University of Ghana http://ugspace.ug.edu.gh DOES GENDER IN CELEBRITY ENDORSED ADVERTISEMENTS INFLUENCE AUDIENCE PERCEPTION? A STUDY OF UNIVERSITY OF GHANA STUDENTS BY ALICE MENSAH 10701186 THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILMENT OF THE REQUIREMENT FOR THE AWARD OF MPHIL IN COMMUNICATION STUDIES DEGREE APRIL, 2021 University of Ghana http://ugspace.ug.edu.gh DECLARATION I declare that this work is the result of an independent enquiry, carried out by the researcher under supervision and has not been presented anywhere by another researcher. All works mentioned to enable the discussion within the study have been recognized appropriately. ……………………….. 30th April, 2021 Alice Mensah (Student) …………………………. 30th April, 2021 Dr. Sarah Akrofi-Quarcoo (Principal supervisor) ………………………… 30th April, 2021 Dr. Abena A. Yeboah-Banin (Second supervisor) i University of Ghana http://ugspace.ug.edu.gh DEDICATION I dedicate this work to the Lord Jesus Christ by whose great mercies this victory has been won. ii University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENTS I praise God the father, the Lord Jesus Christ and the Holy Spirit, my special Helper. Indeed God has been my ever present help in times of troubles. He has been with me when I passed through deep waters and fire. When I was afraid, He became my tower. When I was weak, His power towards me became greater. Dear Lord, I thank you so much for the great mercies you have shown me. I crossed this line because you were determined not to give up on me. May your great name be praised forever! I also appreciate my principal supervisor, Dr. Sarah Akrofi-Quarcoo for the great assistance she freely offered to me. Auntie Sarah, in you I found a mother. You were with me throughout the processes of this research. Your pieces of advice and encouragements spurred me on. You believed in me so much that I also began believing in myself too. I really appreciate your kindness, patience, goodwill and motherly love. May the Almighty God reward you exceedingly! To Dr. Abena Animwaa Yeboah-Banin my second supervisor, your kind is rare. I deeply appreciate your inputs into my work. I am most grateful for your sacrifices, time and encouragements. I always entered your office with a troubled heart and left with a smile and conviction that all will be well. Doc, may God bless you abundantly. To my family, Mr. Simon Peter Gyakye, my mum and siblings, you were my fans on the terrace who refused to keep quiet. You cheered me on with your encouragements. To my sisters, Charlotte and Mary, you were the cheerleaders! I always had renewed sense of hope after speaking to you. To my “little rascals”, John-Paul and Mary Anne, your strong demand for my presence in the iii University of Ghana http://ugspace.ug.edu.gh house, immediately after 31st October, put me on my toes to meet the submission deadline at all cost. I love you guys! To Mrs. Maame Abena JacobPaul, you have been a sister, confidant, prayer partner, advisor and a character mentor to me. You stuck with me even when you had every reason to hold your peace. I thank God for your life and the person you have become. You are the best roommate ever! May God bless you abundantly. I am also grateful to Dr. Gilbert Tietaah, Mr Fidelis Sesenu, Kweku Sakyi Appiah, Nana Ama Gyemaah Otuahene, Mr. Benson Osei Tutu and all who helped me to realize this dream. What shall I say? May the Lord who sees the deeds of men, reward you greatly! iv University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENT Contents Page DECLARATION ............................................................................................................................. i DEDICATION ................................................................................................................................ ii ACKNOWLEDGEMENTS ........................................................................................................... iii TABLE OF CONTENT .................................................................................................................. v LIST OF TABLES ........................................................................................................................ vii LIST OF FIGURES ..................................................................................................................... viii ABSTRACT ................................................................................................................................... ix CHAPTER ONE ............................................................................................................................. 1 INTRODUCTION .......................................................................................................................... 1 1.0 Background to the Study ....................................................................................................... 1 1.1 Problem Statement ................................................................................................................ 5 1.2 Rationale of the study ............................................................................................................ 7 1.3 Objectives of the Study ......................................................................................................... 7 1.4 Research Questions ............................................................................................................... 8 1.5 Significance of Study ............................................................................................................ 9 1.6 Organization of Study ......................................................................................................... 10 CHAPTER TWO .......................................................................................................................... 11 2.0 Introduction ......................................................................................................................... 11 2.1 Source Credibility Model .................................................................................................... 11 2.2 Match-Up Hypothesis Model .............................................................................................. 13 CHAPTER THREE ...................................................................................................................... 16 3.0 Introduction ......................................................................................................................... 16 3.1 Celebrity Endorsements as a Brand Communication Strategy ........................................... 17 3.2 Celebrity Endorsement in Ghana ........................................................................................ 20 3.3 Implications of Gender in Celebrity Endorsement .............................................................. 22 3.4 Endorser and Product Congruence ...................................................................................... 23 3.5 Advertisement Believability ................................................................................................ 27 3.6 Research Hypotheses ........................................................................................................... 30 3.8 Chapter Summary ................................................................................................................ 33 CHAPTER FOUR ......................................................................................................................... 