators measuring the construct in the final model were those of the nature of the exposure that is the extent of forceful or voluntary exposure to online ads (Edwards et al., 2002; Li & Meeds, 2007). The significant relationship found between exposure condition and ATOA indicates that consumers’ perception of the non-interferance of display ads in addition to their non-intrusion on the content they are accessing at the time of exposure have a noteworthy influence on their attitude toward online advertising. As has been reiterated in earlier chapters, the internet is a more goal-oriented medium unlike traditional media. Owing to its goal-orientedness, the disruptions consumers experience during their online activities are correspondingly more amplified as ads become sources of inconvenience hindering them from their actual reasons for being online (Rejón-Guardia & Martínez-López, 2014). For this reason, within the domain of online display advertising, such issues as being able to choose freely what they want to see, not having ads interfere with their online activity, as well as not intruding on the content they wish to access, hold so much significance for consumers or internet users. 217 University of Ghana http://ugspace.ug.edu.gh Evidently, findings in this study confirm the position of previous studies (e.g. Hegner et al., 2015) who have reported negative perceptions of consumers towards forced exposure ads propelled by perceptions of intrusiveness. The study findings are as well in consonance with other schorlarly assertions that when consumers are coerced to watch ads if they wish to continue with their online activities, higher perceptions of ad intrusiveness and more negative attitude toward the ad are produced which then affects other consumer behavioural outcomes (Baek & Morimoto, 2012; Campbell et al., 2017). These assertions have been confirmed on the one hand, within the study context as unforced exposure was found to significantly influence consumer attitude toward online advertising. On the other hand, because prior empirical works in the general online advertising literature have established perceived goal impediment as a predictor of ad avoidance (e.g. Yoon et al., 2011), we expected that unforced exposure may reduce ad avoidance. More so because, Goldfarb and Tucker (2011) found that when ad messages interrupt but do not totally cover the webpage’s content, consmers may view such ads as mere disturbances but may not necessarily avoid such ads. These suppositions were however, not supported in the current study as exposure condition did not to a significant extent influence (reduce) ad avoidance as posited. This affirms that avoidance behaviours of internet users or consumers in Ghana is not directly driven by exposure conditions but may do so through some other underlying mechanism. Notwithstanding, findings regarding ATOA call for advertisers to display online ads with minimal forced exposure or more essentially in voluntary modes. This is more so feasible as the internet by means of evolving and complex technologies aids in generating varying levels of exposure (Hussain et al., 2018) unlike traditional media. 218 University of Ghana http://ugspace.ug.edu.gh 8.2.2 The Mediating Effect of ATOA in the ODA Characteristic-Response Relationship Establishing the influence of ODA characteristics on behavioural responses and ATOA; particularly, acknowledging the lack of direct influence of most of the ODA characteristics on avoidance behaviours, the study also sought to further examine ATOA as an internal process that may mediate the positive and negative relationships between the ODA characteristics and approach behaviours as well as avoidance behaviours of consumers respectively. According to the SOR model, consumers react to stimuli in two stages, the first are internalised affective or cognitive responses indicated as organism, which drive the external response (Chen & Yao, 2018). In order to establish ATOA as a consumer-related conditions under which the influences of ODA features are enhanced, in this thesis, ATOA is operationalised as the organism element of the study, grounded on early scholarly positions (Petty et al., 1983) that attitude formation is an internal process that eventually guides behaviour. This section, therefore, addresses the second research question “how does consumer attitude toward online advertising mediate the relationship between ODA characteristics and consumer behavioural responses?” Generally, results of the mediation analysis provide support for the intervening role of ATOA in the relationship between the ODA characterisitics (Interactivity, Personalisation, Informativeness and Exposure condition) and ad avoidance as well as ad acceptance. Specifically, the findings revealed that ATOA partially mediates the positive relationship between interactivity and ad acceptance as well as between personalisation and ad acceptance; all other relationships were fully mediated (see Figure 6.3). In the case of ad avoidance, these findings suggest that ATOA strongly and significantly mediated the negative influence of the ODA characteristics on consumers’ avoidance behaviours. The inference is that, even though the ODA characteristics do not directly lessen consumers’ avoidance behaviour, they do so indirectly through consumers’ ATOA as a facilitating 219 University of Ghana http://ugspace.ug.edu.gh variable. To put this differently, any possibility of the ODA characteristics negatively influencing ad avoidance are only existent through the attitude they form toward online advertising. These findings, therefore, support the view of the SOR model which fundamentally hinges on the idea that the relationship between any stimuli and responsive behaviour is intervened by the internal processes of the organism exposed to such stimuli (Kamboj et al., 2018). Emperically, the findings support Sung and Cho’s (2012) assertion that consumer attitude plays a critical role in the effectiveness of online advertising; and also corroborate Wang and Sun’s (2010a; 2010b) position that attitude toward online advertising mediates the relationship between consumers belief about online advertising in general, and their online behaviours of ad clicking and shopping frequency. Other past studies have demonstrated that consumers’ attitude toward online advertising which stems from perceptions they hold about online ads (Réjon-Guardia & Martinez-Lòpez, 2014), affect their responses toward online advertising (Saadeghvaziri et al., 2013). Positive attitude toward online advertising has also been associated with more positive behavioural responses toward ads and even purchase intentions (Goodrich, 2011; Goodrich et al., 2015). On the basis of these findings therefore, ATOA is extremely relevant in the context of display advertising in enhancing consumer’s approach behaviours and reducing avoidance behaviours. Consistent with previous studies on online advertising, it seems reasonable to state that, for consumers or internet users to be responsive and attentive to display ads as well as interact with them, they must have generally favourable perceptions of, like, and consider online advertising as essential. Much more pertinently, for ODA characteristics to reduce consumers’ avoidance of display ads, these characteristics must first generate positive attitude toward online advertising. In other words, in the absence of a general 220 University of Ghana http://ugspace.ug.edu.gh ATOA, interactivity, informativeness and exposure condition hold no relevance for consumers avoidance of display ads. Since these findings agree with prior empirical results and lend credibility to the SOR model, the study thus, provides a solid footing for making generalisation particularly in the context of Ghanaian consumers that, attitude toward online advertising is a strong mediator of the ad characteristics-behavioural response relationship. 8.2.3 The Moderating Effect of User Mode As a follow-on to earlier discussions, this section discusses findings from the analysis on the influence of user mode on the nexus between the ODA characteristics and the behavioural responses of consumers as well as between the ODA characteristics and consumer attitude toward online advertising. Consistent with the overaching aim of this study, it was pertinent to examine how a situational or an individual factor such as the goal orientation of consumers driven by their internet usage motives regulates their attitude toward online advertising and their behavioural responses. Internet usage motives are in four categories according to the internet motivation inventory classification framework – researching, shopping, communication, and surfing (Rodgers et al., 2007). These internet usage motives per the reversal theory, are classified into telic and paratelic user modes on the basis that those with researching and shopping motives are more goal-directed/serious- minded (telic) while users that are mostly surfing and communicating are less goal (experience)-directed/playful-minded (paratelic) (Apter, 1984; Rodgers et al., 2007). These two user modes were the focus of this study given that consumer perceptions of, and reactions to online advertising are conditioned by these dimensions (Wang et al., 2013; Kim et al., 2014). This study substantiated this claim by conducting a moderation analysis using the supported direct paths. A multigroup moderation analysis was conducted following the suggestion of 221 University of Ghana http://ugspace.ug.edu.gh Anderson and Gerbing (1988). The results of the analysis revealed that user mode moderated the strength and direction of two paths across the two user groups – the relationship between personalisation and ad acceptance as well as the relationship between informativeness and attitude toward online advertising. Specifically, the effect of informativeness on ATOA was stronger for paratelic users than for telic users. Likewise, personalisation had a significant positive effect on ad acceptance among telic users but not among paratelic users. This study to some extent provides empirical support to prior studies that have demonstrated how user mode regulate the effectiveness of online advertising (see Simola et al., 2011; Eshghi et al., 2017; Bleier & Eisenbeiss, 2015a), and also corroborate assumptions regarding the telic vs. paratelic dimension of the reversal theory (Seyeghorban et al., 2016). Regarding the informativeness-ATOA relationship, it may be reasoned that because telic users are driven by defined objectives and directed search, they focus on specific tasks and are not very attentive to other peripheral activities as much as paratelic users who are driven by the browsing experience and other activities that come with it (Bleier & Eisenbeiss, 2015a). For this reason, telic users are less likely than paratelic users to recognise and recall ads, better still to recollect how informative such ads are (Zanjani et al., 2011). And so, the influence of this perception on subsequent attitude toward online advertising being stronger for paratelic users is thus expected. In the case of personalisation and its impact on ad acceptance, it has been contended by Bleier and Eisenbeiss (2015a) that because personalised display ads catch consumers attention more easily, they are likely to be seen as more distrupting. However, in the context of this study, personaliation had a strong direct impact on ad acceptance generally; and in the moderation analysis, for telic users but not for paratelic users. In providing a probable explanation for this finding, this study argues on the basis of relevance and suggests that telic users who are more goal-directed give more attention to related than unrelated materials when online (Stanaland & Tan, 2010; Janssens 222 University of Ghana http://ugspace.ug.edu.gh et al., 2012), and as such are more receptive towards display ads that are tailored to their shopping needs, that make recommendations matching their needs, thus conferring a sense of uniqueness on them as internet users. In this sense, this finding appears incongruent with Schuman et al. (2014) who found that appealing to reciprocity in lieu of relevance increases acceptance of online advertising among telic users. These results go to show that marketers and advertisers must consider it prudent to examine consumers’ internet usage motive prior to making decisions regarding how to personalise ads to reflect their preferences and interests, as well as deciding on the degree of informativeness with which to imbue such ads. This would be particularly helpful at the individual consumer level because usage motive or task/goal-directedness normally varies among consumers, and also because with complex tracking algorithm and technologies, advertisers have access to large volumes of consumer data even at individual levels. 8.2.4 Behavioural Responses to Product and Service-featured ODAs Generally, product types and peculiarities cause some differences in online advertising effects (Belanche et al., 2017). Evidence from the online advertising literature points to disparities in consumer processing and responses to the online ads of different product categories (e.g. Bradley & Meeds, 2004; Flore et al., 2014; Eshghi et al., 2017). In view of this, from the onset, this study argued that possible disparities in online advertising responses could be theorised for products and services particularly because the extant ODA literature is deficient in emperical evidence on such disparities. Besides, given the increasing imbibing of the service-dominant logic which suggest a diminishing relevance of the goods- service dichotomy (Vargo & Lusch, 2017), this study attempted to understand the disparities (or lack thereof) that might exist in consumer responses to online display ads that feature product brands and those that feature service brands. The study’s outcome in this regard, revealed 223 University of Ghana http://ugspace.ug.edu.gh quite interesting perspectives in the ODA subgroups. First, the ANOVA test revealed that avoidance was higher for service-featured ODAs relative to product-featured ODAs while acceptance was higher among consumers exposed to product-featured ODAs compared to those exposed to service-ODAs. Regarding the intensity of behavioural response, the logistic regression results (see Table 7.15) show that while interactivity, personalisation and exposure condition were all significant influencers of active acceptance for product-featured ODAs, interactivity was the only statistically significant influencer of active ad acceptance for service-featured ODAs. A possible explanation to this lies in the fact that since services are intangible in nature, requiring interactions with both providers and consumers to be produced (Davis, 2007; Skaalsvik & Olsen, 2014), for display ads featuring service brands consumers are driven by the need to interact, thus using interactivity of the ad as a pseudo for the providers. Results also showed that contrary to service-featured ODAs, personalisation for instance, was a significant negative predictor of active avoidance for product-featured ODAs; while exposure condition was a viable predictor of active avoidance for service-featured ODAs relative to product-featured ones. These findings suggest that brand types (in terms of product and services) are relevant and play essential roles in eliciting behavioural responses in varying directions and intensities (Tang et al., 2014; Eshgi et al., 2017). As such, it seems prudent for advertisers to consider their market offering when making choices regarding the ODA characteristics to mix and match so as to elicit the desired behavioural response. This is even more so relevant for marketers of service brands since avoidance was found to be higher for service-featured ODAs. 224 University of Ghana http://ugspace.ug.edu.gh 8.3 CHAPTER SUMMARY This chapter discussed the findings of the study in order to address the research objectives and questions outlined in the introductory chapter, and so, the discussions were structured to address, the effect of the ODA characteristics on consumers’ attitude toward online advertising, and behavioural responses. This was followed by discussions on the mediating role of attitude toward online advertising as well as the moderating effect of user modes on these relationships, and finally discussions on behavioural responses to product and service-featured ODAs. The findings of the study supported the perspectives and assumptions of both the SOR model and the reversal theory and offer important pointers for the design of online display advertising strategies. The next and final chapter presents the research conclusions and the contributions of the study to theory. The study also provides strategic implications the study holds for managerial practice. At the same time the chapter highlights and reflects on the processes employed in the study, acknowledges the limitations of the thesis and also presents avenues that may be the focus of future studies. 225 University of Ghana http://ugspace.ug.edu.gh CHAPTER NINE SUMMARY, CONCLUSIONS AND IMPLICATIONS 9.1 CHAPTER OVERVIEW This, as the final chapter of the study, summarises the research, draws conclusions, presents implications and proffers directions for future research. Particularly, the chapter assesses vital outcomes from the study, and discusses the contributions the study makes to the academic literature as well as managerial practice. As such, beside the overview section, the chapter is divided into five major sections. A succinct summary of the research problem and objectives are provided, and key findings corresponding to these objectives are presented. The next section reflects on the theoretical framework, the conceptual framework and the methodological approaches employed in order to further establish their appropriateness as they pertained to the conduct of this thesis. The chapter then presents contributions this study makes to theory as well as implications for practice. This is followed by the research conclusions and the limitations of the study, and finally vital pointers that future research may pursue in addressing consumer response issues in online advertising. 9.2 SUMMARY OF THE RESEARCH AND MAJOR FINDINGS The overall aim of this study was to examine consumer behavioural responses to online display advertising by delineating and assessing how specific ODA characteristics, in consort with consumer attitude toward online advertising as well as a situational factor such as user mode positively and negatively influence these responses from a developing sub- Sahara African country context. These ad-related and consumer-related issues though of theoretical and practical concern for online advertising, were found to have received limited scholarly attention, and extant research attention to these issues appeared fragmented and 226 University of Ghana http://ugspace.ug.edu.gh less disciplined. A thorough review of the literature on ODA revealed five pertinent ad- characteristics namely, interactivity, placement, informativeness, personalisation and exposure condition that needed further research focus to determine how they affect the effectiveness of online display ads. In addition, to provide further insights into which consumer-related factors may also impact consumer responses to display advertising, the study from the perspective of the stimulus organism response (SOR) model, operationalised consumer attitude toward online advertising (ATOA) as an internal “organism” process that may transmit the effect of the ODA characteristics onto consumers’ approach and avoidance behavioural responses. The key constructs in the study, that is the five ODA characteristics were theorised as stimuli or advertising cues that are vital in eliciting both internal and external consumer responses. Also, from the viewpoint of the reversal theory, the study conceptualised user mode as a vital contingency variable that may confound the relationship between the ODA characteristics (stimuli) and consumer ATOA (organism) as well as ad acceptance and ad avoidance behaviours (responses). In order to achieve its broader aim, the study started off with four key research objectives in the introductory chapter: (1) to determine the relationship between the ODA characteristics and consumer behavioural responses, (2) to examine the intervening role of attitude toward online advertising in the ODA characteristic-behavioural response relationship, (3) to assess the moderating effect of user mode on the relationship between the ODA characteristics and ATOA as well as behavioural responses, and (4) to explore the variations in behavioural responses based on the nature of the advertised brand. Following the introductory chapter, the study context was discussed and recent developments in online advertising in the Ghanaian setting were presented in order to put 227 University of Ghana http://ugspace.ug.edu.gh the entire research work in a suitable perspective. The theoretical framework within which the study was conducted was discussed focusing on the SOR model and the reversal theory as appropriate theoretical foundations that help explain the interrelationships the study seeks to address. Following this was a fairly systematic enquiry into the literature on ODA spanning a ten-year period (2009-2018) in order to understand relevant issues in the field. Various gaps were identified through the review relating to issues, context, theoretical and methodological approaches. Discourses in the three preceding chapters resulted in the development of a research framework that guided the empirical aspect of the study. The assumptions underlying the framework were that, that online display advertising (ODA) is characterised by certain stimuli/features conceptualised in this thesis to include interactivity, placement, informativeness, personalisation, and exposure condition. These ad characteristics are postulated to have a direct influence on consumer behavioural responses (ad avoidance and ad acceptance) to display ads. The framework also suggests that these influences are enhanced (mediated) by consumer attitude toward online advertising. Underpinned by the reversal theory, the framework further purports that the effects of the ODA characteristics on attitude toward online advertising, and behavioural responses are moderated by internet user mode. Springing on these assumptions, various conceptualisations and prior empirical research in the broader online advertising and ODA literature, the study derived hypotheses that were tested using a combination of online and offline surveys of 592 internet users in Ghana who are exposed to online display advertising. The analysis of the empirical data yielded some valuable results which were discussed in relation to extant literature and the study context in the immediate preceding chapter. These findings are recapped or presented in precis based on the research objectives in the Table 9.1 below. 228 University of Ghana http://ugspace.ug.edu.gh Table 9.1 Summary of Key Findings from the Study Research Key Findings Objectives Objective One: The study revealed that interactivity, informativeness and personalisation are essential ODA characteristics that directly elicit approach behaviours (ad Determine the acceptance) among consumers; thus, providing evidence to show that these relationship between characteristics serve a stimuli function that influence attentive processing of ODA characteristics ads as well as consumers’ interaction with such ads. Relevantly, personalisation and consumer emerged as the only ODA characteristic that influences avoidance behaviours behavioural by lessening consumers’ inclination to ignore, skip or scroll away from online responses. display ads. In the case of approach behaviours, interactivity has the strongest influence on ad acceptance, followed by personalisation and informativeness. Objective Two: Outcomes of the study supported the mediating role of consumer ATOA in the ODA characteristic-behavioural response relationship. Particularly, Examine the consumers’ attitude toward online advertising enhances the effects of intervening role of interactivity, personalisation, informativeness and exposure condition on ad consumer ATOA in the acceptance as well as ad avoidance. More definitely, even though most of the relationship between ODA characteristics do not directly influence avoidance behaviours among ODA characteristics consumers, they do so indirectly through ATOA as a facilitating variable. and behavioural responses. These findings evidently support the viewpoint of the SOR model which suggest that the relationship between any stimuli and responsive (approach/avoidance) behaviour is intervened by the internal processes of individuals exposed to such stimuli. By this, ATOA is established as a crucial internal response that facilitates the positive effect of ODA characteristics on approach behaviours; and also transmits their negative effect on consumers’ avoidance behaviours toward online display advertising. Objective Three: The study provided evidence to fairly buttress the telic vs. paratelic perspective of the reversal theory by demonstrating that user mode moderates the Assess the moderating relationship between informativeness and ATOA as well as the relationship effect of user mode on between personalisation and ad acceptance. To be more precise, the effect of the relationships informativeness on ATOA was stronger for paratelic users than for telic users; between ODA while personalisation had a significant positive effect on ad acceptance among characteristics, and ATOA as well as telic users but not among paratelic users. Thus, making the need for advertisers behavioural to examine and keep abreast with consumers’ internet usage motives a vital responses. endeavour. Objective Four: The study’s outcome in this regard, revealed differences in behavioural responses across the ODA subgroups. On the basis of these outcomes, Explore the variations avoidance was higher for service-featured ODAs relative to product-featured in consumer ODAs while acceptance was higher for product-featured ODAs compared to behavioural responses service-featured ODAs. What is more, interactivity, personalisation and based on the nature exposure condition were all significant influencers of active acceptance for 229 University of Ghana http://ugspace.ug.edu.gh (product vs. service) of product-featured ODAs, whereas, interactivity was the only statistically the brand being significant influencer of active ad acceptance for service-featured ODAs. advertised. Results also showed that contrary to service-featured ODAs, personalisation for instance, was a significant negative predictor of active avoidance for product-featured ODAs; while exposure condition was a viable predictor of active avoidance for service-featured ODAs relative to product-featured ones. These findings point to brand types (product and services) as relevant and vital in evoking behavioural responses in varying directions and intensities. 9.3 REFLECTIONS For any valuable research work, it is essential that the researcher does some form of retrospection on the procedures, approaches and routes taken in providing answers to the stated research questions and objectives. The goal for such appraisal is to re-validate and further substantiate the suitability of the various designs and processes employed to arrive at the most useful contributions made by the study relative to alternative approaches that could have been used. In light of this, discussions in the section are centred on the theoretical framework, the conceptual framework, and the methodological approach which represent three key areas on which this study is grounded. 9.3.1 Reflections on Theories The review of extant ODA literature conducted in the third chapter of the thesis pointed to the lack of theory-based approaches to examining online display advertising effects relative to atheoretcal (e.g. category and model-based) approaches. Additionally, the nature of the study necessitated the application of theories that would support the integrated role of ad- related and consumer-related factors as vital to determining online display advertising outcomes. In view of this, the study employed two main theories namely, the Stimulus- Organism-Response model and the Reversal theory as pertinent theoretical perspectives that could shape and direct the conduct of this study. The choice of these theories was predicated on evidence from the literature highlighting both ad-related and consumer-related factors as 230 University of Ghana http://ugspace.ug.edu.gh vital to online display advertising effects. Consequently, the SOR model assisted in understanding and explaining the link or association between the ODA charscterisitics, consumers’ ATOA, and their approach/avoidance behaviours towards online display ads. The study within the framework of the SOR model therefore, conceptualised ad characteristics as major online display advertising simuli, attitude toward online advertising as an essential internal/organismic response that may transmit the effect of the stimuli onto consumers approach and avoidance responses. In retrospect, the SOR model was the most apt theory to help situate ODA characteristics, ATOA, and behavioural responses as key constructs in light of the research objectives and framework. Also, the literature review indicated that the effectiveness of online display advertising, measured in terms of conumers’ reaction to such ads, are largely regulated by their goal-directedness. These assertions derive from the goal orientation of the Internet as a medium since consumers and users go online for various specific purposes (e.g. communicate, surf, shop and search). As such, the reversal theory particularly, the telic vs. paratelic dimension, which offers insights into how these task/goal directions of consumers may affect their responses to display advertising was also applied to the study. Given the researcher’s cognisance of the limited ability of the individual theories to fully explain and provide a comprehensive understanding of the relevant issues addressed in this thesis, the two theories were used in conjunctive manner. The use of these two theories provided a clearer and more comprehensive understanding of the research problem, objectives, the research framework that captured the interrelationships the study examined as well as the methodological procedures uitilised. Though these theories have been sparsely used by extant works on online display advertising (Tang et al., 2014; Bleier & Eisenbess, 2015b; Seyedghorban et al., 2016) particularly, in such complementary manner as done in this study, they are nonetheless not new to the the subject under investigation. It also seems 231 University of Ghana http://ugspace.ug.edu.gh appropriate to point out that insofar as understanding consumer responses to online advertising are concerened, these thoeries are far from being exhaustive, although they were the most useful within the boundaries of this thesis. Albeit it is believed that theories such as the elaboration likelihood model, the mere exposure effects theory, selective attention theory, and reactance theory among others could have revealed added viewpoints, it is also the researcher’s contention that such theories would not have provided deeper and more ample understanding of the research objectives particularly, in light of the research framework. This would have restricted the contributions the thesis seeks to make, that is enhance understanding of behavioural response issues in ODA research. For this reason, their exception from this thesis did not by any means undermine or weaken the findings of this study, given that the applied theories provided valuable support for the study and its outcomes. 9.3.2 Reflections on Conceptual Framework and Post-Study Framework As has been reiterated in several sections of this research, following the specification of the theoretical framework, and a thorough review of the ODA literature, in the fifth chapter of the thesis, a conceptual framework was developed to guide the empirical conduct of the study. The framework used relevant theories, concepts and constructs from the literature review to establish interconnections and propositions among ODA characteristics, attitude toward online advertising (ATOA), and behavioural responses of consumers. The framework also incorporated user mode as a contigency variable that moderated the other relationships, as well as control variables such as gender, age, internet usage rate and familiarity with online advertising. The results or outcomes of the study provided support for the constructs in the conceptual framework and confirmed most of the hypotheses posited in the study in the sense that one 232 University of Ghana http://ugspace.ug.edu.gh ODA characteristic (placement), and some other hypothesised relationships illustrated in the initial conceptual framework presented in chapter five, were not supported. As such, the proposed conceptual framework was revised to reflect results from the data analysis. The post-study framework pictured in Figure 9.1 relative to the initial framework, best depicts relevant ODA characteristics that influence Ghanaian consumers’ or internet users’ attitude toward online advertising and, in turn, their behavioural responses. The framework as well depicts the two paths or relationships (personalisation– ad acceptance, and informativeness– attitude toward online advertising) that are moderated by internet user mode. In essence, the post-study framework is a diagrammatic recapitulation of the study outcome. Fig ure 9.1 Post-Study Framework Behavioural Responses ODA Characteristics Ad Acceptance Interactivity Personalisation Attitude toward Online Advertising Informativeness Ad Avoidance Exposure Condition User Mode Although these key variables included in the conceptual framework were representative and essential to achieving the research aim and providing answers to the research problem, it is the submission of this thesis that, they were not all-inclusive. Given the wide array of ODA elements and characteristics, it seems within reason to suggest that the inclusion of other variables such as entertainment, animation and size among others could have surged or improved the predictive value of the research framework. Also, from the perspective of the consumer-related factors, there is also a likelihood that variables such as privacy concerns, 233 University of Ghana http://ugspace.ug.edu.gh perceptions of intrusiveness as alternative intermediate variables could have further clarified the relationship between consumers’ perceptions of ODA characteristics and their approach or avoidance behaviours. In spite of this, the outcome of the study which confirmed the stimuli effect of four out of the five ODA characteristics, and the strong empirical support provided for ATOA as a significant intervener in the focal relationship goes a long way to show that the choice of these variables was not misplaced, and the exclusion of other possible alternatives casts no doubts on the outcome of the study, neither does it provide grounds on which to demean the study findings. Particularly, as far as these variables were derived from a thorough review of the literature and appear congruent with the underpinning theories as well as other conceptualisations in prior studies, the resultant post-study framework is considered vital in offering insights for online advertising practice. 9.3.3 Reflections on Methodological Approach Since the study was conducted within the framework of the positivist paradigm, all other methodological choices were made in a logical fashion to align with the chosen paradigmatic stance. As such, a quantitative approach, and a survey strategy were adopted through the use of structured questionnaires to obtain the required information from respondents. The quantitative approach, and its corresponding survey strategy are justifiable on the basis that, the study sought to provide understanding regarding specific factors and their interconnections and associations with one another by testing relevant hypotheses derived from existing theories, and prior empirical works (Creswell, 2014; Saunders et al., 2012). Moreover, the research required the use of a larger sample of internet users to enable generalisation of the study ouctomes regarding consumer responses to display advertising in the online context through statistical explanations of the extent of associations that exists among the ODA characteristics, consumer attitude toward online advertising, internet user mode, and the approach/avoidance behavioural responses of consumers. 234 University of Ghana http://ugspace.ug.edu.gh It is possible that an experimental design or strategy may have provided much observable evidence particularly concerning consumers’ behavioural reactions in real media enviroments, and also increase the internal validity of the study. However, this would have offered a stagnant view of the reality as well as limit the sample diversity in terms of certain demographics (e.g. respondents’ age and occupation) and internet usage profile (e.g. usage frequency and familiarity with online advertising); particularly since the study controlled for some of these variables. Again, although an experimental design would have reduced the social desirability of responses that may have occurred given the survey strategy adopted, the quantitative analysis allowed for the controlling of outlying as well as skewed responses. It may also be argued that a qualitative approach to addressing the research problem and objectives would have revealed some added insights from internet users especially since calls have been made for qualitative approaches to addressing online advertising effects (Drossos et al., 2011; Tang et al., 2014). Notwithstanding, the quantitiative survey appeared more suited to the study goals given the large sample of respondents needed, and even a qualitative approach would have made it rather challenging to objectively test and confirm the hypothesised relationships. By adopting a positivist stance, and a quantitative surevy design, the study objectively examined the relevant issues pertinent to the research problem and through its rigourous and sequential analytical approach, provided more suited and valuable answers to the research questions than any alternative method would have allowed. 9.4 CONTRIBUTIONS AND IMPLICATIONS OF THE RESEARCH This study set out to provide a theoretical and practical understanding of consumer behavioural responses to online display advertising by offering insight into how ODA characteristics drive approach and avoidance behaviours, as well as explaining how 235 University of Ghana http://ugspace.ug.edu.gh consumer-specific factors like attitude toward online advertising and user mode can enhance ad acceptance and minimise ad avoidance among consumers. It must be stated that, the efforts made in this study by employing the stimulus organism response (SOR) model and the reversal theory to examine online display advertising and its effects, which resulted in the theory-based framework that guided our empirical results have produced vital theoretical contributions to the online advertising literature and managerial implication for online advertising practice which are discussed in the following sections. 9.4.1 Theoretical Contributions and Implications It is argued by Corley and Gioia (2011) that any research work that possesses originality and utility is fit to make valid theoretical contributions to knowledge. The authors maintain that originality speaks to the extent to which the research offers revelatory or incremental insights; and utility concerns the scientific and/or practical usefulness of the study. We consider contributions made by this research as offering incremental value to extant knowledge on online display advertising, and as such we also deem it useful for improving current research and advertising practice in the online domian. These study makes three key contributions to knowledge in the online display advertising field and the online advertising literature. Foremost, the study took stock of research works on ODA through a fairly systematic review of the literature which pinpointed relevant themes and issues, and also established pertinent gaps in these issues, theoretical and methodological approaches, and context. In so doing, this study establishes knowledge gaps that may guide future research in the area. The incremental value of the review conducted in this study, relative to other reviews in the extant literature (e.g. Kim & Macmillan, 2008; Ha, 2008; Liu-Thompkins, 2019) stems from its specific focus on ODA, the fastest growing category of online advertsing. This defined focus, allowed the study to delineate specific ODA issues, and 236 University of Ghana http://ugspace.ug.edu.gh provide deeper insights into the assumptions and rationales underlying their usage and the effects they have. Second, the review of extant literature and the review of the theoretical foundations called for more theory-based approaches to understanding how advertising works (Laczniak, 2015; Faber, 2015) as well as more studies to test these theories in the online advertising context (Jung et al., 2014; Bleier & Esenbeiss, 2015b). This study as a result of the reviews synthesised the diversive literature on online display advertising and developed a conceptual framework which integrates two relevant theories namely, stimulus organism response model and reversal theory. The integration of these two theories in this current study, responds to these pressing calls and demonstrates that reversal theory can complement the SOR model. A number of phenomena in the broader marketing literature have been studied using the SOR model and reversal theory but both theories have been separately and sporadically applied. Our findings suggest that the complementarity between the two theories can best be understood by considering the telic-paratelic dimension of the reversal theory as applied in this study, as a contingency consumer-variable that regulates the relationship between a stimulus (S) and the organism (O) as well as a stimulus and the responses (R). And so, to provide a more complete depiction of the effects of a stimulus, this study suggests the reversal theory is used to balance the SOR model as a possible contingency factor that causes variations in the direction and magnitude of both internal and external consumer responses elicited by environmental stimuli particularly in the online media context. This study passes as one of the few marketing studies to provide empirical backing to the combined applicability of these two theories in explaining behavioural responses to online display advertising. 237 University of Ghana http://ugspace.ug.edu.gh Additionally, the theory-based framework identifies, incorporates and describes how ad characteristics in consort with consumer-related factors enhance and/or mitigate consumers’ behavioural responses to online display advertising. The application of these two theories enabled the study to ascertain and highlight how relevant ODA characteristics (interactivity, informativeness, personalisation and exposure condition) could be employed and exploited as useful stimuli in eliciting approach behaviours, and assuaging avoidance behaviours toward display ads among consumers. By examining the effects of these ad-chacracteristics individually, the study provides detailed appreciation of their individualised effects on consumers’ evaluation of online advertising and their approach and avoidance behaviours. Much more profoundly, by confirming attitude toward online advertising (ATOA) as a vital internal consumer process or response that intervenes in the relationship between the ad stimulus (ODA characteristics) and behavioural responses of consumers, the study has demonstrated that when ad features do not directly influence behavioural responses significantly, the favourable attitude consumers have or form toward online advertising, can facilitate (enhance) such effects or influences. This research, therefore, provides vital insights to progress our understanding of how consumers’ favourable perceptions and evaluations of online advertising mediate the influence of ad characteristics on their approach and avoidance behaviours. The perspective of the reversal theory as applied in this study also provides clarifications on how telic and the paratelic user mode may vary the effect of specific ad stimuli on the direction and magnitude of the behavioural response of consumers. With heightened focus on interactive and informative advertising in the online environment (Mahmoud, 2014; Zha et al., 2015), this study throws more light on how such display ads can be valuable to advertisers through consumers’ favourable perceptions and acceptance of them. Thus, as a further contribution, this study demonstrates to a fair extent, the importance of 238 University of Ghana http://ugspace.ug.edu.gh understanding consumers’ motive (goal-directed/serious-minded or experience- directed/playful minded) for internet usage given the moderating effect these motives have on the effects of interactivity and informativeness on attitude toward online advertising and ad acceptance respectively. This is considered an extension of extant literature on online display advertising and the extant literature on reversal theory in the online advertising environment. The third important contribution of this study lies in the provision of empirical evidence and support for understanding consumer behavioural responses to online display advertising and the test of the above theories from an emerging economy context from sub-Saharan Africa. Since the review of extant ODA literature points to limited studies from developing country setting, and more specifically no apparent studies from Africa, the current study represents a response to the call for an investigation into different context using data from emerging markets on online advertsing effects (Valaei et al., 2016; Eshghi et al., 2017). The relative dearth of research into online display advertising and its effects from emerging market settings, and the seeming lack from an African perspective is telling. This is particularly so because, the literature indicates that individuals’ attitude and responses to advertising and their online behaviours are context-dependent and vary extensively by country (Wang & Sun, 2010b; De Mooji & Hofstede, 2010). Given such contextual disparities, this study has presented a different contextual perspective to the online advertising literature from a lower- middle income economy such as Ghana, and as one of the few of such, this study therefore, posseses contextual originality. 9.4.2 Implications for Online Display Advertising Practice A vital aim of any internet advertising research is to unearth and understand ways in which online advertising can be acceptable to consumers and in turn be applicable and effective to 239 University of Ghana http://ugspace.ug.edu.gh advertisers as well as appropriate for publishers (Li & Leckenby, 2004). The research in this regard, provides some actionable pointers or implications for online display advertising practice. A major challenge faced by advertising professionals and publishers is identifying executional features and design elements that can capture consumer attentention and ensure that display ads generate favourable reactions (Bleier & Esenbeiss, 2015b; Kim, 2018). This current study shows that an understanding of the direct and indirect (through consumer- related factors) influence of pertinent ODA characteristics can be valuable for making guided decisions concerning the sorts of ads to design and display. First, this research shows that interactivity, personalisation and informativeness are the relevant ODA characteristics that directly influence approach and avoidance behaviours of consumers toward display ads. Interactive display ads were found to have the most positive influence on consumers’ acceptance of ads. Inherent in this finding is the need for advertisers to embrace the knowledge that, infusing display ads with features that allow consumers to click ads for further informations or provide instantaneous information when consumers request as well as mechanisms that allow consumers the opportunity to provide feedback is a key step in engendering approach behaviours towards such ads which may be manifested in attentiveness, focus and interaction with the ads. This finding underscores interactivity as a requisite ODA stimulus or feature that consumers expect if they are to find display ads as acceptable, and calls on advertisers and publishers to leverage the constantly evolving interactivity of the internet as an advertising medium appropriatetely to this effect. Similar implications may apply in the case of informative display ads. Consumers consider as informative display ads that serve as good and convenient sources of information as well as provide complete and up to date information that keep them abreast of the product/service category and brand. For this reason, informativeness is also recommended as a vital driver of approach behaviours among consumers. Information search is one of the four major 240 University of Ghana http://ugspace.ug.edu.gh motives for internet usage (Rodgers & Thorson, 2000; Rhoades et al., 2008). And since internet tracking technologies allow publishers and advertisers access to consumers browsing history, beyond focus on product types, advertisers need to carefully consider the types of product-specific information consumers search for in order to provide such detailed information that consumers consider relevant in their display ads. In reference to personalisation, advertisers need to understand its dual stimuli function. This is to mean that unlike interactivity and informativeness which only positively influence ad acceptance, personalisation was found to positively influence acceptance behaviours and negatively influence avoidance behaviours as well. The findings in this regard provide indications that to effectively reach consumers, get them to not ignore, skip or scroll away from display ads as well as be attentive to and interact with ads, there is the need for advertisers to make ads personally relevant to consumers. This can be achieved by enhancing ad personalisation through tailoring ads to meet consumers search and shopping need, which makes them feel unique as internet users. Although personalisation has relatively less influence on ad acceptance than did interactivity and informativeness, and its negative influence on ad avoidance is even more so less significant, its twofold function promises that more benefits may accrue to advertisers and publishers if they personalised their display ads more. Second, we found evidence to support ATOA as a consumer-related condition under which the influences of ODA characteristics are ehanced or amplified. These findings confirm that while advertisers with a focus on emerging markets like the study setting can influence consumers acceptance of their display ads, and minimise avoidance through ad characteristics, such outcomes are much more feasible if consumers form positive ATOA. Springing on the evidence in the context of this study, advertisers and marketers are well 241 University of Ghana http://ugspace.ug.edu.gh advised that, albeit necessary, it might not be sufficient to imbue display ads with interactive, personalised, informative and unforced exposure features. These features must first result in positive consumer attitudes toward online advertising which will then translate into desired behavioural responses; particularly, if advertisers aim to reduce consumers’ avoidance behaviours toward their display ads because the effects of interactivity, informative and exposure condition for instance, are only present or manifest through attitude toward online advertising. In essence, the goal of engendering approach behavioural responses and lessening avoidance behaviours toward online display advertising can be achieved through ODA features that generate positive ATOA. Third, the importance of understanding consumers’ internet usage motive is underlined in this study. Illustratively, the positive influence of informativeness on attitude toward online advertising was stronger among paratelic or experience-oriented users than for telic or goal- oriented users. Likewise, personalisation was found to have a positive effect on ad acceptance only among telic users. In view of this, advertisers must give careful considerations to consumers’ motivation for internet usage before designing ads to persuade them. Specifically valuable for advertisers seeking to reach consumers driven by shopping and (re)searching motives (telic users), findings from this study suggest that they can enhance acceptance behaviours among this group of consumers by presenting to them display ads that are personally relevant or consistent with their usage goals. Such an approach could be effective in educing positive behavioural responses and minimizing avoidance behaviours. Fourth, researchers recognise that advertising by itself independent of the advertised product or service brand can and should be valuable to consumers (e.g. Cunningham & Haley, 2000). In spite of this, product peculiarities have also been known to influence the effectiveness of 242 University of Ghana http://ugspace.ug.edu.gh online ads (Flore et al., 2014; Eshghi et al., 2017). While our findings lend credence to the former position, they also provide empirical evidence to the fact that, (1) ad acceptance is higher for ODAs featuring product brands relative to those featuring service brands and ad avoidance is higher for service-featured ODAs compared to product-featured ODAs; (2) personalisation, interactivity and exposure condition were significant influencers of active acceptance for product-featured ODAs while interactivity was the only significant influencer of active acceptance for service-featured ODAS; (3) personalistion and exposure condition were the only negative influencers of active ad avoidance in the case of product- featured ads and service-featured ads respectively. Against these non-hypothesised evidences, it therefore makes good business sense for marketers and publishers to promote product brands using personalised and interactive display ads in voluntary exposure modes, in order to enhance acceptance behaviours among consumers. It similarly behoves firms operating in service sectors or marketers of service brands to focus on interactivity and exposure condition as relevant features to elicit acceptance as well as dull avoidance behaviours of consumers toward their display ads. Basically, in order to stimulate the desired behavioural responses, advertsisers must consider their market offering (product or services) when making choices regrding the ODA characteristics to blend. This is especially important for service brand marketers since avoidance was found to be generally higher among consumers exposed to service-featured ODAs. These implications emphasise the significance of prudently controlling and coordinating the design and types of display ads presented to consumers, more especially in terms of providing two way communications, imbuing ads with relevant information, matching ads to consumers needs and preferences and making sure consumers have control over their 243 University of Ghana http://ugspace.ug.edu.gh viewing experience and ads do not interfere with their online activities. Over and above these ad-related issues, practitioner attention should focus on consumers attitude toward online advertising and their internet usage motives as pertinent consumer-related factors, and the firms market offering as conditions under which online display ads with the above features may be more effective. 9.5 CONCLUSIONS A number of conclusions may be drawn from this research since results support the core proposition of the study that, the role played by both ad-related and consumer-related factors are very vital to the design of online display ads in influencing the behavioural responses of consumers. The current study shows that internal consumer processes can act as intermediate variables in the relationship between ODA characteristics and behavioural responses of consumers toward display ad. Thus, the study particularly submits that the deployment of display ads designed to be informative, personalised, interactive and presented in unforced exposure modes may elicit acceptance behaviours and reduce avoidance behaviours through consumers’ attitude toward online advertising as a facilitating factor. By springing on the theoretical insights derived from the SOR model (Mehrabian & Russell, 1974), this study supports and mirrors the position of Rodgers and Thorson (2000) that both “advertiser-controlled” and ‘consumer-controlled” factors are central to internet users’ perceptions and processing of online ads as well as advertising outcomes. In light of these, it seems logical to argue that reliance on ODA characteristics or design elements although may be quite adequate in eliciting positive behavioural responses, may not be sufficient in lessening avoidance behaviours toward display ads; rather how these ODA stimuli generate positive consumer ATOA is more crucial. 244 University of Ghana http://ugspace.ug.edu.gh The study further asserts that, given the role user mode underpinned by the reversal theory is found to play in these relationships as depicted in the study findings and post-study framework, understanding consumers’ motive for internet usage, becomes crucial for advertisers and publishers if they hope to appeal to the various user groups with the appropriate display ads imbued with the right features to generate the required attitude and responses. It has also been established from this study that an understanding of the product/service brands advertisers seek to promote is required if the appropriate ad features are to be selected in designing display ads that may suit such brands. In other words, the type or nature of brand (product or service) can determine the ODA characteristics that may be effective for an advertiser or firm. Finally, this research contends from a theoretical and empirical standpoint that, the effectiveness of online display ads measured in terms of consumer behavioural responses, is dependent on marketers or advertisers’ ability to blend the right features which need to be well-identified, harnessed, and deployed in a way that enhances approach behaviours of consumers and lessens avoidance behaviours through generating positive consumer attitude toward online advertising, as well as identifying consumers’ internet usage motive and giving consideration to the product/service brand being advertised. In precis, it is the conclusion of this research that the SOR model in consort with the reversal theory provide a useful framework for understanding consumer behavioural responses in the online advertising environment; and points to ODA characteristics together with attitude toward online advertising, user moder (consumers’ internet usage motive) as relevant research variables with direct implications for online advertising practice, and so does the nature of the brand (product or service). 9.6 RESEARCH LIMITATIONS AND AVENUES FOR FUTURE RESEARCH As with any academic research work, although this study makes useful contributions to the online advertising literature by providing relevant insights into online display advertising, 245 University of Ghana http://ugspace.ug.edu.gh these contributions should be viewed in light of the limitations amid which the study was conducted thus, pointing out avenues to be addressed by future research. To begin with, conceptually, the variables examined in the research model are vital to the current study’s aims and objectives and so are considered illustrative. They, however, are not exhaustive because other likely ad-related and consumer-related factors could explain and influence the interrelationships assessed in this study. First, the thesis in responding to the gaps identified in literature, focused on five ODA characteristics -interactivity, placement, personalisation, informativeness and exposure condition. Yet given the contended contextual originality of the study, other features such as entertainment, animation, and size though have been considerably studied, could have been incorporated into the model in order to compare results obtained with extant findings in the literature. This is even more so relevant since internet penetration and online advertising uptake in Ghana as mentioned early on (see chapter one) are progressively picking up. Moreover, several of these factors synchronously (work together to) influence online advertising outcomes. As such, research efforts that extend the study model to include these additional constructs as well as examine their interaction effects, can notably progress understanding of their influence from a less advanced technological context and offer comparable insights for online advertising strategies and design. Second, this study conceptualised user mode as two distinct groups of internet users. However, the reversal theory also suggests that user mode of internet users reverses during an online session. That is to say internet users’ metamotivational state can alter from paratelic to telic states or the other way around during their online activities (Mehrabian & Russell, 1974; Jung et al., 2014). Demonstrating this reversal effect was outside the remits of this current study owing to the research design and methodological approaches employed. As a viable area for further research, future studies on online display advertising through 246 University of Ghana http://ugspace.ug.edu.gh experimental designs could observe and validate the reversal of metamotivational or user mode over a controlled time period. Moreover, user mode was used as a consumer-related moderator in this study, but the influence of ODA characteristics on consumers’ ATOA and behavioural responses could be moderated by context-related factors as well. Auschaitraku and Mucherjee (2017) in their recent study establish that website types influence the effectiveness of online display advertising, and as found in this current study, consumers are exposed to ODA on various types of websites; informational websites (e.g. blogs and news sites), social media site (e.g. Facebook, Twitter, Instagram, YouTube), commercial website (e.g. Amazon, Tonaton, Jumia, OLX, Kaymu) and search engine (e.g. Google Adwords). Whether the effect of ODA characteristics on consumers behavioural responses and ATOA varies across these websites is an interesting question for future research to explore. Third, future research could also examine alternative mediating variables. This study informed by the SOR model conceptualised organism response using attititude toward online advertising (ATOA) which is an affective state. Findings from this current study provided evidence to support the significant mediational role of ATOA, which was examined from an explicit unidimensional perspective. However, it has been argued in quite recent studies (e.g. Goodrich 2011) that the existence of dual attitudes toward online advertising is possible; an implicit attitude formed without conscious awareness and control, and an explicit attitude based on intentionally generated evaluations which can be self reported. Because implicit attitudes are said to be distinct from their explicit equivalents on the basis of their formation, storage, retrieval and operations (e.g. Serenko & Turel, 2018), future research attention in the online advertising literature should be given to studying implicit attitude as a major attitudinal dimension or organism response that may predict behavioural responses, and not explicit attitude alone. This focus on dual attitudes or 247 University of Ghana http://ugspace.ug.edu.gh implicit attitudes is even more so warranted because researchers (e.g. Cowley, 2007) arguing within the framework of the SOR model have pointed out that there may be inaccuracies in consumers recall and report of internal responses. Since implicit unlike explicit attitudes are examined using implicit tests and not self-reports, examining the differential internvening role of dual attitudes is a research venture worth the efforts of future studies. Still speaking from a mediational standpoint, the online environment as argued by some authors (e.g. Demangeot & Broderick, 2007; Rosen & Purinton, 2004 ) is a “cognitive landscape” typified by stimuli that that can be processed cognitively, and this provides another avenue for the examination of the possibility that consumers’ behaviours to online display advertising could be intervened by internal cognitive responses. Particularly, consumers’ approach and avoidance behaviours toward ODA may be further explained by examining the intermediate role of such cognitive factors as advertising recognition, recall and memorisation. Future studies could therefore investigate these variables from the perspective of the SOR model and/or through cognition-primed models. It would also be instructive for future studies on online display advertising to consider the bi-lateral view of the organism response to extend the model presented in the current study by incorporating a cognitively mediated response. Fourth, attempts made by this study to investigate the differences in behavioural responses on the basis of the nature of the advertised brands featured in the ODAs were at best exploratory or tentative. Generally, findings show significant differences in acceptance and avoidance behaviours for product-featured ODAs and service-featured ODAs as well as differences in which ODA characteristics are likely to influence active acceptance and avoidance behaviours for the two types of ODAs. Since the study focused on the broad 248 University of Ghana http://ugspace.ug.edu.gh product-service dichotonomy, it would be enlightening for future research to extend the current study by investigating possible variations that may exist in ODA effects for the different service processing categories: people (e.g. healthcare, hospitality), possession (e.g. repair or maintenance), mental stimulus (e.g. education), and Information (e.g. banking, insurance) as has been done for physical product categories by prior studies (e.g. Flore et al., 2014; Eshghi et al., 2017; Belanche et al., 2017). This current study also bears a few methodological limitations that are worth mentioning. The study was a cross-sectional survey-based research which assessed consumers’ behavioural responses using self-reported measures. The advantage of survey-based designs is that they have acceptable external validity as they are centred on real occurences experienced by the consumers (Lewis & Rao, 2015) over time. The survey design also allowed the study to rule out the eccentric effect that emanates from using one ad for a particular brand of a specific product category. Nonetheless, researchers (e.g. Wang & Minor 2008) argue that cross-sectional studies are limited in their ability to determine true causality. Also, the self-reported measurement of behavioural responses limited the identification of “subtle or unconscious” behaviours of which consumers may be unaware. Although the study attempted to curb this drawback and provide a more refined view of approach and avoidance behaviours by examining both the intensity (active and passive) and dire direction (ad acceptance and avoidance) of behavioural responses, only the behavioural direction was determined at the exploration and model refinement stage as relevant in the current study setting. As such future studies could tackle these shortcomings by employing scenario-based methods or field experiments and alternative data collection approaches such as secondary/clickstream data or by using longitudinal designs which would offer significant additional insights. Also, the study data was obtained from a relatively large respondent sample (n=592) using purposive and snowball sampling which 249 University of Ghana http://ugspace.ug.edu.gh have elements of convenience. It is therefore, acknowledged that the study does not encompass all internet users. While the sampling techniques used are appropriate approaches for studies that purport to test theory as was done in this study, and the theoretical as well as methodological approaches to reduce this limitation were applied, it is still prudent that caution is exercised when generalizing the results of the research to all internet users. Lastly with the contextual disparities identified in extant ODA literature, prospects exist for future research to replicate the theory-based research framework presented in this study along side other avenues pointed out in settings with similar economic and consumer characterics as well as other emerging markets or developing economies with relatively higher internet penetration and online advertising uptake. It is the contention of this study that evidence from these other economic settings as well as cross-economy studies will provide considerable amount of data, added perspectives, and a broad base for comparisons toward developing theoretically grounded insights for improving online advertising strategies and practice. 250 University of Ghana http://ugspace.ug.edu.gh REFERENCES Abend, G. (2008). The meaning of theory. Sociological Theory, 26(2), 173–199. Africa News (June, 2018). Digital in 2018: Africa’s internet users increase by 20%. Retrieved, September 20, 2018, from http://www.africanews.com/2018/02/06/ digital-in-2018-africa-s-internet-users-increase-by-20-percent// Agarwal, A., Hosanagar, K., & Smith, M. D. (2011). Location, location, location: An analysis of profitability of position in online advertising markets. Journal of marketing research, 48(6), 1057-1073. Aguinis, H., Edwards, J. R., & Bradley, K. J. (2016). Improving our understanding of moderation and mediation in strategic management research. Organizational Research Methods, 20(4), 665-685. Aguirre, E., Mahr, D., Grewal, D., de Ruyter, K., & Wetzels, M. (2015). Unraveling the personalization paradox: The effect of information collection and trust-building strategies on online advertisement effectiveness. Journal of Retailing, 91(1), 34-49. Ajzen, I. (2001). Nature and operation of attitudes. Annual review of psychology, 52(1), 27- 58. Aliyu, A. A., Muhammad, U. B., Rozilah, K., & David, M. (2014). Positivist and non- positivist paradigm in social science research: conflicting paradigms or perfect partners? Journal of Management and Sustainability, 4(3), 79-95. Anderson, J. Y., & Gerbing, D. W. (1988). Structural equation modelling in practice: a review and recommended two-step approach. Psychological Bulletin, 103(3), 411- 423. Anees-ur-Rehman, M., Wong, H. Y., & Hossain, M. (2016). The progression of brand orientation literature in twenty years: a systematic literature review. Journal of brand management, 23(6), 612-630. Apter, M. J. (1984). Reversal theory and personality: A Review. Journal of Research in Personality, 18(3), 265–88. Apter, M. J. (1992). The Dangerous Edge. New York, NY: Free Press. Apter, M. J. (2001) Motivational styles in everyday life: A guide to reversal theory. Washington, D.C.: American Psychological Association. Apter, M. J. (2007). Reversal theory: The dynamics of motivation, emotion, and personality (2nd ed.). Oxford, England: Oneworld. Auschaitrakul, S., & Mukherjee, A. (2017). Online Display Advertising: The Influence of Web Site Type on Advertising Effectiveness. Psychology & Marketing, 34(4), 463- 480. 251 University of Ghana http://ugspace.ug.edu.gh Babbie, E. R. (2004). The Practice of Social Research (10th ed.). Belmont, CA: Wadsworth Thomson Learning. Baek, T. H., & Morimoto, M. (2012). Stay away from me. Examining the determinants of consumer avoidance of personalized advertising. Journal of advertising, 41(1), 59- 76. Bagozzi, R. P. (1986). Principles of marketing management. Chicago: Science Research Associates. Bagozzi, R. P., & Yi, Y. (2012). Specification, evaluation, and interpretation of structural equation models. Journal of the Academy of Marketing Science, 40(1), 8-34. Bakshi, A. C. (2015). Why and how to regulate native advertisement in online news publications. Journal of Media Law & Ethics, 4(3/4), 4-27. Barnes, S. J. (2002). Wireless digital advertising: nature and implications. International journal of advertising, 21(3), 399-420. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173-1182. Baron, S. D., Brouwer, C., & Garbayo, A. (2014). A model for delivering branding value through high-impact digital advertising. Journal of Advertising Research, 54(3), 286-291. Bartlett, M. S. (1954) A note on the multiplying factors for various chi square approximation. Journal of the Royal Statistical Society, 16(Series B), 296-298. Becker-Olsen, K. L. (2003). And now, a word from our sponsor--a look at the effects of sponsored content and banner advertising. Journal of Advertising, 32(2), 17-32. Belanche, D., Flavián, C., & Pérez-Rueda, A. (2017). Understanding Interactive Online Advertising: Congruence and Product Involvement in Highly and Lowly Arousing, Skippable Video Ads. Journal of Interactive Marketing, 37, 75-88. Bellman, S., Kemp, A. Haddad, H., & Varan, D. (2014). The effectiveness of advergames compared to television commercials and interactive commercials featuring advergames. Computers in Human Behaviour, 32, 276-283. Bentler, P.M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238-246. Benton, J. (2014). Like it or not, native advertising is squarely inside the big news tent. Retrieved 15, December 2017, from http://www.niemanlab.org/2014/09/like-it-or- not-nativeadvertising-is-squarely-inside-the-big- news-tent Beverland, M., & Lindgreen, A. (2010). What makes a good case study? A positivist review of qualitative case research published in Industrial Marketing Management, 1971– 2006. Industrial Marketing Management, 39(1), 56-63. 252 University of Ghana http://ugspace.ug.edu.gh Blaikie, N. W. H. (2010) Designing Social Research (2nd ed.). Campbridge, UK: Sage Publications. Blaydon, M. J., Lindner, K. J., & Kerr, J. H. (2004). Metamotivational characteristics of exercise dependence and eating disorders in highly active amateur sport participants. Personality and Individual Differences, 36(6), 1419–1432. Bleier, A., & Eisenbeiss, M. (2015a). Personalized online advertising effectiveness: The interplay of what, when, and where. Marketing Science, 34(5), 669-688. Bleier, A., & Eisenbeiss, M. (2015b). The importance of trust for personalized online advertising. Journal of Retailing, 91(3), 390-409. Boone, G., Secci, J., & Gallant, L. (2010). Emerging trends in online advertising. Doxa communicacion, 5(2), 241-253. Bornmann, L. (2011). Scientific peer review. Annual review of information science and technology, 45(1), 197-245. Boso, N., Story, V. M. & Cadogan, J. W. (2013). Entrepreneurial orientation, market orientation, network ties, and performance: Study of entrepreneurial firms in a developing economy. Journal of Business Venturing, 28, 708-727. Botha, E., & Van der Waldt, D. L. R. (2011). Relationship outcomes as measurement criteria to assist communication strategists to manage organizational relationships. Innovar, 21(40), 5-16. Boyd, B. K., Haynes, K. T., Hitt, M. A., Bergh, D. D., & Ketchen, D. J. (2012). Contingency hypotheses in strategic management research: Use, disuse, or misuse? Journal of Management, 38(1), 278-313. Brackett, L. K., & Carr, B. N. (2001). Cyberspace advertising vs. other media: Consumer vs. mature student attitudes. Journal of advertising research, 41(5), 23-32. Brajnik, G., & Gabrielli, S. (2010). A review of online advertising effects on the user experience. International Journal of Human-Computer Interaction, 26(10), 971– 997. Brehm, J. W. (1966). A theory of psychological reactance. New York: Academic Press. Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: a theory of freedom and control. New York: Academic press. Brettel, M., & Spilker-Attig, A. (2010). Online advertising effectiveness: a cross-cultural comparison. Journal of research in interactive marketing, 4(3), 176-196. Breuer, R., & Brettel, M. (2012). Short-and long-term effects of online advertising: Differences between new and existing customers. Journal of Interactive Marketing, 26(3), 155-166. Breuer, R., & Brettel, M. (2017). Time lags and synergies of online advertising. In C. L. Campbell (Ed.), The customer is NOT always right? Marketing orientations in a 253 University of Ghana http://ugspace.ug.edu.gh dynamic business world. Developments in Marketing Science. Proceedings of the Academy of Marketing Science. Springer, Cham. Breuer, R., Brettel, M., & Engelen, A. (2011). Incorporating long-term effects in determining the effectiveness of different types of online advertising. Marketing Letters, 22(4), 327-340. Briggs, R., & Hollis, N. (1997). Advertising on the web: Is there response before click- through?” Journal of Advertising Research, 37(2), 33–45. Bright, L. F., & Daugherty, T. (2012). Does customization impact advertising effectiveness? An exploratory study of consumer perceptions of advertising in customized Online Environments. Journal of Marketing Communications, 18(1), 19-37. Bruce, N. I., Murthi, B. P. S., & Rao, R. C. (2017). A dynamic model for digital advertising: The effects of creative format, message content, and targeting on engagement. Journal of marketing research, 54(2), 202-218. Bryman, A. (2001). Social Research Methods. Oxford: Oxford University Press. Bryman, A., & Bell, E. (2011). Business Research Methods (3rd Ed.). New York: Oxford University Press Inc. Bryman, A., & Bell, E. (2015). Business Research Methods. Oxford: Oxford university press. Burns, K. S., & Lutz, R. J. (2006). The function of format: Consumer responses to six on- line advertising formats. Journal of Advertising, 35(1), 53-63. Burns, K. S., & Lutz, R. J. (2008). Web users' perceptions of and attitudes toward online advertising formats. International Journal of Internet Marketing and Advertising, 4(4), 281-301. Burns, R. P., & Burns, R. (2008). Business research methods and statistics using SPSS. London: Sage Publication. Byrne, B. M. (2004). Testing multigroup invariance using AMOS graphics: A road less travelled. Structural Equation Modelling, 11(2), 272-300. Byrne, B. M. (2010). Structural equation modeling with AMOS (2nd ed.). New York: Routledge. Byrne, B. M. (2013). Structural equation modelling with AMOS: Basic concepts, applications, and programming. New York: Routledge. Cacioppo, J. T., & Petty, R. E. (1984). The elaboration likelihood model of persuasion. In C. K. Thomas (Ed.), Advances in Consumer Research. (pp.673-675) Provo: UT: Association for Consumer Research. 254 University of Ghana http://ugspace.ug.edu.gh Cai, Y., & Mehari, Y. (2015). The use of institutional theory in higher education research. In Theory and Method in Higher Education Research (pp. 1-25). Emerald Group Publishing Limited. Campbell, C., Mattison Thompson, F., Grimm, P. E., & Robson, K. (2017). Understanding why consumers don't skip pre-roll video ads. Journal of Advertising, 46(3), 411-423. Campbell, D. E., & Wright, R. T., Clay, P. F. (2010). Deconstruction and operationalizing interactivity: An online advertising. Journal of Information technology Theory and Application, 14(4), 29-53. Campbell, D. E., & Wright, R. T. (2008). Shut-up I don't care: understanding the role of relevance and interactivity on customer attitudes toward repetitive online advertising. Journal of Electronic Commerce Research, 9(1), 62-76. Carter, S. (2005). Reversal theory: changing you and your organization for the better. Training Journal, 2(x), 20–22. *Chan, J. C., Jiang, Z., & Tan, B. C. (2010). Understanding online interruption-based advertising: Impacts of exposure timing, advertising intent, and brand image. IEEE Transactions on Engineering Management, 57(3), 365-379. Chan, T. K., Cheung, C. M., & Lee, Z. W. (2017). The state of online impulse-buying research: A literature analysis. Information & Management, 54(2), 204-217. Chang, C. (2017). Methodological Issues in Advertising Research: Current Status, Shifts, and Trends. Journal of Advertising, 46(1), 2-20. Chang, H. H., & Chen, S. W. (2008). The impact of online store environment cues on purchase intention: Trust and perceived risk as a mediator. Online Information Review, 32(6), 818-841. Chang, H. J., Eckman, M., & Yan, R. N. (2011). Application of the Stimulus-Organism- Response model to the retail environment: the role of hedonic motivation in impulse buying behavior. The International Review of Retail, Distribution and Consumer Research, 21(3), 233-249. Chang, H.H., & Chen, S.W. (2008). The impact of online store environment cues on purchase intentions: Trust and perceived risk as a mediator. Online information review, 32(6), 818–841. Chang, Y.P. and Zhu, D.H. (2011) Understanding social networking sites adoption in China: A comparison of pre-adoption and post-adaption. Computers in Human Behavior, 27(5), 1840-1848. *Chapelle, O., Manavoglu, E., & Rosales, R. (2015). Simple and scalable response prediction for display advertising. ACM Transactions on Intelligent Systems and Technology (TIST), 5(4), 61. 255 University of Ghana http://ugspace.ug.edu.gh Chatterjee, P. (2008). Are unclicked ads wasted? Enduring effects of banner and pop-up ad exposures on brand memory and attitude. Journal of Electronic Commerce Research 9(1), 51-61. Chen, C. C., & Yao, J. Y. (2018). What drives impulse buying behaviours in a mobile auction? The perspective of the Stimulus-Organism-Response model. Telematics and Informatics, 35(5), 1249-1262. Ching, R. K., Tong, P., Chen, J. S., & Chen, H. Y. (2013). Narrative online advertising: identification and its effects on attitude toward a product. Internet Research, 23(4), 414-438. Cho, C-H. (2003). The effectiveness of banner advertisements: Involvement and click- through. Journalism and Mass Communication Quarterly, 80(3), 623-645. Cho, C-H., & Cheon, H. J. (2004). Why do people avoid advertising on the Internet? Journal of Advertising, 33(4), 89–97. Cho, C.-H., & Khang., H. (2006). The state of Internet-related research in communications, marketing, and advertising: 1994-2003. Journal of Advertising, 35(3), 143-63. Churchill, G. A., & Iacobucci, D. (2009). Marketing research: Methodological foundations. Mason: Cenage Learning. Clark, R., Ezell, J., Clark, J., & Sheffield, D. N. (2009). Stay or leave: Applying approach- avoidance theory to virtual environments. Journal of Database Marketing and Customer Strategy Management, 16(4), 231–240. Companiesandmarkets.com (2014) Online advertising—Global strategic business report. Retrieved, January 25, 2016, from http://www.companiesandmarkets.com/Market/ Media/Market-Research/Online-AdvertisingGlobalStrategic-Business- Report/RPT1268916. Cooper, D. R., & Schindler, P. S. (2006). Marketing research. New York: McGraw-Hill. Cooper, D. R., Schindler, P. S., & Sun, J. (2006). Business Research Methods. New York: McGraw-Hill. Corley, K. G., & Gioia, D. A. (2011). Building theory about theory building: What constitutes a theoretical contribution? Academy of Management Review, 36(1), 12– 32. Cowley, E. (2007). How enjoyable was it? Remembering an affective reaction to a previous consumption experience. Journal of Consumer Research, 34(4), 494–505. Craighead, C. W., Ketchen, D. J., Dunn, K. S., & Hult, G. T. (2011). Addressing common method variance: Guidelines for survey research on information technology, operations, and supply chain management. IEEE Transactions on Engineering Management, 58(3), 578-588. Creswell, J. W. (2014). Research Design: Qualitative, Quantitative and Mixed Approaches. Thousand Oaks, CA: Sage Publications. 256 University of Ghana http://ugspace.ug.edu.gh Creswell, J. W. & Clark, V. L. P. (2010). Designing and conducting mixed methods research. Thousand Oaks, C.A.: Sage. Creswell, J. W., & Clark, V. L. P. (2007). Designing and Conducting Mixed Methods Research. California: Sage Publications. Crossan, M. M., Lane, H. W., & White, R. E. (1999). An organizational learning framework: From intuition to institution. Academy of management review, 24(3), 522-537. Crotty, M. (2003). The Foundations of Social Research: Meaning and Perspective in the Research Process (3rd ed.). London: Sage Publications. Citifmonline (2016). Biggest telecom merger in the offing as Airtel engages Tigo for a deal. Retrieved, May 23, 2018 from http://citifmonline.com/2017/01/biggest-telecom- merger-in-the-offing-as-airtel-engages-tigo-for-a-deal/ Cunningham, A., & Haley, E. (2000). A look inside the world of advertising-free publishing: A case study of Ms. Magazine. Journal of Current Issues & Research in Advertising, 22(2), 17-30. Dahlen, M., & Rosengren, S. (2016). If advertising won't die, what will it be? Toward a working definition of advertising. Journal of Advertising, 45(3), 334-345. Dahlen, M., Rosengren, S., Torn, F. & Ohman, N. (2008). Could placing ads wrong be right? Advertising effects of thematic incongruence. Journal of Advertising, 37(3), 57–67. Davis, J. C. (2007). A conceptual view of branding for services. Innovative Marketing, 3(1), 7- 13. Davis, R. (2009). Do consumers experience a reversal state when encountering mobile commerce services? Proceedings of 2009 ANZMAC. Australia: Melbourne. Retrieved March 30, 2011, from www.duplication.net.au/ANZMAC09/papers/ ANZMAC2009-110.pdf De Mooij, M., & Hofstede, G. (2010). The Hofstede model: applications to global branding and advertising strategy and research. International Journal of Advertising, 29(1), 85-110. de Pelsmacker, P., &Neijens, P. C. (2012). New advertising formats: How persuasion knowledge affects consumer responses. Journal of Marketing Communications, 18(1), 1-4. de Pelsmacker, P., Geuens, M., & Van Den Bergh, J. (2004). Marketing communications: A European perspective. UK: Pearson Education Limited. Demangeot, C., & Broderick, A. J. (2007) Conceptualising consumer behaviour in online shopping environments. International Journal of Retail & Distribution Management, 35(11), 878-894. 257 University of Ghana http://ugspace.ug.edu.gh Denscombe, M. (2008). Communities of practice a research paradigm for the mixed methods approach. Journal of Mixed Methods Research, 2(3), 270-283. Devellis, R. F. (2003). Scale development: Theory and applications (2nd ed.). Newbury Park, CA: Sage Publications. Digital Content Africa (October, 2017). Africa: Ghana's digital advertising industry begins to take off - big international brands take online and mobile seriously. Retrieved, May 25, 2018 from http://allafrica.com/stories/201710270788.html Dong, P. & Siu, N. Y. (2013). Servicescape elements, customer predispositions and service experience: the case of theme park visitors. Tourism Management, 36, 541-531. Donovan, R. J., & Rossiter, J. R. (1982). Store atmosphere: An environmental psychology approach. Journal of Retailing, 58(1), 34-57. Donovan, R. J., Rossiter, J. R., Marcoolyn, G., & Nesdale, A. (1994). Store atmosphere and purchasing behavior. Journal of Retailing, 70(3), 283-294. Draganska, M., Hartmann, W. R., & Stanglein, G. (2014). Internet versus television advertising: A brand-building comparison. Journal of Marketing Research, 51, 578– 590. *Drossos, D. A., Fouskas, K. G., Kokkinaki, F., & Papakyriakopoulos, D. (2011). Advertising on the internet: perceptions of advertising agencies and marketing managers. International Journal of Internet Marketing and Advertising, 6(3), 244- 264. Duarte, P. A. O., & Raposo, M. L. B. (2010). A PLS model to study brand preference: An application to the mobile phone market. In Handbook of partial least squares (pp. 449- 485). Springer, Berlin, Heidelberg. Ducoffe, R. H. (1996a). Advertising value and advertising on the web. Journal of advertising research, 21-35. Ducoffe, R. H. (1996b). How consumers assess the value of advertising. Journal of Current Issues and Research in Advertising, 17(1), 1-18. Duff, B. R. L., & Faber, R. J. (2011). Missing the mark. Journal of Advertising, 40(2), 51– 62. Easterby-Smith, M., Thorpe, R., & Jackson, P. R. (2012). Management Research. London: Sage Publication. Edwards, S. M., H. Li, J.-H. Lee. (2002). Forced exposure and psychological reactance: Antecedents and consequences of the perceived intrusiveness of pop-up ads. Journal of Advertising, 31(3), 83–95. Eldabi, T., Irani, T., Paul, R. J., & Love, P. E. D. (2000). Quantitative and qualitative decision-making methods in simulation modelling. Management Decision, 40(1), 64-73. 258 University of Ghana http://ugspace.ug.edu.gh eMarketer (2018) Digital ad spending worldwide, by region, 2018 (billions) [charts and tables]. Retrieved, April 10, 2018, from https://www.emarketer.com/Chart/Digital- Ad-Spending-Worldwide-by-Region-2018-billions/217835. Eroglu, S. A., Machleit, K. A., & Davis, L. M. (2001). Atmospheric qualities of online retailing: A conceptual model and implications. Journal of Business Research, 54(2), 177-184. Eroglu, S. A., Machleit, K. A., & Davis, L. M. (2003). Empirical testing of a model of online store atmospherics and shopper responses. Psychology and Marketing, 20(2), 139- 150. Eshghi, A., Sarkar, J. G., & Sarkar, A. (2017). Impact of online advertising on adolescent’s brand attitudes. Marketing Intelligence & Planning, 35(6), 706-723. Faber, R. J. (2015). Peeking under the curtain and over the horizon: The reflections of another former editor. Journal of Advertising, 44(3), 289–295. Field, A. (2013). Discovering Statistics Using IBM SPSS Statistics (4th Ed.). London: Sage Publications. Fink, A. (2009). How to Conduct Surveys (4th ed.). Thousand Oaks, CA: Sage Publication. Florenthal, B., & Shoham, A. (2010). Four-mode channel interactivity concept and channel preferences. Journal of Service Marketing, 24(1), 29-41. *Flores, W., Chen, J. C. V., & Ross, W. H. (2014). The effect of variations in banner ad, type of product, website context, and language of advertising on Internet users’ attitudes. Computers in Human Behavior, 31(1), 37-47. Fornell, C., & Larcker, D.F. (1981). Evaluation Structural Equation Models with Unobservable Variables and Measurement Error, Journal of Marketing Research, 18(1): 39-50. Frankfort-Nachmias, C., & Nachmias, D. (1996). Research Methods in the Social Sciences (5th ed.). London: Arnold. Fransen, M. L., Verlegh, P. W. J., Kirmani, A., & Smit, E. G. (2015). A typology of consumer strategies for resisting advertising, and a review of mechanisms for countering them. International Journal of Advertising, 34(1), 6–16. *Fridgeirsdottir, K., & Najafi-Asadolahi, S. (2018). Cost-per-impression pricing for display advertising. Operations Research, 66(3), 653-672. Frith, K. T. (1995). Advertising and Mother Nature. In A. N. Valdivia (Ed.), Feminism, multiculturalism, and the media: Global diversities (pp. 185–196). Thousand Oaks, CA: Sage Publications. Fruchart, E., Rulence-Pâques, P., & Mullet, E. (2018). Watching high-risk sports on television: the reversal theory’s concept of protective frame. Sport in Society. Retrieved, January 5, 2019, from https://doi.org/10.1080/17430437.2018.1487404 259 University of Ghana http://ugspace.ug.edu.gh Fuchs, C., & Horak, E. (2008). Africa and the digital divide. Telematics and informatics, 25(2), 99-116. *Fulgoni, G. M., & Mörn, M. P. (2009). Whither the click? How online advertising works. Journal of Advertising Research, 49(2), 134-142. Fuller, C. M., Simmering, M. J., Atnic, G., Atnic, Y., & Babin, B. J. (2016). Common methods variance detection in business research. Journal of Business Research, 69(8), 3192-3198. Furnham, A., Jenny, B., & Barrie, G. (2002). Memory for television advertisements as a function of advertisement–programme congruity. Applied Cognitive Psychology, 16(5), 525–45. Gao, Q., Rau, P. L. P., & Salvendy, G. (2009). Perception of interactivity: Affects of four key variables in mobile advertising. International Journal of Human-Computer Interaction, 25(6), 479-505. Ghauri, P., & Gronhaug, K. (2005). Research methods in business studies: a practical guide (3rd ed.). Essex: Prentice Hall/Financial times. Gill, J., & Johnson, P. (2010). Research Methods for Managers. Thousand Oaks, CA: Sage Publications. Gioia, D. A., & Pitre, E. (1990). Multiparadigm perspectives on theory building. Academy of Management Review, 15(4), 584–602. Goh, K. Y., & Ping, J. W. (2014). Engaging consumers with advergames: An experimental evaluation of interactivity, fit and expectancy. Journal of the Association for Information Systems, 15(7), 388-421. Goldfarb, A. (2014). What is different about online advertising? Review of Industrial Organization, 44(2), 115-129. *Goldfarb, A., & Tucker, C. E. (2011a). Online display advertising: Targeting and obtrusiveness. Marketing Science, 30(3), 389-404. Goldstein, D. G., Suri, S., McAfee, R. P., Ekstrand-Abueg, M., & Diaz, F. (2014). The economic and cognitive costs of annoying display advertisements. Journal of Marketing Research, 51(6), 742–752. Goodrich, K. (2010). What’s up? Exploring upper and lower visual field advertising effects. Journal of Advertising Research, 50(1), 91–106. Goodrich, K. (2011). Anarchy of effects? Exploring attention to online advertising and multiple outcomes. Psychology & marketing, 28(4), 417-440. Goodrich, K. (2013). Effects of age and time of day on Internet advertising outcomes. Journal of Marketing Communications, 19(4), 229-244. 260 University of Ghana http://ugspace.ug.edu.gh Goodrich, K. (2014). The gender gap: Brain-processing differences between the sexes shape attitudes about online advertising. Journal of Advertising Research, 54(1), 32-43. Goodrich, K., Schiller, S. Z., & Galletta, D. (2015). Consumer Reactions to Intrusiveness of Online-Video Advertisements. Journal of Advertising Research, 55(1), 37-50. Grant, C., & Osanloo, A. (2014). Understanding, selecting, and integrating a theoretical framework in dissertation research: Creating the blueprint for your “house”. Administrative Issues Journal, 4(2), 12-26. Graphic Online (February, 2018). Over 10 million Ghanaians use the Internet – Report. Retrieved , September 20, 2018, from https://www.graphic.com.gh/news/general- news/over-10-million-ghanaians-using-the-internet-report.html Green, N. (2008). Formulating and refining a research question. In N. Gilbert (Ed.), Researching Social Life (3rd ed.) (pp.43-61). London: Sage Publications. Ha, L. (2008). Online advertising research in advertising journals: A review. Journal of Current Issues and Research in Advertising, 30(1), 31-48. Ha, L., & McCann, K. (2008). An integrated model of advertising clutter in offline and online media. International Journal of Advertising, 27(4), 569-592. Hair Jr., J. F., Black, W. C., Babin, B. J., Anderson, R. E. & Tatham, R. L. (2006). Multivariate Data Analysis. New Jersey: Pearson. Hair, J. F., Black, W. C., Barbin, B. J., & Anderson, R. E. (2010). Multivariate Data Analysis (7th ed.). Upper Saddle River, NJ: Printice Hall. Hair, J. F., Ringle, C. M., & Sarstedt, M. (2011). PLS-SEM: Indeed, a silver bullet. Journal of Marketing theory and Practice, 19(2), 139-152. Ham, C. D. (2017). Exploring how consumers cope with online behavioral advertising. International Journal of Advertising, 36(4), 632-658. Hambrick, D. C. (2007). The field of management’s devotion to theory: Too much of a good thing? Academy of Management Journal, 50(6),1346–1352. Hanafizadeh, P., Behboudi, M., Ahadi, F., & Ghaderi Varkani, F. (2012). Internet advertising adoption: a structural equation model for Iranian SMEs. Internet Research, 22(4), 499-526. Harker, D. (2008). Regulating online advertising: the benefit of qualitative insights. Qualitative Market Research: An International Journal, 11(3), 295-315. Hatch, M. J., & Cunliffe, A. L. (2006). Organization Theory (2nd ed.). Oxford: Oxford University Press. Hayes, A. F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Communication monographs, 76(4), 408-420. 261 University of Ghana http://ugspace.ug.edu.gh Hayes, A. F. (2015). An index and test of linear moderated mediation. Multivariate Behavioral Research, 50(1), 1-22. Healy, M., & Perry, C. (2000). Comprehensive criteria to judge validity and reliability of qualitative research within the realism paradigm. Qualitative Market Research: An International Journal, 3(3), 118-126. Heeks, R., & Bailur, S. (2007). Analyzing e-government research: perspectives, philosophies, theories, methods, and practice. Government Information Quarterly 24(2), 243-265. Hegner, S., Kusse, D., & Pruyn, A. H. (2015). Watch it! The influence of forced pre-roll video ads influence on consumer perceptions. In P.W.J. Verlegh, H.A.M. Voorveld, & M. Eisend (Eds.), Advances in advertising research: The digital, the classic, the subtle and the alternative (Volume VI). (pp.63-73). Berlin: Springer Wiesbaden. Heinberg, M., & Taube, M. (2015). Challenges and chances for International Portfolio Acquisition Brands (IPA Brands) in developing countries. In Marketing Dynamism & Sustainability: Things Change, Things Stay the Same… (pp. 474-474). Cham: Springer. Heiser, R. S., Sierra, J. J., & Torres, I. M. (2008). Creativity via cartoon spokespeople in print ads: Capitalizing on the distinctiveness effect. Journal of Advertising, 37(4), 75-84. Hinton, P. R. (2014). Statistics Explained (3rd ed.). East Sussex, UK: Routledge. *Hoban, P. R., & Bucklin, R. E. (2015). Effects of internet display advertising in the purchase funnel: Model-based insights from a randomized field experiment. Journal of Marketing Research, 52(3), 375–393. Hoezel, M. (2014). Spending on native advertising is soaring as marketers and digital media publishers realize the benefits. Business Insider. Retrieved, October 10, 2017 from http://uk.businessinsider.com/spending-on-native-ads-will-soar-as-publishers- andadvertisers-take-notice-2014-11?r=US&IR=T Hof, R. (2013). Display ads to eclipse search as mobile revenues take off. Retrieved, September 9, 2018, from https://www.forbes.com/sites/roberthof/2013/01/17/display- ads-to-eclipse-search-as-mobile-revenues-take-off/. Holbert, R. L., & Stephenson, M. T. (2003). The importance of indirect effects in media effects research: Testing for mediation in structural equation modeling. Journal of Broadcasting & Electronic Media, 47(4), 556-572. Hollis, N. (2005). Ten years of learning on how online advertising builds brands. Journal of Advertising Research, 45(2), 255–268. Hornik, R., Jacobsohn, L., Orwin, R., Piesse, A., & Kalton, G. (2008). Effects of the national youth anti-drug media campaign on youths. American Journal of Public Health, 98(12), 2229-2236. 262 University of Ghana http://ugspace.ug.edu.gh Hsieh, Y. C., & Chen, K. H. (2011). How different information types affect viewers’ attention on internet advertising. Computers in Human Behavior, 27(2), 935–945. Huang, M.-H. (2003). Modeling virtual exploratory and shopping dynamics: An environmental psychology approach. Information Management, 41(1), 39–47. Huck, S. (2012). Reading Statistics and Research (6th ed.). Boston, MA: Pearson. *Hussain, R., Ferdous, A. S., & Mort, G. S. (2018). Impact of web banner advertising frequency on attitude. Asia Pacific Journal of Marketing and Logistics, 30(2), 380- 399. Hussey, J. & Hussey, R. (1997). Business Research: A practical Guide for Undergraduate and Post-graduate Students. London: MacMillan Press Ltd. Iacobucci, D. (2010). Structural equations modeling: Fit indices, sample size, and advanced topics. Journal of consumer psychology, 20(1), 90-98. Institute of Statistical, Social and Economic Research, ISSER (2017). State of the Ghanaian Economy Report 2016. Accra: ISSER. Internet Advertising Bureau. (2018). Digital ad spend reaches an all-time high of $88 billion in 2017, with mobile upswing unabated, accounting for 57% of revenue. Retrieved, March 29, 2018, from https://www.iab.com/wp-content/uploads/2018/05/IAB- Internet-Advertising-Revenue-Report-FY-2017-Draft-Media-Alert-v9.pdf Internet World Stats (June, 2018). Internet usage statistics. The Internet big picture: World Internet users and 2018 population stats. Retrieved, March 29, 2018 from https://www.internetworldstats.com/stats.htm Internet World Stats (June, 2019). Internet users statistics for Africa. Retrieved, July 25, 2019 from https://www.internetworldstats.com/stats1.htm Islam, J. U., & Rahman, Z. (2017). The impact of online brand community characteristics on customer engagement: An application of Stimulus-Organism-Response paradigm. Telematics and Informatics, 34(4), 96-109. IT News Africa (September, 2015). Ghana’s top 12 ecommerce websites. Retrieve, September 20, 2018, from http://www.itnewsafrica.com/2015/09/ghanas-top-12- ecommerce-websites/ Jacoby, J. (2002). Stimulus-organism-response reconsidered: An evolutionary step in modeling (consumer) behavior. Journal of Consumer Psychology, 12(1), 51–57. Janiszewski, C. (1993). Preattentive mere exposure effects. Journal of Consumer Research, 20(3), 376–392. Janssens, W., De Pelsmacker, P., & Geuens, M. (2012). Online advertising and congruency effects: It depends on how you look at it. International Journal of Advertising, 31(3), 579-604. 263 University of Ghana http://ugspace.ug.edu.gh Jeong, S.W., Fiore, A.M., Niehm, L. S., & Lorenz, F. O. (2009). The role of experiential value in online shopping: The impacts of product presentation on consumer responses towards an apparel web site. Internet Research, 19(1), 105–124. Jeong, Y., & King, C. M. (2010). Impacts of website context relevance on banner advertisement effectiveness. Journal of Promotion Management, 16(3), 247–264. Johar, G. V. (2016). Mistaken inferences from advertising conversations: A modest research agenda. Journal of Advertising, 45(3), 318-325. Johnson, J. P. (2013). Targeted advertising and advertising avoidance. RAND Journal of Economics, 44 (1), 128–44. Johnson, R. B., Onwuegbuzie, A. J., & Turner, L. A. (2007). Toward a definition of mixed methods research. Journal of Mixed Methods Research, 1(2), 112-133. Jonathan J. Wright, J. J., Wright, S., Sadlo, G., & Stew, G. (2014) Exploring optimal experiences: A reversal theory perspective of flow and occupational science. Journal of Occupational Science, 21(2), 173-187. *Jung, J. M., Chu, H., Min, K. S., & Martin, D. (2014). Does telic/paratelic user mode matter on the effectiveness of interactive internet advertising? A reversal theory perspective. Journal of Business Research, 67(6), 1303-1309. *Jung, J. M., Min, K. S., & Kellaris, J. J. (2011). The games people play: How the entertainment value of online ads helps or harms persuasion. Psychology & Marketing, 28(7), 661–681. Kafka, P., & Molla, R. (2017). 2017 was the year digital ad spending finally beat TV. Retrieved, January 3, 2019, from https://www.recode.net/2017/12/4/16733460/2017 -digital-ad-spend-advertising-beat-tv. Kamboj, S., Sarmah, B., Gupta, S., & Dwivedi, Y. (2018). Examining branding co-creation in brand communities on social media: Applying the paradigm of Stimulus- Organism-Response. International Journal of Information Management, 39(1), 169-185. Katona, Z., & Sarvary, M. (2010). The race for sponsored links: Bidding patterns for search advertising. Marketing Science, 29(2), 199-215. Kelloway, E. K. (1998). Using LISREL for structural equation modeling: A researcher’s guide. Thousand Oaks, CA: Sage Kerr, J. H. (2001). Sport and exercise. In M. J. Apter (Ed.), Motivational styles in everyday life: A guide to reversal theory. (pp.187–213). Washington, DC: American Psychological Association. Kerr, J. H. (2005). Rethinking aggression and violence in sport. London, UK: Routledge. 264 University of Ghana http://ugspace.ug.edu.gh Kerr, J. H. (2007). Sudden withdrawal from skydiving: A case study informed by reversal theory’s concept of protective frame. Journal of Applied Sport Psychology 19(x), 337–351. Ketelaar, P. E., Konig, R., Smit, E. G., & Thorbjørnsen, H. (2015). In ads we trust’: Religiousness as a predictor of advertising trustworthiness and avoidance. Journal of Consumer Marketing, 32(3),190–98. *Kim, C., Kwon, K., & Chang, W. (2011). How to measure the effectiveness of online advertising in online marketplaces. Expert Systems with Applications, 38(4), 4234- 4243. Kim, C., Park, S., Kwon, K., & Chang, W. (2012). How to select search keywords for online advertising depending on consumer involvement: An empirical investigation. Expert Systems with Applications, 39(1), 594-610. Kim, J. U., Kim, W. J., & Park, S. C. (2010). Consumer perceptions on web advertisements and motivation factors to purchase in the online shopping. Computers in Human Behaviour, 26(5), 1202-1222. Kim, J., & Lennon, S. J. (2013). Effects of reputation and website quality on online consumers' emotion, perceived risk and purchase intention: Based on the stimulus- organism-response model. Journal of Research in Interactive Marketing, 7(1), 33- 56. Kim, J., & McMillan, S. J. (2008). Evaluation of Internet advertising research: A bibliometric analysis of citations from key sources. Journal of Advertising 37(1), 99- 112. Kim, K., Hayes, J. L., Avant, J. A., & Reid, L. N. (2014). Trends in advertising research: A longitudinal analysis of leading advertising, marketing, and communication journals, 1980 to 2010. Journal of advertising, 43(3), 296-316. Kim, N. Y. (2018). The Effect of Ad Customization and Ad Variation on Internet Users' Perceptions of Forced Multiple Advertising Exposures and Attitudes. Journal of Interactive Advertising, 18(1), 15-27. Kim, S. B., Kim, D. Y., & Wise, K. (2014). The effect of searching and surfing on recognition of destination images on Facebook pages. Computers in Human Behavior, 30, 813-823. Kim, S., Youn, S., & Yoon, D. (2018). Consumers’ responses to native vs. banner advertising: moderation of persuasion knowledge on interaction effects of ad type and placement type. International Journal of Advertising, 1-30. Kim, W. G., & Moon, Y. J. (2009). Customers’ cognitive, emotional, and actionable response to the servicescape: A test of the moderating effect of the restaurant type. International Journal of Hospitality Management, 28(1), 144-156. 265 University of Ghana http://ugspace.ug.edu.gh Kireyev, P., Pauwels, K., & Gupta, S. (2015). Do display ads influence search? Attribution and dynamics in online advertising. International Journal of Research in Marketing, 33(3), 475-490. Kline, R. B. (2011). Principles and Practice of Structural Equation Modeling (3rd ed.). New York: Guilford. Kline, R.B. (2005), Principles and Practice of Structural Equation Modelling (2nd ed.). New York: The Guilford Press. Knoll, J. (2015). Advertising in social media: a review of empirical evidence. International Journal of Advertising, 35(2), 266-300. Ko, H., Cho, C. H., & Roberts, M. S. (2005). Internet uses and gratifications: A structural equation model of interactive advertising. Journal of advertising, 34(2), 57-70. Kononova, A., & Yuan, S. (2015). Double-dipping effect? How combining YouTube environmental PSAs with thematically congruent advertisements in different formats affects memory and attitudes. Journal of Interactive Advertising, 15(1), 2- 15. Koo, C., Chung, N., & Kim, H. W. (2015). Examining explorative and exploitative uses of smartphones: a user competence perspective. Information Technology & People, 28(1), 133-162. Koo, D. M., & Ju, S. H. (2010). The interactional effects of atmospherics and perceptual curiosity on emotions and online shopping intention. Computers in human behavior, 26(3), 377–388. Korgaonkar, P., & Wolin, L. D. (2002). Web usage, advertising, and shopping: Relationship patterns. Internet Research, 12(2), 191–204. Krauss, S. E. (2005). Research paradigms and meaning making: A primer. The qualitative report, 10(4), 758-770. Krugman, H. E. (1977). Memory without recall, exposure without perception. Journal of Advertising Research, 17(4),7-12. Kuhn, T. (1970). The Structure of Scientific Revolutions. Chicago: University of Chicago Press. Kuisma, J., Simola, J., Uusitalo, L., & Öörni, A. (2010). The effects of animation and format on the perception and memory of online advertising. Journal of Interactive Marketing, 24(4), 269-282. Laczniak, R. N. (2015). The Journal of Advertising and the development of advertising Theory: reflections and directions for future research. Journal of Advertising, 44 (4), 429–433. 266 University of Ghana http://ugspace.ug.edu.gh Laczniak, R. N., & Muehling, D. D. (1993). The relationship between experimental manipulations and tests of theory in an advertising message involvement context. Journal of Advertising, 22(3), 59-74. Lam, L. W., Chan, K. W., Fong, D., & Lo, F. (2011). Does the look matter? The impact of casino servicescape on gaming customer satisfaction, intention to revisit, and desire to stay. International Journal of Hospitality Management, 30(3), 558-567. Lambrecht, A., & Tucker, C. (2013). When does retargeting work? Information Specificity in online advertising. Journal of Marketing Research, 50(5), 561-576. Lascu, D. N., Marcheva, M., & Thieringer, K. (2016). Magazine online advertising in France and the United States. Journal of Fashion Marketing and Management, 20(1), 120-135. Lavrakas, P. J., Mane, S., & Laszlo, J. (2010). Does anyone really know if online ad campaigns are working? Journal of Advertising Research, 50(4), 354-373. Lee, S. Y., & Cho, Y. S. (2010). Do web users care about banner ads anymore? The effects of frequency and clutter in web advertising. Journal of Promotion Management, 16(3), 288-302. Lee, S., Ha, S., & Widdows, R. (2011). Consumer responses to high-technology products: Product attributes, cognition, and emotions. Journal of Business Research, 64(11), 1195–1200. Leppäniemi, M., & Karjaluoto, H. (2008). Exploring the effects of gender, age, income and employment status on consumer response to mobile advertising campaigns. Journal of Systems and Information Technology, 10(3), 251-265. Lewis, R. A., & Rao, J. M. (2015). The unfavorable economics of measuring the returns to advertising. The Quarterly Journal of Economics, 130(4), 1941-1973. Lewis, R. A., Rao, J. M., & Reiley, D. H. (2011, March). Here, there, and everywhere: correlated online behaviors can lead to overestimates of the effects of advertising. In Proceedings of the 20th international conference on World wide web (pp. 157- 166). ACM. Li-Ming, A. K., Wai, T. B., Hussin, M., & Mat, N. K. N. (2013). The predictors of attitude towards online advertising. International Journal of Applied Psychology, 3(1), 7-12. Li, C., & Meeds, R. (2007). Factors affecting information processing of internet advertisements: a test on exposure condition, psychological reactance, and advertising frequency. In American Academy of Advertising. Conference. Proceedings (Online) (p. 93). American Academy of Advertising. Li, H., & Leckenby, J. (2004). Internet advertising formats and effectiveness. Center for Interactive Advertising. Retrieved August 23, 2018, from http://www.ciadvertising. org/studies/reports/measurement/ad_format_print.pdf 267 University of Ghana http://ugspace.ug.edu.gh Li, H., & Lo, H-Y. (2015). Do you recognize its brand? The effectiveness of online in- stream video advertisements. Journal of Advertising, 44(3), 208–18. Li, H., Edwards, S. M., & Lee, J.-H. (2002). Measuring the intrusiveness of advertisements: scale development and validation. Journal of Advertising, 31(2), 37–47. Li, H., Li, A., & Zhao, S. (2009). Internet advertising strategy of multinationals in China: a cross-cultural analysis. International Journal of Advertising, 28(1), 125-146. Li, M., Dong, Z.Y., & Chen, X. (2012). Factors influencing consumption experience of mobile commerce. Internet Research, 22(2), 120–141. Little, T. D., Card, N. A., Bovaird, J. A., Preacher, K. J., & Crandall, C. S. (2007). Structural equation modeling of mediation and moderation with contextual factors. In T. D. Little, J. A. Bovaird, & N. A. Card (Eds.), Modeling contextual effects in longitudinal studies (pp. 207–230). Mahwah: Lawrence Erlbaum Associates Publishers. Liu-Thompkins, Y. (2019). A Decade of online advertising research: What we learned and what we need to know. Journal of Advertising, 48(1) 1-13. Liu, S. Q., & Mattila, A. S. (2017). Airbnb: Online targeting advertising, sense of power, and consumer decisions. International Journal of Hospitality Management, 60(1), 33-41. Liu, Y., & Shrum, L. J. (2002). What is interactivity and is it always such a good thing? Implications of definition, person, and situation for the influence of interactivity on advertising effectiveness. Journal of advertising, 31(4), 53-64. Lombardi, L., & Pastore, M. (2012). Sensitivity of fit indices to fake perturbation of ordinal data: a sample by replacement approach. Multivariate behavioral research, 47(4), 519-546. Lutz, B. J., MacKenzie, S. B., & Belch, G. E. (1983). Attitude toward the ad as a mediator of advertising effectiveness: determinants and consequences. Advances in Consumer Research, 10, 532–539. Lu, Y., Zhou, L., Bruton, G., & Li, W. (2010). Capabilities as a mediator linking resources and the international performance of entrepreneurial firms in an emerging economy. Journal of International Business Studies, 41(3), 419-436. Mackenzie, S. H., Hodge, K., & Boyes, M. (2011). Expanding the flow model in adventure activities: A reversal theory perspective. Journal of Leisure Research, 43(4), 519- 544. MacKinnon, D. P., & Fairchild, A. J. (2009). Current directions in mediation analysis. Current directions in psychological science, 18(1), 16-20. Mahmoud, A. B. (2014). Linking information motivation to attitudes towards Web advertising. Journal of Islamic Marketing, 5(3), 396-413. 268 University of Ghana http://ugspace.ug.edu.gh Malhotra, N. K. (2007). Marketing research: An applied orientation (5th ed.). New Jersey: Pearson Education. Malhotra, N. K. (2010). Marketing research: An applied orientation (6th ed.). Upper Saddle River, NJ: Pearson Education. Malhotra, N. K., & Birks, D. F. (2007). Marketing Research: An Applied Approach (3rd European Ed.). Spain: Pearson Educational Limited. Manganari, E. E., Siomkos, G. J., & Vrechopoulos, A. P. (2009). Store atmosphere in web retailing. European Journal of Marketing, 43(9/10). 1140-1153. Marshall, C., & Rossman, G. B. (2014). Designing Qualitative Research. London: Sage publications. Martín-Santana, J. D., & Beerli-Palacio, A. (2012). The effectiveness of web ads: rectangle vs contextual banners. Online Information Review, 36(3), 420-441. Martin, R. A., Kuiper, N. A., Olinger, L. J., & Dobbin, J. (1987). Is stress always bad? Telic versus paratelic dominance as a stress-moderating variable. Journal of Personality and Social Psychology, 53(5), 970–982. Maslowska, E., Smit, E. G., & van den Putte, B. (2016). It is all in the name: A study of consumers’ responses to personalized communication. Journal of Interactive Advertising, 16(1),74-85. Mathieu, J. E., DeShon, R. P., & Bergh, D. D. (2008). Mediational inferences in organizational research; Then, now, and beyond. Organizational Research Methods, 11(2), 203-223. Maxwell, J. A., & Mittapalli, K. (2008). The Sage Encyclopaedia of Qualitative ResearchMethods. Thousand Oaks: Sage publications, Inc. Mazaheri, E., Richard, M. O., & Laroche, M. (2011). Online consumer behaviour: Comparing Canadian and Chinese website visitors. Journal of Business Research, 64(9), 958–965. McCoy, S., Everard, A., Galletta, D. F., & Moody, G. D. (2017). Here we go again! The impact of website ad repetition on recall, intrusiveness, attitudes, and site revisit intentions. Information & Management, 54(1), 14-24. McCoy, S., Everard, A., Galletta, D., & Moody, G. (2012). A rational choice theory approach towards: A causal model of online advertising intrusiveness and irritation. ECIS Proceedings. Paper 124. McCoy, S., Everard, A., Polak, P., & Galletta, D. F. (2008). An experimental study of antecedents and consequences of online ad intrusiveness. International Journal of Human–Computer Interaction, 24(7), 672-699. McQuitty, S., & Wolf, M. (2013). Structural equation modeling: A practical introduction. Journal of African Business, 14(1), 58-69. 269 University of Ghana http://ugspace.ug.edu.gh Mehrabian, A., & Russell, J. A. (1974). An approach to environmental psychology. Cambridge: MIT Press. Mehta, A. (2000). Advertising attitudes and advertising effectiveness. Journal of Advertising Research, 40(3), 67–72. Meyers, H., & Gerstman, R. (2001). The strategic role of e-branding. In branding@ thedigitalage (pp. 62-75). London: Palgrave Macmillan. Miller, K. (2005). Communication theories perspective, processes, and contexts (2nd Ed.). New York, NY: McGraw-Hill. Milne, G. R., Rohm, A. J., & Bahl, S. (2004). Consumers’ protection of online privacy and identity. Journal of Consumer Affairs, 38(2), 217-232. *Miralles-Pechuán, L., Ponce, H., & Martínez-Villaseñor, L. (2018). A novel methodology for optimizing display advertising campaigns using genetic algorithms. Electronic Commerce Research and Applications, 27(1), 39-51. Mitchell, A., & Olson, J. (1981). Are product attribute beliefs the only mediator of advertising effects on brand attitude. Journal of Marketing Research, 18(3), 318- 332. Mollen, A., & Wilson, H. (2010). Engagement, telepresence and interactivity in online consumer experience: Reconciling scholastic and managerial perspectives. Journal of Business Research, 63(9-10), 919-925. Montgomery, A. L., & Smith, M. D. (2009). Prospects for Personalization on the Internet. Journal of Interactive Marketing, 23(2), 130-137. Montoya, A. K., & Hayes, A. F. (2017). Two-condition within-participant statistical mediation analysis: A path-analytic framework. Psychological Methods, 22(1), 6. Mummalaneni, V. (2005) An empirical investigation of web site characteristics, consumer emotional states and on-line shopping behaviors. Journal of business research, 58(4), 526–532. Myers, M.D. (2013). Qualitative Research in Business and Management. Thousand Oaks, CA: Sage Publications. *Nasir, V. A. (2017). Identification of web user segments based on beliefs about online ads. Journal of Internet Commerce, 16(3), 231-254. NCA News (2016). List of internet service providers (ISPs) as at first quarter of 2016. Retrieved, June 25, 2019 from https://www.nca.org.gh/media-and-news/news/list- of-authorized-internet-service-providers-isps-as-at-first-quarter-2016/ Ndofor, H. A., Sirmon, D. G., & He, X. (2011). Firm resources, competitive actions and performance: Investigating a mediated model with evidence from in-vitro diagnostics industry. Strategic Management Journal, 32(6), 640-657. 270 University of Ghana http://ugspace.ug.edu.gh Nelson, P. (1974). Advertising as information. Journal of Political Economy, 82(4), 729- 754. Neuman, W. L. (2014). Social research methods: Qualitative and quantitative approaches. Essex: Pearson Education. Nexus (2013). Find adverting, marketing and PR expertise in Ghana. Retrieved January 5, 2020, from http://www.commonwealthofnations.org/sectors- ghana/business/advertising_marketing_and_pr/ Ngai, E. W. (2005). Customer relationship management research (1992-2002): an academic literature review and classification. Marketing Intelligence & Planning, 23(6), 582- 605. Nihel, Z. (2013). The effectiveness of internet advertising through memorization and click on a banner. International Journal of Marketing Studies, 5(2), 93-101. Nordea (2019). Ghana: Buying and selling - Advertising and marketing in Ghana. Retrieved March 4, 2020, from https://www.nordeatrade.com/en/explore-new- market/ghana/marketing?vider_sticky=oui Nottorf, F. (2014). Modeling the clickstream across multiple online advertising channels using a binary logit with Bayesian mixture of normals. Electronic Commerce Research and Applications 13(1):45–55. Nunnally, J. O. (1978). Psychometric theory. New York, NY: McGraw-Hill. Nunthiphatprueksa, A., & Suntrayuth, S. (2018). The application of Stimulus-Organism- Response Paradigm: The role of social media in Thailand’s destination Image and Behavioural intentions. ASEAN Journal of Management & Innovation, 5(1), 15-29. Okoli, C. (2015). A Guide to Conducting a standalone systematic literature review. Communications of the Association for Information Systems: 37(43), 879-910. Pace, V. L. (2010). Method variance from the perspectives of reviewers: poorly understood problem or overemphasized complaint? Organizational Research Methods, 13(3), 421–34. Pagani, M., Hofacker, C. F., & Goldsmith, R. E. (2011). The influence of personality on active and passive use of social networking sites. Psychology and Marketing, 28(5), 441–456. Pallant, J. (2011). SPSS survival manual a step by step guide to data analysis using SPSS (4th ed.). Crows Nest: Allen & Owen. Pallant, J. (2013). SPSS Survival Manual A Step by Step Guide to Data Analysis Using IBM SPSS (5th ed.). England, Berkshire: McGraw Hill. Palmgreen, P. (1984). Uses and gratifications: A theoretical perspective. Annals of the International Communication Association, 8(1), 20-55. 271 University of Ghana http://ugspace.ug.edu.gh Parra-Arnau, J., Achara, J. P., & Castelluccia, C. (2017). MyAdChoices: Bringing transparency and control to online advertising. ACM Transactions on the Web (TWEB), 11(1), 7. Pashkevich, M., Dorai-Raj, S., Kellar, M., & Zigmond, D. (2012). Empowering online advertisements by empowering viewers with the right to choose. Journal of Advertising Research, 52(4), 451-457. Percy, L. & Elliott, R. (2012). Strategic advertising management (4th ed.). Oxford: Oxford University Press. Pescher, C., Reichhart, P., & Spann, M. (2014). Consumer decision-making processes in mobile viral marketing campaigns. Journal of interactive marketing, 28(1), 43-54. Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. Journal of Consumer Research, 10(2), 135-146. Petty, R. E., Cacioppo, J. T., Strathman, A. J., & Priester, J. R. (2005). To think or not to think: exploring two routes to persuasion. In T. C. Brock, & M. C. Green (Eds.), Persuasion: Psychological Insights and Perspectives (2nd ed.). (pp. 81-116.). Thousand Oaks, CA: Sage Publications. Petty, R. E., Tormala, Z. L., Brinol, P., & Jarvis, W. B. G. (2006). Implicit ambivalence from attitude change: an exploration of the PAST model. Journal of personality and social psychology, 90(1), 21. Piotrowski, P. (2011). Street robbery offenders: Shades of rationality and reversal theory perspective. Rationality and Society, 23(4), 427-451. Pitt, L. F., Berthon, P., Caruana, A., & Berthon, J. P. (2005). The state of theory in three premier advertising journals: a research note. International Journal of Advertising, 24(2), 241-249. Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of method bias in social science research and recommendations on how to control it. Annual Review of Psychology, 63(10), 539-569. Podsakoff, P. M., Mackenzie, S. B., Bachrach, D. G., & Podsakoff, N. P. (2005). The influence of management journals in the 1980s and 1990s. Strategic Management Journal, 26(5), 473-488. Ponterotto, J. G. (2005). Qualitative research in counseling psychology: A primer on research paradigms and philosophy of science. Journal of counseling psychology, 52(2), 126. Portell, M., & Mullet, E. (2014). Why do people enjoy watching natural disasters and human violence on television? A reversal theory perspective. Journal of Motivation, Emotion, and Personality, 2(1), 38-49. 272 University of Ghana http://ugspace.ug.edu.gh Preacher, K. J., & Hayes, A. F. (2008). Contemporary approaches to assessing mediation in communication research. In A. F. Hayes, M. D. Slater, & L. B. Snyder (Eds.), The Sage Sourcebook of Advanced Data Analysis Methods for Communication Research (pp. 13-54). Thousand Oaks, CA: Sage. Proctor, T. (2005). Essentials of marketing research (4th ed.). USA: Prentice Hall. Quarshie, H. O., & Ami-Narh, J. (2012). The growth and usage of Internet in Ghana. Journal of Emerging trends in computing and information sciences, 3(9), 1302-1308. Rajaguru, R. (2014). Motion picture-induced visual, vocal and celebrity effects on tourism motivation: Stimulus organism response model. Asia Pacific Journal of Tourism Research, 19(4), 375-388. Rappaport, S. D. (2007). Lessons from online practice: New advertising models. Journal of Advertising Research, 47(2), 135–141. Rejón-Guardia, F., & Martínez-López, F. J. (2014). Online advertising intrusiveness and consumers’ avoidance behaviors. In Handbook of strategic e-business management (pp. 565-586), Berlin, Heidelberg: Springer. Resnick, M., & Albert, W. (2014). The Impact of advertising location and user task on the emergence of banner ad blindness: An eye-tracking study. International Journal of Human-Computer Interaction, 30(3), 206-219. Rhoades, E.B., Irani, T., Telg, R. & Myers, B. E. (2008). Internet as an information source attitudes and usage of students enrolled in a college of agriculture course. Journal of Agricultural Education, 49(2), 108-117. Richardson, H. A., Simmering, M. J., & Sturman, M. C. (2009). A tale of three perspectives: Examining post hoc statistical techniques for detection and correction of common method variance. Organizational Research Methods, 12(4), 762-800. Robinson, H., Wysocka, A., & Hand, C. (2007). Internet advertising effectiveness: the effect of design on click-through rates for banner ads. International Journal of Advertising, 26(4), 527-541. Robson, C. (2002). Real world research (2nd ed.). Oxford: Blackwell Publishing. Rodgers, S., & Thorson, E. (2000). The interactive advertising model: How users perceive and process online ads. Journal of interactive advertising, 1(1), 41-60. Rodgers, S., Wang, Y., Rettie, R. & Alpert, F. (2007). The web motivation inventory. International Journal of Advertising, 26(4), 447–476. Rosen, D. E., & Purinton, E. (2004). Website design: Viewing the web as a cognitive landscape. Journal of Business Research, 57(7), 787-794. Rosengren, S., Dahlén, M., & Modig (2013). Think outside the Ad: Can advertising creativity benefit more than the advertiser? Journal of advertising 42(4), 320-330. 273 University of Ghana http://ugspace.ug.edu.gh Rosenkrans, G. (2009). The creativeness and effectiveness of online interactive rich media advertising. Journal of interactive advertising, 9(2), 18-31. Rosenkrans, G. (2010). Maximizing user interactivity through banner ad design. Journal of Promotion Management, 16(3), 265-287. Rotzoll, K. B., Haefner, J. E., & Sandage, C. H. (1990). Advertising in contemporary society. Cincinnati, OH: South-Western. *Saadeghvaziri, F., Dehdashti, Z., & Askarabad, M. R. K. (2013). Web advertising: Assessing beliefs, attitudes, purchase intention and behavioral responses. Journal of Economic and Administrative Sciences 29(2), 99–112. Sahni, N. S., Wheeler, C. S., & Chintagunta, P. (2018). Personalization in email marketing: The role of noninformative advertising content. Marketing Science, 37(2), 236–58. Saunders, M. N., Saunders, M., Lewis, P., & Thornhill, A. (2011). Research Methods For Business Students (5th ed.). India: Pearson Education. Saunders, M., Lewis, P., & Thornhill, A. (2009). Research Methods for Business Students (5th ed.). Essex: Pearson Education Limited. Saunders, M., Lewis, P., & Thornhill, A. (2012). Research Methods for Business Students (6th Ed.). England, UK: Pearson. Schlosser, A. E., Shavitt, S., & Kanfer, A. (1999). Survey of Internet users’ attitudes toward Internet advertising. Journal of interactive marketing, 13(3), 34-54. Schneider, L-P., & Cornwell, B. T. (2005). Cashing in on crashes via brand placement in computer games. International Journal of Advertising, 24 (3), 321–43. Schumacker, R. E., & Lomax, R. G. (2010). A beginner's guide to structural equation modelling (3rd ed.). New York: Taylor & Francis Group. Schumann, J. H., von Wangenheim, F., & Groene, N. (2014). Targeted online advertising: Using reciprocity appeals to increase acceptance among users of free web services. Journal of Marketing, 78(1), 59-75. Segev, S., Wang, W., & Fernandes, J. (2014). The effects of ad–context congruency on responses to advertising in blogs: Exploring the role of issue involvement. International Journal of Advertising, 33(1), 17-36. Serenko, A., & Turel, O. (2018). A Dual-Attitude Model of System Use: The Effect of Explicit and Implicit Attitudes. Information & Management. DOI: https://doi.org/10.1016/j.im.2018.10.009 Seyedghorban, Z., Tahernejad, H., & Matanda, M. J. (2016). Reinquiry into advertising avoidance on the internet: A conceptual replication and extension. Journal of Advertising, 45(1), 120-129. 274 University of Ghana http://ugspace.ug.edu.gh Shaouf, A., Lü, K., & Li, X. (2016). The effect of web advertising visual design on online purchase intention: An examination across gender. Computers in Human Behaviour, 60, 622-634. Shaughnessy, J., Zechmeister, E. B., & Zechmeister, J. S. (2012). Research Methods Psychology (9th ed.). New York: McGraw-Hill. Sheng, H., & Joginapelly, T. (2012). Effects of web atmospheric cues on users’ emotional responses in e-commerce. AIS Transactions on Human-Computer Interaction, 4(1), 1-24. Sherman, E., Mathur, A., & Smith, R. B. (1997). Store environment and consumer purchase behaviour: Mediating role of consumer emotions. Psychology & Marketing. 14(4), 361-78. Sheth, J. N. (2011). Impact of emerging markets on marketing: Rethinking existing perspectives and practices. Journal of Marketing, 75(4), 166-182. Siemsen E., Roth, A., & Oliveira, P. (2010). Common method bias in regression models with linear, quadratic, and interaction effects. Organanisational Research Methods 13(3), 456–476. Silverman, D. (2001). Interpreting Qualitative Data (2nd ed.). London: Sage Publication Ltd. Simola, J., Kuisma, J., Öörni, A., Uusitalo, L., & Hyönä, J. (2011). The impact of salient advertisements on reading and attention on web pages. Journal of Experimental Psychology: Applied, 17(2), 174-190. Sit, C. H. P., & Lindner, K. J. (2006). Situational state balances and participation motivation in youth sport: A reversal theory perspective. British Journal of Educational Psychology, 76(x), 369-384. Skaalsvik, H., & Olsen, B. (2014). Service branding: suggesting an interactive model of service brand development. Kybernetes, 43(8), 1209-1223. Smit, E. G., Van Noort, G., & Voorveld, H. A. (2014). Understanding online behavioural advertising: User knowledge, privacy concerns and online coping behaviour in Europe. Computers in Human Behavior, 32(1), 15-22. Smith, M. S., & Albaum, S. G. (2005). Fundamentals of marketing research. Thousand Oaks, CA: Sage Publications. Smyth, R. (2004). Exploring the usefulness of a conceptual framework as a research tool: A researcher’s reflections. Issues in Educational Research, 14(2), 167-180. Sobel, M. E. (1986). Some new results on indirect effects and their standard errors in covariance structure models. In N. Tuma (Ed.), Sociological Methodology (pp.159- 186). Washington, DC: American Sociological Association. 275 University of Ghana http://ugspace.ug.edu.gh Song, J. H., & Zinkhan, G. M. (2008). Determinants of perceived web site interactivity. Journal of Marketing, 72(2), 99- 113. Song, P., Xu, H., Techatassanasoontorn, A., & Zhang, C. (2011). The influence of product integration on online advertising effectiveness. Electronic Commerce Research and Applications, 10(3), 288-303. Souiden, N., Chtourou, S., & Korai, B. (2017). Consumer Attitudes toward Online Advertising: The Moderating Role of Personality. Journal of Promotion Management, 23(2), 207-227. Spalding, L., Cole, S., & Fayer, A. (2009). How rich-media video technology boosts branding goals different online advertising formats drive different brand- performance metrics. Journal of Advertising Research, 49(3), 285-92. Spangenberg, E. R., Sprott, D. E., Grohmann, B., & Tracy, D. L. (2006). Gender-congruent ambient scent influences on approach and avoidance behaviors in a retail store. Journal of Business Research, 59(12), 1281-1287. Spilker-Attig, A., & Brettel, M. (2010). Effectiveness of online advertising channels: A price-level-dependent analysis. Journal of Marketing Management, 26(3-4), 343- 360. Sridhar, S., Germann, F., Kang, C., & Grewal, R. (2016). Relating online, regional, and national advertising to firm value. Journal of Marketing, 80(4), 39-55. Stanaland, A. J. S., & Tan, J. (2010). The impact of surfer/seeker mode on the effectiveness of website characteristics. International Journal of Advertising, 29(4), 569–595. Statista (2019). Internet advertising revenue in Ghana from 2013 to 2023. Retrieved, January 5, 2020 from, https://www.statista.com/statistics/889964/ghana-internet- ad-revenue/ Steenkamp, J.-B. E. M., & Baumgartner, H. (1992). The role of optimum stimulation level in exploratory consumer behavior. The Journal of Consumer Research, 19(3), 434- 448. Stipp, H. (2018). How context can make advertising more effective. Journal of Advertising Research, 58(2), 138-145. Streiner, D. L. (2013). 22 A checklist for evaluating the usefulness of rating scales. A guide for the statistically perplexed: Selected readings for clinical researchers. Toronto: University of Toronto Press. Sun, S., & Wang, Y. (2010). Familiarity, beliefs, attitudes, and consumer responses toward online advertising in China and the United States. Journal of Global Marketing, 23(2), 127-138. Sung, J. & Cho, K. (2012). The Influence of media type on attitude toward mobile advertisements over time. Cyberpsychology, Behavior, and Social Networking, 15(1), 31–36. 276 University of Ghana http://ugspace.ug.edu.gh Sweeney, J. C., & Wyber, F. (2002). The role of cognitions and emotions in the music approach- avoidance behavior relationship. Journal of Services Marketing, 16(1), 51-69. Tabachnick, B, G., & Fidell, L, S. (2007). Using Multivariate Statistics (5th Ed.). Boston, MA: Pearson International Education. Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th ed.). Boston, MA: Houghton Mifflin. Tam, K. Y., & Ho, S. Y. (2006). Understanding the impact of web personalization on user information processing and decision outcomes. MIS Quarterly, 30(4), 865–890. Tang, J., Zhang, P., & Wu, P. F. (2015). Categorizing consumer behavioral responses and artifact design features: The case of online advertising. Information Systems Frontiers, 17(3), 513-532. Tang, Z., Kreiser, P. M., Marino, L., & Weaver, K. M. (2010). Exploring proactiveness as a moderator in the process of perceiving industrial munificense: A field study of SMEs in four counties. Journal of Small Business Management, 48(2), 97–115. Taylor, D. G., Lewin, J. E., & Strutton, D. (2011). Friends, fans, and followers: do ads work on social networks?: how gender and age shape receptivity. Journal of advertising research, 51(1), 258-275. Te’eni-Harari, T., Lehman-Wilzig, S. N., & Lampert, S. I. (2009). The importance of product involvement for predicting advertising effectiveness among young people. International Journal of Advertising, 28(2), 203-29. Teng, N., Ye, N., Yu, Y., & Wu, X. (2014). Effects of culturally verbal and visual congruency/incongruency across cultures in a competitive advertising context. Journal of Business Research, 67(3), 288–94. The Report: Ghana 2012. Retrieved, May 25, 2018, from https://oxfordbusinessgroup. com/ghana-2012/media-advertising Tucker, C. E. (2014). Social networks, personalized advertising, and privacy controls. Journal of Marketing Research 51(5):546–562. Tutaj, K., & van Reijmersdal, E. A. (2012). Effects of online advertising format and persuasion knowledge on audience reactions. Journal of Marketing Communications, 18(1), 5-18. Valaei, N., & Rezae, S. (2016). The Effect of Culture on Attitude Toward Online Advertising and Online Brands: Applying Hofstede’s Cultural Factors to Internet Marketing. International Journal of Internet Marketing and Advertsing, 10(4), 270- 300. Van de Waldt, D. L. R, Schleritzko, N E. A., & Van Zyl, K. (2007). Paid versus unpaid celebrity endorsement in advertising: an exploration. Africa Journal of Business Management, 1(7), 185-191. 277 University of Ghana http://ugspace.ug.edu.gh van Doorn, J., & Hoekstra, J. C. (2013). Customization of online advertising: The role of intrusiveness. Marketing Letters, 24(4), 339-351. van Reijmersdal, E. A., Rozendaal, E., Smink, N., van Noort, G., & Buijzen, M. (2017). Processes and effects of targeted online advertising among children. International Journal of Advertising, 36(3), 396-414. Vargo, S. L., & Lusch, R. F. (2017). Service-dominant logic 2025. International Journal of Research in Marketing, 34(1), 46-67. Vieira, V. A. (2013). Stimuli–organism-response framework: A meta-analytic review in the store environment. Journal of Business Research, 66(9), 1420-1426. Wahyuni, D. (2012). The research design maze: Understanding paradigms, cases, methods and methodologies. Journal of applied management accounting research, 10(1), 69- 80. Wakolbinger, L.M., Denk, M., & Oberecker, K. (2009). The effectiveness of combining online and print advertisements: is the whole better than the individual parts? Journal of Advertising Research, 49(3), 360-72. Wang, J., & Calder, B. J. (2006). Media transportation and advertising. Journal of Consumer Research, 33(3), 152-162. Wang, K. Y., Shih, E., & Peracchio, L. A. (2013). How banner ads can be effective: Investigating the influences of exposure duration and banner ad complexity. International Journal of Advertising, 32(1), 121-141. Wang, K., Wang, E. T., & Farn, C. K. (2009). Influence of web advertising strategies, consumer goal-directedness, and consumer involvement on web advertising effectiveness. International Journal of Electronic Commerce, 13(4), 67-96. Wang, Y. J., Hernandez, M. D., & Minor, M. S. (2010). Web aesthetics effects on perceived online service quality and satisfaction in an e-tail environment: The moderating role of purchase task. Journal of Business Research, 63(9-10), 935-942. Wang, Y., & Sun, S. (2010a). An Online Advertising Model: Comparing China and the US. Journal of Current Issues & Research in Advertising, 32(2), 101-115. Wang, Y., & Sun, S. (2010b). Assessing beliefs, attitudes, and behavioral responses toward online advertising in three countries. International Business Review, 19(4), 333-344. Wang, Y., & Sun, S. (2010c). Examining the role of beliefs and attitudes in online advertising: A comparison between the USA and Romania. International Marketing Review, 27(1), 87-107. Wang, Y., & Sun, S. (2010d). Modeling online advertising: A cross-cultural comparison between China and Romania. Journal of Marketing Communications, 16(5), 271- 285. 278 University of Ghana http://ugspace.ug.edu.gh Wang, Y., Sun, S., Lei, W., & Toncar, M. (2009). Examining beliefs and attitudes toward online advertising among Chinese consumers. Direct Marketing: An International Journal, 3(1), 52-66. Wang, Y. J., & Minor, M. S. (2008). Validity, reliability, and applicability of psychophysiological techniques in marketing research. Psychology & Marketing, 25(2), 197-232. Webster, J. and Watson, R. T. (2002). Analysing the past to prepare for the future: Writing a literature review. MIS Quarterly, 26(2), 13–23. Wells, W., Burnett, J., Moriarty, S. E., & Pearce, R. C. (1992). Advertising: principles and practice. Englewood Cliffs, NJ: Prentice Hall. Williams, J., & MacKinnon, D. P. (2008). Resampling and distribution of the product methods for testing indirect effects in complex models. Structural Equation Modeling: A Multidisciplinary Journal, 15(1), 23-51. Wimmer, D. R., & Dominick, R. J. (2011). Qualitative research methods. Mass media research: An introduction, 9(1), 114-154. Wojdynski, B. W., & Evans, N. J. (2015). Going native: Effects of disclosure position and language on the recognition and evaluation of online native advertising. Journal of Advertising, 45(2), 157-168. Wolin, L., Korgaonkar, P., & Lund, D. (2002). Beliefs, attitudes and behavior towards Web advertising. International Journal of Advertising, 21(1), 87–113. Wong, Y. T., Osman, S., Jamaluddin, A., & Yin-Fah, B. C. (2012). Shopping motives, store attributes and shopping enjoyment among Malaysian youth. Journal of retailing and Consumer Services, 19(2), 240-248. World Bank (2014). World Development Indicators. New York, NY: World Bank World Bank (2019). The World Bank in Ghana. Retrieved January 5, 2020, from https://www.worldbank.org/en/country/ghana/overview Wynn Jr, D., & Williams, C. K. (2012). Principles for conducting critical realist case study research in information systems. MIS Quarterly, 36(3), 87-810. Xu, L., Duan, J. A., & Whinston, A. (2014). Path to purchase: A mutually exciting point process model for online advertising and conversion. Management Science, 60(6), 1392-1412. Yadav, M. S., & Varadarajan, R. (2005). Interactivity in the electronic marketplace: an exposition of the concept and implications for research. Journal of the Academy of Marketing Science, 33(4), 585-603. Yang, K. C., Huang, C. H., Yang, C., & Yang, S. Y. (2017). Consumer attitudes toward online video advertisement: YouTube as a platform. Kybernetes, 46(5), 840-853. 279 University of Ghana http://ugspace.ug.edu.gh Yang, S., & Ghose, A. (2010). Analyzing the relationship between organic and sponsored search advertising: Positive, negative, or zero interdependence?. Marketing Science, 29(4), 602-623. Yao, S., & Mela, C. F. (2011). A dynamic model of sponsored search advertising. Marketing Science, 30(3), 447-468. Yaveroglu, I., & Donthu, N. (2008). Advertising repetition and placement issues in on-line environments. Journal of Advertising, 37(2), 31-44. Yeu, M., Yoon, H-S., Taylor, C.R. and Lee, D-H. (2013) ‘Are banner advertisements in online games effective? Journal of Advertising, 42(2/3), 241–250. Yin, R. K. (2009). Case study research: design and method (4th ed.). London: Sage Publications. Yin, R. K. (2012). Applications of Case Study Research (3rd ed.).Thousand Oaks, CA: Sage Publications. Ying, L., Korneliussen, T., & Grønhaug, K. (2009). The effect of ad value, ad placement and ad execution on the perceived intrusiveness of web advertisements. International Journal of Advertising, 28(4), 623-638. Yoon, S., Choi, Y. K., & Song, S. (2011). When intrusive can be likable. Journal of Advertising, 40(2), 63–76. Yoo, C. Y., & Kim, K. (2005). Processing of animation in online banner advertising: The roles of cognitive and emotional responses. Journal of Interactive marketing, 19(4), 18-34. Zaichkowsky, J. L. (1985). Measuring the involvement construct. Journal of consumer research, 12(3), 341-352. Zajonc, R. (1980). Feeling and thinking: preferences need no inferences. American Psychologist, 35(2), 151–175. Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 9(2/2), 1-27. Zanjani, S. H. A., Diamond, W. D., & Chan, K. (2011). Does ad–context congruity help surfers and information seekers remember ads in cluttered e-magazines? Journal of Advertising, 40(4), 67–83. Zha, X., Li, J., & Yan, Y. (2015). Advertising value and credibility transfer: attitude towards web advertising and online information acquisition. Behaviour & Information Technology, 34(5), 520-532. Zikmund, W. G. (2003). Business research methods (7th ed.). Ohio: Thomson, South- Western. 280 University of Ghana http://ugspace.ug.edu.gh Zikmund, W., Babin, B., Carr, J., & Griffin, M. (2009). Business research methods. USA: South, Western College Publishing. Zhao, X., Lynch Jr, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and truths about mediation analysis. Journal of consumer research, 37(2), 197-206. 281 University of Ghana http://ugspace.ug.edu.gh APPENDICES Appendix A1 ODA Thematic Areas and Methodological Approaches Major Themes in Qualitative & ODA Research Quantitative Approaches Mixed Methods Controlled Experiment Field Experiment S u rvey & Content Secondary Analysis Market Data Antecedents to 1, 7, 11, 15, 18, 21, 2, ,4, 34, 40, 41, 36, 43, 63 6, 28 49, 52 Online Advertising 22, 24, 26, 29, 31, 44, 42 Effectiveness 39, 50, 54, 55 Assessing Online 1, 20, 21, 24, 39, 54, 2, 3, 13, 14, 16, 37 6, 8, 12, 25, Advertising 55, 62 19, 23, 30, 41, 28, 37, 47, Effectiveness 44, 45, 61, Attitude Toward 5, 10, 18, 53 17, 32, 35, 38, 46, 48, 9, 49 Online Advertising 49, 51, 56, 57, 58, 59, 60 Other Themes 33* Note: Bold: Mixed methods study Underlined: Qualitative study Asterisked (*): Content Analysis University of Ghana http://ugspace.ug.edu.gh Appendix A2 Numbered Articles used for the Review of Literature No. Author(s) No. Author(s) 1. Auschaitraku & Mucherjee (2017) 33. Li et al. (2009) 2. Belanche et al. (2017) 34. Liu & Mattila (2017) 3. Bleier & Eisenbeiss (2015a) 35. Mahmoud (2014) 4. Bleier & Eisenbess (2015b) 36. Martin-Santana & Beerli-Palacio (2012) 5. Bright & Daugherty (2012) 37. Miralles-Pechuan et al. (2017) 6. Bruce et al. (2016) 38. Nasir (2017) 7. Chan (2010) 39. Nihel (2013) 8. Chapelle et al. (2014) 40. Pashkevich et al. (2012) 9. Drossos et al. (2011) 41. Rosenkrans (2009) 10. Eshghi et al. (2017) 42. Segev et al. (2014) 11. Flores et al. (2014) 43. Seyedghorban et al. (2016) 12. Fridgeirsdottir & Najafi-Asadolahi (2018) 44. Simola et al. (2011) 13. Fugoni & Morn (2009) 45. Song et al. (2011) 14. Goldfarb & Tucker (2011) 46. Souiden et al. (2017) 15. Goodrich (2011) 47. Spilkerattig & Brettel (2010) 16. Goodrich (2013) 48. Sun & Wang (2010) 17. Goodrich (2014) 49. Tang et al. (2014) 18. Goodrich et al. (2015) 50. Tutaj & van Reijmersdal (2012) 19. Hoban & Bucklin (2015) 51. Valaei et al. (2016) 20. Hsieh & Chen (2011) 52. Van Dorn & Hoekstra (2013) 21. Hussain et al. (2018) 53. Van Reijmersdal et al. (2016) 22. Janssens et al. (2012) 54. Wang et al. (2013) 23. Jung et al. (2014) 55. Wang et al. (2009) 24. Jung et al. (2011) 56. Wang & Sun (2010a) 25. Kim et al. (2011) 57. Wang & Sun (2010b) 26. Kim (2018) 58. Wang & Sun (2010c) 27. Kim et al. (2018) 59. Wang & Sun (2010d) 28. Kireyev et al. (2015) 60. Wang et al. (2009) 29. Kuisma et al. (2010) 61. Xu et al. (2014) 30. Lambrecht & Tucker (2013) 62 Yeu et al. (2013) 31. Lee & Cho (2010) 63. Ying et al. (2009) 32. Li-Ming et al. (2013) 283 University of Ghana http://ugspace.ug.edu.gh Appendix A3 Publication by Journals and Years Journal Name and Category 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Total % Marketing 44 69.8 International Journal of Advertising √√ - - √ √ √ - √ - √ 7 Journal of Marketing Communication - √ - √√√ √ - - - - - 4 Journal of Advertising Research √ - - - - √ √ - - - 4 Journal of Marketing Research - - - - √ - √ √ - - 3 Journal of Interactive Advertising √ - - - - - - - - √ 2 Journal of Advertising - - - - √ - - √ - - 2 Journal of Interactive Marketing - √ - - - - - - √ - 2 Journal of Promotion Management - √ - - - - - √ - 2 Psychology & Marketing - - √ - - - - - √ - 2 Marketing Science - - √ - - - √ - - - 2 International Journal of Internet Marketing and Advertising - - √ - - - - √ - - 2 Marketing Letters - - - - √ - - - - - 1 Marketing Intelligence & Planning - - - - - - - - √ - 1 Journal of Research in Interactive Marketing √ - - - - - - - - - 1 Journal of Retailing - - - - - - √ - - - 1 Journal of Current Issues and Research in Advertising - √ - - - - - - - - 1 Journal of Marketing Management - √ - - - - - - - - 1 Journal of Global Marketing - √ - - - - - - - - 1 Journal of Islamic Marketing - - - - - √ - - - - 1 International Marketing Review - √ - - - - - - - - 1 International Journal of Research in Marketing - - - - - - √ - - - 1 Asia Pacific Journal of Marketing & Logistics - - √ - - - - - - - 1 International Journal of Marketing Studies - - - - √ - - - - - 1 Information Management 8 12.7 Computers in Human Behaviour - - √ - - √√ - - - - 3 Electronic Commerce Research and Applications - - √ - - - - - √ - 2 Expert Systems with Application - - √ - - - - - - - 1 Information Systems Frontiers - - - - - √ - - - - 1 International Journal of Electronic Commerce √ - - - - - - - - - 1 General Management, Economics, Operations Research, Sector & Area Studies 11 17.5 Journal of Business Research - - - - - √ - - - - 1 Management Science - - - - - √ - - - - 1 ACM Transactions on Intelligent Systems and Technology - - - - - √ - - - - 1 Operations Research - - - - - - - - - √ 1 Online Information Review - - - √ - - - - - - 1 Journal of Internet Commerce - - - - - - - - √ - 1 International Journal of Hospitality Management - - - - - - - - √ - 1 International Journal of Applied Psychology - - - - √ - - - - - 1 International Business Review - √ - - - - - - - - 1 Journal of Experimental Psychology: Applied - - √ - - - - - - - 1 IEEE Transactions on Engineering Management - √ - - - - - - - - 1 Total 6 9 8 5 7 9 5 4 7 3 63 100 284 University of Ghana http://ugspace.ug.edu.gh Appendix A4: Summary of Study Focus, Measured Variables, Findings and Gaps in ODA Research Study Focus and Measured/Studied Variables Major Findings Relevant Research Gaps Identified “Auschaitraku & Influence of website type on ODA effectiveness: ODA is more effective in terms of attitude Future research could: Mucherjee (2017) Website type, product involvement, brand toward the ad and brand on commercial than on 1. replicate these findings on behavior in the field by familiarity, attitude toward the ad, attitude social websites; as well as on brand compared to examining click stream and purchase data from toward the brand, and processing fluency. personal pages of the latter. actual ODA being shown on social and commercial Web sites. 2. investigate ad-related (source of the ad) and context-related (type of websites not used in the current study e.g. information or entertainment) moderators. Belanche et al. Skippable ad effectiveness: high-arousal ads are watched for longer time, Future research could: (2017) High and low-arousal stimuli, ad attitude, ad and are more effective in congruent 1. use actual brands in ad design as well as acceptance, brand attitude, ad intrusiveness, contexts. Users' product involvement determines longitudinal measures that may focus on brand context congruency, and product involvement. the intrusiveness of high- and low-arousal recall. skippable ads. 2. focus on elements that differentiate advertising from other online persuasive messages 3. examine how users’ active roles may determine ad effectiveness Bleier & Eisenbeiss Personalised online ad effectiveness -interplay of Personalisation increases click-through at early Future research could: (2015a) what, when and where: information stage of purchase decision and does 1. focus on other directly observable indicators Degree of content personalization, website so irrespective of banner ad-website congruency (duration of shop visit, spending per purchase) or congruency, web browsing mode (Goal-directed It however, increases view through only on implicit indicators (attitude toward ad, firm, ad vs experiential), click through, view through. motive congruent websites, but decreases it on recall) incongruent websites. 2. extend the degree of content personalisation to shopping actions (product placed in shopping carts or on wish lists) Bleier & Eisenbeiss Importance of trust in personalised online Trusted retailers can increase personalisation of Future research could: (2015b) advertising: ads through high depth and narrow breadth 1. consider other moderators (e.g. browsing mode, Depth of personalisation, Breadth of without eliciting reactance or privacy concern. age, personality traits, shopping habits etc.) personalisation, trust in the retailer, usefulness, For less trusted retailer’s high depth banners reactance, privacy concern, click-through. trigger reactance and privacy concerns reflecting in click-throughs. Bright & Daugherty Does customisation impact advertising: Customized environments create a sense of Future research could: (2012) Customisation desire for control (High/low), ad engagement for consumers; subjects who 1. determine additional individual differences format (Banner ad/Keyword ad/No ad), attitude thought they were exposed to a customized (including Internet self-efficacy and degree of toward advertising, content recognition, media environment had greater behavioural information overload) that may impact effects in behavioural intention for ad interaction. intention for interacting with advertising, and customized online environments. those who thought the environment was non- 2. explore how choices regarding personal agency and customized had a more positive attitude toward system-driven customizations impact the perception advertising. Ad recognition was high for those of advertising within customized online with low desire for control. environments.” 285 University of Ghana http://ugspace.ug.edu.gh “Bruce et al. (2016) The Effects of creative formats, message content Carryover effects for dynamic formats are Future research could: and targeting on engagement: greater than for static formats; yet static format 1. use individual or cookie level data Ad size, animation, content, and targeting. can still be effective for price ads and re- 2. build a hierarchical dynamic model using targeting. Re-targeted ads are effective only if demographics and retargeting data to define they offer price incentives segment level distributions from which individual behaviour could arise. Chan (2010) Online interaction-based advertising: Ads designed with implicit advertising intent Future research could: Exposure timing, advertising intent, brand compared to explicit intent are more effective in 1. examine other design factors to complement image, and purchase intention. the pre-decisional shopping phase. Brand image existing results: e.g. contextual factors relating to is found to moderate the effects of advertising websites (type and reputation) in which ads are intent on consumer’s purchase intention. launched. Drossos et al. Perceptions of advertising agencies and Reluctance toward online advertising due to Future research could: (2011) marketing managers: highly perceived disadvantages such as low 1. use qualitative approaches to examine perceptions Enablers and motivators, awareness, readiness to levels of; internet penetration, click-through of practitioners from countries with differing levels use, and perceptions. rates, online purchases, education in interactive of internet penetration and maturity. channels. Increase in online advertising expenditures would be driven by increase in internet penetration, and growth in internet buyers and sellers. Eshghi et al. (2017) Impact of online advertising on brand attitude: Effects of ad copy type and task orientation on Future research could: Ad copy type (narrative vs. factual), task brand attitude is mediated by ad message 1. conduct cross-cultural research across other orientation (researchers vs. surfers), Ad message involvement. Narrative ad copies generate emerging or developed economies to enhance the involvement, product type (technical vs. non- greater involvement than factual ad copies, generalizability technical), attitude toward the brand. regardless of task orientation or product type. 2. explore other possible moderators (e.g. utilitarian vs. hedonic products) 3. examine the potential for replication in the context of services Flores et al. (2014) Effects of variation in banner ads: An interaction suggested that the high- Future research could: Product involvement (high - smartphone vs. low involvement product was seen as fairly more 1. expand the sample pool and include other design involvement – news magazine), website context appealing if advertised with display ads rather elements as well as interact this features or factors. (video vs. newspaper), ad type and shape, than text-only ads; the low-involvement product language (English vs. Spanish), and attitude was slightly more appealing if advertised with toward the brand. text-only ads. Finally, products were slightly more appealing if advertised on ‘highly congruent’ websites. Goldfarb & Tucker Online display advertising targeting and Matching an ad to website content, and Future research could: (2011) obtrusiveness: increasing an ad’s visibility independently 1. consider actual purchase data Contextual targeted ads, obtrusiveness ads, increase purchase intent, but are ineffective 2. explore behavioural processes that stimulate privacy concerns, purchase intent when combined. These results are moderated by consumer privacy concerns. privacy concern. 3. explore responses to behavioural targeted ads to generate a theoretical framework for understanding how behavioural targeting and privacy concerns interact.” 286 University of Ghana http://ugspace.ug.edu.gh “Goodrich (2011) Attention to online advertising and multiple Attention to an ad is affected by ad type Future research could: outcomes: and the interaction between ad location, 1. use live webpages for better understating of user Ad type (pictorial vs. text), ad location (left vs. and page. attention is positively related to aided navigation and clickstreams. right), page type (image-oriented vs. textual), recall and to purchase intention, but negatively 2. Use diverse products and page contexts to enhance attention, aided recall, brand attitude, purchase associated with brand attitude. understanding of their effects. intention. Goodrich (2013) Effects of age and time of day on internet Older adults pay greater attention than younger Future research could: advertising outcomes: adults to online banner ads regardless of time of 1. use explore different advertised products to better Time of day, age, gender, attention, brand day. A mere exposure effect, with lower understand varied effects. attitude, and purchase intention. attention associated with more favorable brand 2. examine effects of product familiarity and attitudes, is found for the entire sample, but is involvement to reveal other insights. not particularly distinct for older adults. 3. study gender issues or effects across a wide range Purchase intention is highest for older adults, of products. mainly later in the day when processing resources are very low. Goodrich (2014) Gender brain-processing differences about online Results of this study support basic gender Future research could: advertising: variances in the processing of advertising 1. explore different Other potentially interesting areas Advertising location (left or right), gender, information: males have higher attention to for future research include using neuromarketing to attention to ad, and brand attitude. online ads and are more susceptible to mere test gender processing differences and exploring exposure effect than females, ads positioned on gender-related processing differences with higher the left of a page generate more favourable attention appeals. attitudes from males as rightward positioned ads do for females, and lower ad attention generates more favourable attitudes. Goodrich et al. Consumer reactions to online-video ads: Informative and humorous ads reduce (2015) Length, informativeness, humour, intrusiveness, intrusiveness; longer pre-roll ads heighten ad attitude toward the ad, brand and host website, ad recall; more intrusive ads generate more abandonment, ad recall, website revisit intention, abandonment, recall and negatively influence ad and purchase intentions. attitudes and reduce attitude toward the host website; attitude toward the brand influences purchase intention; and attitude toward the host website correlates with site revisit intentions. Hoban & Bucklin Effects of internet display advertising in the display ads positively affect visitation to the Future research could: (2015) purchase funnel: firm’s website for users in three of the four 1. use larger, more diverse data sets to substantiate Stages of purchase funnel (non-visitor, visitor, stages of the purchase funnels studied, but not results or the underlying mechanisms driving authenticated user, converted customer). Website for those who previously visited the site without differences in display advertising response by visits creating an account. Expected visits increase funnel stage are fully understood. nearly 10 percent when display ad impressions are partially reallocated from non-visitors and visitors to authenticated users. Hsieh & Chen Effects of information type on attention to online Different information types of browsing content Future research could: (2011) advertising: affect the number and intensity of attention to 1. use a large-scale sample beyond college students to ads from the strongest to the weakest were in provide a representative view of internet users.” 287 University of Ghana http://ugspace.ug.edu.gh Information type (text-based, text-picture mixed, this order: video-based, picture-based, text- 2. verify the influence of other web content types (e.g. picture-based, and video-based web pages), user picture mixed, and text-based. interactive format or audio format webpage) attention to banner ad “Hussain et al. Impact of web banner advertising frequency on Ad type moderated the influence of frequency Future research could: (2018) attitude: on brand attitude; the moderating effect of ad 1. ascertain what happens from sixth or more Exposure frequency, Ad type (pop-up vs. static), appeal manifested at a higher exposure exposure in order to establish the wear-out effect as ad appeal (emotional vs. rational); brand frequency. well as decide on the optimal frequency level to attitude. enhance suitable designs and executions. Janssens et al. Online advertising and congruency effects: Undivided attention benefits web ads that are Future research could: (2012) Type of exposure (ad before page vs. page before congruent with the web page in which they are 1. use several ads for different brands across different ad), context congruency (congruent vs. embedded, but divided attention benefits those product categories and test them in different incongruent), attitude toward the ad, and click that are incongruent with the web page. contexts to exclude the idiosyncratic effects of intention, and divided attention. testing one ad for one brand of a product category. 2. explore further processed by which congruency and incongruency effects emerge Jung et al. (2014) Telic/para-telic influence on internet advertising Telic state consumers form more positive Future research could: effectiveness: attitudes toward a low-level interactive ad, 1. conduct laboratory experiment to demonstrate User mode, perceived advertising interactivity, while para-telic state consumers form more reversal for different activities. attitude toward the ad, arousal seeking tendency. positive attitudes toward a high-level interactive 2. further explore how reversal theory explains online ad. Arousal seeking tendency mediates the behaviour. meta-motivational state's impact on ad attitude. Jung et al. (2011) How entertainment value affects persuasion: show that more favorable brand attitudes and Future research could: Perceived advertising value, recall, need for more positive purchase intentions are formed 1. examine entertainment value independent of cognitive closure, versatility of internet usage, when consumers are exposed to an ad that interactivity. attitude toward the brand and purchase intention. generates a high (game ad), rather than a low 2. further explore and clarify the relationship among (banner ad) level of entertainment value. goal accessibility, cognitive elaboration, and However, such effects are qualified by persuasion. consumers’ shopping goals. 3. use high involvement products. Kim (2018) Effect of ad customisation and variation on user Advertising content control through a Future research could: perceptions: customization feature was an influential factor 1. test what circumstances create a strong association Ad customisation, ad variation, perceived that led to positive attitudes toward multiple between attitudes toward the websites in a more intrusiveness, perceived irritation, attitude exposures to the ads. In addition, ad variation natural setting. toward the ad, attitude toward the website, induces users’ positive attitudes toward multiple 2. use a broader population and multiple products in ad exposures. Furthermore, users’ perceived competitive exposure instances. intrusiveness and feelings of irritation seemed to play an underlying mechanism in ad variation and customization -effectiveness relationship. Kim et al. (2018) Ad type (native vs. banner), placement type (solo In the solo condition, native advertising was Future research could: vs. duo), persuasion knowledge (high vs. low), evaluated more favourably than banner 1. use more diverse set of participants and ads/ perceived fit, ad credibility, brand attitude, and advertising in terms of perceived fit, ad 2. use natural exposure design and compare findings click intentions. credibility, brand attitudes and click intention. to find points of similarities and/or differences.” 288 University of Ghana http://ugspace.ug.edu.gh Significant interaction effects between ad type and placement type were found on all dependent variables, with native advertising showing a significant decrease in duo placement. These interaction effects appeared to be more manifest for consumers with high persuasion knowledge, confirming its moderating role. Kuisma et al. (2010) Effects of animation and format on perception There is a strong interaction effect between Future research could: and memory of online advertising: animation and ad format, which suggests that 1. investigate how task and task involvement influence Animation, ad format (banner and skyscraper), the effect of animation is conditioned by ad the perception of ads and attract visual attention. attention, and memorization (recognition and format: animation can increase attention to a 2. employ brain imaging on hemispheric recall). skyscraper, but not to a banner. Animation can lateralization, selective attention, and attention also increase memorizing of an ad, especially control to enrich the models of perception of recognition of banners. advertising Lambrecht & When does retargeting work: Dynamic retargeted ads are, on average, less Future research could: Tucker (2013) Dynamic retargeted and generic retargeted ads, effective than their generic equivalents. When 1. explicitly address the specific of dynamic purchase and review site visit consumers have narrowly construed preferences, retargeting ad design; e.g. which products should they respond positively (purchase) toads that be highlighted. display detailed product information (dynamic 2. explore how competitive ads moderate the retargeting). effectiveness of dynamic retargeting. Lee & Cho (2010) The effects of frequency and clutter in banner Exposure frequency influenced recall, attitudes Future research could: advertising: toward brand, and trial intention. behaviour. 1. examine influences of task-oriented and casual Exposure frequency, ad clutter, ad recall, ad However, contrary to expectation, banner browsers or users on ad effectiveness. recognition, attitude toward ad, attitude toward clutter does not lead to negative effects on 2. examine diverse formats and shapes in real life brand, and trial intention. recall, attitudes, and behaviour. Negative impact situations. of banner clutter was significant only on ad recognition. Li-Ming et al. Predictors of Attitude toward online advertising: Usability, trust and information are positive (2013) Usability, trust, and information, and attitude predictors of consumer attitude toward online toward online advertising advertising. Li et al. (2009) Internet advertising strategies of MNCs in China: Both Eastern and Western companies Future research could: Creative, placement and budget strategies. dominantly use individualist appeals 1. explore similar perspectives in different country for internet advertising in China, a collectivist contexts with different levels of economic country. However, Eastern multinationals also development. rely on emotional appeals, whereas Western companies generally adopt rational appeals. Liu & Mattila Online targeted advertising: Powerless individuals exhibit higher click- Future research could: (2017) Appeal (belongingness vs. uniqueness), sense of through and purchase intention to belongingness 1. use field experiments and measure consumers’ power (high vs. low), self-brand connection, appeal, while powerful individuals react more actual behaviours such as click-through rate and click-through intention, purchase intention positively to the uniqueness appeal. Self-brand purchase. connection is the underlying mechanism 2. use longitudinal studies to help check consistency (mediator) that explains these effects. across waves. 289 University of Ghana http://ugspace.ug.edu.gh “Mahmoud (2014) Linking information motivation to attitude Information motivation predicts three Future research could: toward web advertising: dimensions of beliefs about web advertising, i.e. 1. investigate or include other types of internet usage Information motivation, beliefs (information, positively for information and entertainment; motivation (e.g. social escapism, fun). entertainment, irritation, and value corruption), and negatively for irritation. Information 2. examine the sequential effects of several/arrays of and attitude toward web advertising motivation positively influences consumers’ ads. attitudes towards advertising. Also, information, entertainment and irritation are found to partially mediate the relationship between information motivation and attitude towards Web advertising. Martin-santana & Effectiveness of web ads: A direct relation between measurements of Future research could: Beerli-Palacio Ad format (rectangle vs. contextual banners), effectiveness and CTR emerged; there were 1. investigate and extend study results to other online (2012) product involvement, duration of website visit, differences in the effectiveness of the two platforms, use divers’ products and advertising attitude toward the website, unaided and aidede advertising formats explained by attitude toward formats. recall, aided, recognition, attitude toward the ad the web site, involvement with the product and and brand, and CTR. duration of web site visit. Nasir (2017) Identification of web user segments based on Web users were segmented into three Future research could: beliefs about online advertising: groups based on their beliefs about online ads 1. explore the impact of culture in web user segments Beliefs about online advertising, attitude toward (supporters, neutrals, and opponents). Web with positive or negative beliefs about online online advertising, willingness to purchase and users who hold affirmative beliefs about online advertising. pay more advertising have high levels of variety-seeking, innovativeness, & market-mavenism personality traits, when compared to those who hold negative beliefs. Nihel (2013) Effectiveness of internet advertising: Memorisation of a banner ad is largely affected Position, animation, size, profession, duration, by ad location, size and animation; and banner images, colours, memorisation, and click. click is influenced by colours used in the banner, size and clarity of the message. Rosenkrans (2009) Creativeness and effectiveness of online The interactive, rich media ad earned Future research could: interactive rich media advertising: significantly higher click-through rates than the 1. continue to focus on online ad placement Interactivity, ad placement, click-through, and non-interactive, rich media ads. It also 2. focus on measure of mouse rollover independent of mouse rollover. generated more user engagement and interactivity encouraged more user interactivity, thus increasing user involvement, as indicated by mouse rollovers. Segev et al. (2014) Effect of ad-context congruency on advertising A banner ad that is thematically congruent with Future research could: responses: the blog’s context generates more favourable 3. test participant in a more realistic setting of blog Ad-context congruency, attitude toward the ad, responses than an ad that is not congruent with reading using a mock blog. attitude toward the brand, purchase intention, the context. However, issue involvement 1. test the moderating effect of issue involvement on a issue involvement. moderates the effect of congruency. wide variety of issues.” 290 University of Ghana http://ugspace.ug.edu.gh “Seyedghorban et al. Re-inquiry into internet advertising avoidance: A positive relationship between perceived ad Future research could: (2016) Perceived goal impediment, perceived ad clutter, clutter, negative prior experience and ad 1. focus on other specific internet platforms (e.g., prior negative experience, user mode, and ad avoidance is supported but was weaker for gaming and social networking platforms), or media avoidance. perceived goal impediment. User mode (e.g., free web services). moderates the association between perceived 2. examine and compare avoidance toward different goal impediment and ad avoidance only among ad formats (e.g. banners, skyscrapers, interstitials telic users, and the impact of prior negative etc.). experience and ad avoidance was stronger 3. focus on examining differences across the three among paratelic users. No interaction effects of types of ad avoidance among telic and paratelic ad clutter emerged. users. Simola et al. (2011) Impact of salient advertisements on reading and Ads were overtly attended during reading and attention: dwell times on ads were the longest when the ad Ad position, animation, attention, view time above was static and the ad positioned right was animated; and ads close to the text capture more overt attention. Salient ads attract overt visual attention and disrupt reading and were viewed more frequently and for longer during casual browsing than during reading. Song et al. (2011) Influence of product integration on online The integration level influences the strength of Future research could: advertising effectiveness: the perceived tie which in turn has a significant 1. replicate the study and include task characteristics Integration type, perceived tie, attitude toward impact on advertising effectiveness. Product and other contextual factors. promotion, usage intention, and click-through integration level also has a direct impact on 2. investigate the influence of bounded choice through rates. advertising effectiveness. product bundling on user adoption behaviour. Souiden et al. Consumer attitude toward online advertising: Attitude toward advertising in general has a Future research could: (2017) attitude toward advertising, personality positive and significant impact on attitude 1. broaden the sample scope in terms of age (extroversion vs. introversion), attitude toward toward online advertising. Introversion is found 2. consider how other personality traits (e.g. advertising, attitude toward online advertising. to have no moderating effect on the relationship openness) may explain consumers’ attitude to between both attitudes. However, online advertising. extroversion moderates this relationship. Spilker-Attig & Channel Presence: e-mail, affiliate banner, The results show that push channels have a Future research could: Brettel (2010) affiliate price comparison, affiliate loyalty, and minor effect on sales compared with pull 1. focus on the details of individual ads (e.g. content SEM, price group, ad impressions, ad clicks, channels. Additionally, we show that products and size format etc.). site visitations and orders generated. at different price levels can be promoted via 2. analyse online advertising in combination with selective advertising channels. offline marketing efforts in order to gain insight into a lean and efficient marketing mix 3. examine the effect of social networks as well as interactions between the ad channels on orders. Sun & Wang (2010) Consumer response to online advertising in For U.S. consumers, familiarity was a positive Future research could: China and the United States: predictor of online shopping but not a 1. conduct longitudinal studies to offer further insights Familiarity with online advertising, beliefs about significant predictor of persuasion. For Chinese into the relationships among the evolving belief and online advertising, attitude toward online consumers, familiarity was a positive predictor attitudinal factors. advertising, and consumer responses (persuasion, of persuasion but not a significant predictor of 2. focus on how other social and individual factors and shopping experience). online shopping. Familiarity did not influence such as economic development level, any belief factors in the U.S. sample but demographics, lifestyle, and Internet experience, 291 University of Ghana http://ugspace.ug.edu.gh emerged as a significant predictor of all five and development stage of online advertising come belief factors in the Chinese sample. For both together to influence online advertising.” samples, ATOA positively predicted persuasion (stronger for US consumers), which positively predicted online shopping. “Tang et al. (2014) Categorising consumer behavioural responses in All four types of consumer behaviors were Future research could: online advertising: present, and all behaviors identified can be 1. provide more rigorous validation about the Ad content, ad form, and ad action, active classified into one of the four types. interplay between ad features, judgement of ad approach behaviour, active avoidance behaviour, categorization of the three types of ad design features and behavioural responses. passive approach behaviour, passive avoidance features can also guide the understanding of 2. focus on other design features like repetition, behaviour consumers’ judgments of ads, which may timing etc. function as a bridge of ad design features’ influence on consumer behaviours. Tujat & Reijmersdal Effect of online advertising format on audience Participants find sponsored content more Future research could: (2012) reaction: informative, more amusing, and less irritating 1. explore the mediating or moderating effects of Ad format (sponsored content and banner ad), than the banner ad. With respect to persuasion involvement and web experience. perceived advertising value, advertising knowledge, recognition of the advertising 2. investigate other formats (e.g. advergames, pop- recognition, understanding of persuasive and format, understanding of persuasive intent, and ups) to provide more evidence for format selling intent, and advertising skepticism. ad skepticism are higher for banner ads than for effectiveness sponsored content. Ad skepticism seems to be 3. construct and validate an overall measurement scale strongly related to perceived advertising value. for persuasion knowledge. Valaei et al. (2016) Effect of culture on attitude toward advertising Individualism and long-term orientation are Future research could: and online brands: predictors of ATOA and ATOB. In addition, 1. explore the results of inter-cultural difference in Hofstede’s cultural factors, attitude toward uncertainty avoidance does not have a positive diverse countries. online advertising, and attitude toward online relationship with ATOA and ATOB; and while brands. there is no association between power distance and ATOB. Van Dorn & Customisation of online advertising: Higher degrees of personalization such as Hoekstra (2013) Personalization (use of name, use of transaction adding personal identification or transaction information), privacy concern, intrusiveness, information to browsing data is considered a and purchase intentions. double-edged sword; it increases purchase intention but also increases feelings of intrusiveness, which negatively affects purchase intentions. respondents with higher levels of privacy concerns experience stronger feelings of intrusiveness. Van Reijmersdal et Processes and effects of targeted online Product targeting results in greater brand liking Future research could: al. (2016) advertising among children: and increased purchase intention. These 1. compare children, teenagers, and adults in a single Profile targeting (colour vs. product), ad liking, effects are explained by increased ad liking. study to gain more insights into the type of perceived personal relevance, targeting Targeting colors in online advertisements had processing each uses. recognition, brand attitude, purchase intention. no effect. Children did not think that profile- 292 University of Ghana http://ugspace.ug.edu.gh targeted ads are more relevant, and they also did 2. examine whether other types of targeting have not seem to understand the targeting tactic. stronger impacts., that is, whether different types of targeting differ their effects. 3. use real targeted online advertisements based on children’s profile pages. 4. employ a larger sample size and test hypotheses using SEM 5. use qualitative methods, allowing children to talk about this practice through use of tools such as focus groups or in-depth interviews” “Wang et al. (2013) How banner ads can be effective: When a banner ad is difficult to process, Future research could: Exposure duration, ad complexity, attitude increase in exposure duration in the priming 1. explore these effects in online media environment toward the ad, attitude toward the brand, phase linearly heightens attitude toward the ad 2. determine the optimal exposure frequency of and brand in the testing phase. When the ad is banner ads moderately difficult, and easy to process, n 3. further investigate the possibility that the degree of inverted-U pattern, and U pattern occur ad-context congruity has a meaningful impact on respectively. the effects of exposure duration and banner ad complexity. Wang et al. (2009) Influence of web advertising strategies on Product involvement moderated the relationship Future research could: effectiveness: between ad variation and Attitude toward the 1. verify results across diverse product categories, and Ad variation strategy (substantive vs. cosmetic), ad. Specifically, under substantive variation, website types message appeal strategy (emotional vs. high-involvement consumers showed more 2. incorporate other ad formats to enrich outcomes informational), and goal-directedness, positive attitude toward the ad than low- and examine behavioural measures (e.g. click- Consumer involvement, and attitude toward the involvement consumers. Results were similar in through rate, eye tracking) to extend findings. ad the instance of informational appeal Wang & Sun Cross cultural comparison of beliefs, attitude, Culture influences belief factors: compared to Future research could: (2010a); (2010b) and responses to online advertising: Chinese, Americans believe online advertising 1. examine a broader profile of online consumers Belief factors (information seeking, is more informative, credible and less beneficial besides students and compare online advertising entertainment, credibility, economy, value to the economy. across different profiles. corruption), attitude toward online advertising, 2. conduct longitudinal studies to offer further insights ad clicking, and shopping frequency The influence of ATOA on ad clicking is into the relationships among the evolving belief and stronger among Chinese than Americans attitudinal factors. (2010a). 3. focus on how other social and individual factors such as economic development level, Romanians believed online advertising was demographics, lifestyle, and Internet experience more informative and credible; and they held conspire to influence online advertising. more positive attitudes toward online advertising compared Americans and Chinese consumers (2010b). Wang & Sun Cross cultural comparison of beliefs, attitude, all five belief factors significantly predicted (2010c); (2010d) and responses to online advertising: ATOA, which in turn predicted online ad clicking, and online shopping frequency. 293 University of Ghana http://ugspace.ug.edu.gh Belief factors (information seeking, entertainment, credibility, economy, value Compared to the Chinese, Romanians held more corruption), attitude toward online advertising, positive ATOA and were more likely to click on ad clicking, and shopping frequency ads, while Chines were more likely to purchase online than Romanians (2010c). Compared to Americans, Romanians held a more positive ATOA and are more likely to click advertisements, while Americans are more likely than Romanians to make online purchases (2010d). Wang et al. (2009) Belief and attitude toward online advertising: Five belief factors that underlie Chinese Belief factors, attitude toward online advertising, consumers’ ATOA were identified: ad clicking, and shopping frequency entertainment, information seeking, credibility, economy, and value corruption. Information seeking, economy and value corruption were significant predictors of ATOA. ATOA was significantly predicted of ad clicking and online shopping frequency. Yeu et al. (2013) Are banner advertisements in online games Banner ads have the ability to be noticed and Future research could: effective? either explicitly or implicitly remembered in the 1. examine a affective and behavioural responses in Recall, recognition, and implicit memory. context of an online advergame. Both explicit addition to the cognitive measures. and implicit memory does not vary based on 2. Consider alternative methods of encouraging achievement level in the game. involvement. Ying et al. (2009) Effects of ad value, placement and execution on Future research could: intrusiveness of web advertising: 1. consider broader samples of internet users beyond Ad value (content congruence, entertainment) ad students as well as large sample sizes and control placement (frequency and quantity), ad execution for demographics. (sound, size, and animation), and intrusiveness. 2. use real-life surfing situations to better capture user reactions and investigate whether the degree of involvement in the ad moderates intrusiveness.” Note: Italicized=Mediator Bold=Moderators 294 University of Ghana http://ugspace.ug.edu.gh Appendix B: Survey Instrument UNIVERSITY OF GHANA BUSINESS SCHOOL DEPARTMENT OF MARKETING AND ENTREPRENEURSHIP SURVEY INSTRUMENT Dear Sir/Madam, I am a PhD candidate at the University of Ghana Business School examining Consumer behavioural responses to online display advertising in Ghana. Information provided for the purposes of this research will be treated confidentially and used for academic purposes only. I would be grateful if you would take a few minutes (approx.: 7-10 mins) to fill out this questionnaire by ticking (√) where appropriate. For any further clarifications and enquiries, kindly contact me via this email address – pmensah047@st.ug.edu.gh NB: Online display advertising or display ads are graphic (visual) advertisements that are shown on websites, apps, and social media sites (e.g. Facebook, Twitter, YouTube etc.). They are mostly made of images, animations, flash, video, audio and text etc., and are meant to deliver general brand advertising messages to website visitors or internet users. Section A: Internet Usage Information 1. How often do you use the Internet? [ ] Never or almost never [ ] Once or twice a month [ ] Several times a week [ ] Every or almost every day [ ] Several times a day 2. What is your major motive or reason for using the Interent? [ ] I go online to make purchases and buy things - Shopping [ ] I use the internet to search for any kind of information – (Re)searching [ ] When online, I’m mostly browsing for fun or exploring new sites - Surfing [ ] I go online to chat with friends, send and check emails – Communication 3. How familiar are you with advertising on the internet (online advertising)? [ ] Not familiar [ ] Slightly familiar [ ] Somewhat familiar [ ] Familiar [ ] Very Familiar 4. How often are you exposed to online display ads while using the internet? [ ] Never [ ] Rarely [ ] Sometimes [ ] Often 295 University of Ghana http://ugspace.ug.edu.gh [ ] Always 5. How long ago did you view an online display ad and/or make a purchase after viewing one? [ ] Below 1 month [ ] 1-2 months [ ] 3-4 months [ ] 5-6 months [ ] Above 6 months 6. What is the nature of the web page/site on which you saw the ad? [ ] Informational website (e.g. blogs, news site etc.) [ ] Social media site (e.g. Facebook, Twitter, Instagram, YouTube etc.) [ ] Commercial website (e.g. Tonaton, Jumia, OLX, Kaymu Amazon etc.) [ ] Search engine (e.g. Google Adwords) 7. What was the nature of the advertised brand? [ ] Product [ ] Service 8. Kindly state the specific product or service (e.g. mobile phone, clothing, restaurant, telecom) ………………………………………………………………… Section B: On a scale of 1-7, please indicate by ticking (√) or circling, the extent to which you agree or disagree with the following statements regarding the online display ad you recently saw and/or for which you made the purchase. 1=Strongly Disagree, 2=Disagree, 3=Somewhat Disagree, 4=Neutral, 5=Somewhat Agree, 6= Agree, 7=Strongly Agree No. Statement Scales Strongly Strongly Interactivity: the online ad I saw… Disagree Agree 1. had interactive features 1 2 3 4 5 6 7 2. had links I could click for further information 1 2 3 4 5 6 7 3. provided opportunity for me to give my feedback 1 2 3 4 5 6 7 4. gave instantaneous information when I requested 1 2 3 4 5 6 7 5. made me feel the advertiser wants to listen to customers 1 2 3 4 5 6 7 6. allowed me a lot of control over my ad viewing experience 1 2 3 4 5 6 7 7. allowed me to choose the timing of the ad 1 2 3 4 5 6 7 8. provided two-way communication 1 2 3 4 5 6 7 Strongly Strongly Placement: the online ad I saw… Disagree Agree 1. was a good fit/match for the webpage/site on which it appeared 1 2 3 4 5 6 7 2. was consistent with the webpage/site on which it was featured 1 2 3 4 5 6 7 3. and the webpage/site on which it appeared belong together 1 2 3 4 5 6 7 4. had a lot in common with the webpage/site on which I saw it 1 2 3 4 5 6 7 5. matched the content of the webpage/website 1 2 3 4 5 6 7 6. was made more credible by the webpage on which it was featured 1 2 3 4 5 6 7 7. was similar to other ads on the webpage/site 1 2 3 4 5 6 7 296 University of Ghana http://ugspace.ug.edu.gh Strongly Strongly Informativeness: the online ad I saw… Disagree Agree 1. was a good source of product/service information 1 2 3 4 5 6 7 2. was a convenient source of product/service information 1 2 3 4 5 6 7 3. supplied relevant information which was of value to me 1 2 3 4 5 6 7 3. had information worth paying attention to 1 2 3 4 5 6 7 5. gave me new ideas about the product/service 1 2 3 4 5 6 7 6. helped keep me up to date with the product/service category 1 2 3 4 5 6 7 7. made product information readily accessible 1 2 3 4 5 6 7 8. supplied complete product/service information 1 2 3 4 5 6 7 Strongly Strongly Personalisation: the online ad I saw … Disagree Agree 1. was tailored to my shopping situation at the time 1 2 3 4 5 6 7 2. made recommendations that matched my needs at the time 1 2 3 4 5 6 7 3. made me feel unique as an internet user 1 2 3 4 5 6 7 4. was related to my search history at the time 1 2 3 4 5 6 7 5. was useful and meaningful to me 1 2 3 4 5 6 7 6. provided information based on my real-time location 1 2 3 4 5 6 7 7. used my personal information (e.g. name & gender) 1 2 3 4 5 6 7 Strongly Strongly Exposure Condition: the online ad I saw… Disagree Agree 1. permitted me to choose freely what I wanted to see 1 2 3 4 5 6 7 2. did not interfere with my online activity at the time 1 2 3 4 5 6 7 3. did not intrude on the content I was accessing 1 2 3 4 5 6 7 4. was not forced upon me 1 2 3 4 5 6 7 5. was not repeated while I was on the webpage 1 2 3 4 5 6 7 6. was not shown more than once during my activities on the webpage 7. did not stay on the screen for long 1 2 3 4 5 6 7 8. was not on the screen for up to 30 seconds 1 2 3 4 5 6 7 Section C: On a scale of 1-7, please indicate by ticking (√) or circling, the extent to which you agree or disagree with the following statements regarding your general disposition toward online advertising, and your reactions to online display advertising 1=Strongly Disagree, 2=Disagree, 3=Somewhat Disagree, 4=Neutral, 5=Somewhat Agree, 6= Agree, 7=Strongly Agree Strongly Strongly Attitude toward Online Advertising Disagree Agree 1. My general opinion of online advertising is favourable 1 2 3 4 5 6 7 2. I consider online advertising very essential 1 2 3 4 5 6 7 3. Online advertising is interesting and fun to see 1 2 3 4 5 6 7 4. I appreciate seeing advertising messages on the Internet 1 2 3 4 5 6 7 5. Overall, I like online advertising 1 2 3 4 5 6 7 Strongly Strongly Ad Avoidance Disagree Agree 297 University of Ghana http://ugspace.ug.edu.gh Passive Avoidance 1. I intentionally ignore online ads when using the internet 1 2 3 4 5 6 7 2. I look away from online ads when I’m using the internet 1 2 3 4 5 6 7 3. I wait for online ads to go away, then I continue with what I’m doing 1 2 3 4 5 6 7 Active Avoidance 4. I click/scroll away from or leave webpages displaying online ads 1 2 3 4 5 6 7 5. I skip/close online ads that appear on my screen while I’m online 1 2 3 4 5 6 7 6. I use ad blockers on my computer 1 2 3 4 5 6 7 Strongly Strongly Ad Acceptance Disagree Agree Passive Acceptance 1. I pay attention to online advertisements 1 2 3 4 5 6 7 2. I carefully read the content of online advertisements 1 2 3 4 5 6 7 3. I watch online/read online advertisements to the end 1 2 3 4 5 6 7 Active Acceptance 4. I click on online advertisements or links provided in online ads 1 2 3 4 5 6 7 5. I bookmark online ads when I’m using the internet 1 2 3 4 5 6 7 6. I sign up or give feedback if the ad provides the option 1 2 3 4 5 6 7 Section D: Demographic Information Please respond to the following demographic questions, and we are done. 1. Age: [ ] 18-30 years [ ] 31-40 years [ ] 41-50 years [ ] Above 50 years 2. Gender: [ ] Male [ ] Female 3. Educational Level: [ ] Basic (Primary up to JHS) [ ] Secondary/Vocational/Technical [ ] HND/Diploma/Undergraduate Degree [ ] Postgraduate Degree [ ] Professional Degree 4. Employment Status: [ ] Unemployed [ ] Self-Employed [ ] Salaried worker (Full-time) [ ] Salaried worker (Part-time) [ ] College/Tertiary Student 5. Nationality: [ ] Ghanaian [ ] Other, please specify ……………………………… 6. Region of Residence (e.g. Greater Accra, Volta etc.): …………………………… Thank you for your time! 298 University of Ghana http://ugspace.ug.edu.gh Appendix C: Ethical Approval 299