34 4.0 Introduction ......................................................................................................................... 34 v University of Ghana http://ugspace.ug.edu.gh 4.1 Research Paradigm .............................................................................................................. 34 4.2 Research Design .................................................................................................................. 34 4.3 Empirical Setting and Sample ............................................................................................. 36 4.4 Sampling Procedure ............................................................................................................ 36 4.5 Data Collection Procedure .................................................................................................. 38 4.5.1 Pre-Test ........................................................................................................................ 39 4.5.2 Main Study .................................................................................................................... 40 4.6 Construct Measures ............................................................................................................. 42 4.7 Data Analysis Procedure ..................................................................................................... 43 4.8 Ethical Consideration .......................................................................................................... 44 4.9 Chapter Summary ................................................................................................................ 44 CHAPTER FIVE .......................................................................................................................... 46 FINDINGS .................................................................................................................................... 46 5.0 Introduction ......................................................................................................................... 46 5.1 Sample Description ............................................................................................................. 46 5.2 Analysis of Construct Measures .......................................................................................... 49 3.3 Hypothesis Testing .............................................................................................................. 62 CHAPTER SIX ............................................................................................................................. 72 DISCUSSION OF FINDINGS ..................................................................................................... 72 6.0 Introduction ......................................................................................................................... 72 6.2 The Relationship between Endorser Gender and Advertisement Believability .................. 73 6.3 Endorser Gender -Product Gender Fit Implications for Believability ................................ 75 6.4 Chapter Summary ................................................................................................................ 79 SUMMARY, CONCLUSIONS AND RECOMMENDATIONS ................................................ 80 7.0 Introduction ......................................................................................................................... 80 7.1.1 Summary ....................................................................................................................... 80 7.2 Conclusions/recommendations ............................................................................................ 82 7.3 Limitations/recommendations ............................................................................................. 85 7.5 Chapter Summary ................................................................................................................ 87 REFERENCES ............................................................................................................................. 88 APPENDIX I: EXPERIMENTAL TREATMENTS AND QUESTIONNAIRE ......................... 97 APPENDIX II: ETHICAL CLEARANCE ................................................................................. 115 vi University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Table 5.1: Perceived product gender ............................................................................................ 49 Table 5.2: Initial Factor Solution .................................................................................................. 51 Table 5.3: Result of Final Factor Solution .................................................................................... 54 Table 5.4: Result of Reliable Tests ............................................................................................... 56 Table 5.5: Result of Discriminant Validity Test ........................................................................... 59 Table 5.6: Participant’s Response to Relevant Constructs ........................................................... 60 Table 5.7: Relationship between Endorser Gender and Ad Believability .................................... 62 Table 5.8: Model Summary for hypothesis one ............................................................................ 63 Table 5.9: ANOVA table for hypothesis one ................................................................................ 63 Table 5.10: TABLE 10: Coefficients table for hypothesis one .................................................... 64 Table 5.11: Model Summary for hypothesis two .......................................................................... 64 Table 5.12: ANOVA table for hypothesis two ............................................................................. 65 Table 5.13: Coefficients table for hypothesis two ........................................................................ 65 Table 5.14: Model Summary for hypothesis three ........................................................................ 66 Table 5.15: ANOVA table for hypothesis three ........................................................................... 66 Table 5.16: Coefficients table for hypothesis three ...................................................................... 67 Table 5.17: Model Summary for hypothesis four ......................................................................... 67 Table 5.18: ANOVA table for hypothesis four ............................................................................. 68 Table 5.19: Coefficients table for hypothesis four ....................................................................... 68 Table 5.20: Summary Of The Hypotheses Test Results ............................................................... 69 vii University of Ghana http://ugspace.ug.edu.gh LIST OF FIGURES Figure 3.1: The researcher’s Conceptual model describing the relationship between the variables under investigation ........................................................................................................................ 32 Figure 5.1: Age ............................................................................................................................. 47 Figure 5.2: Level of Education ..................................................................................................... 47 Figure 5.3: Participants’ Gender ................................................................................................... 48 Figure 5.4: Empirical Model ......................................................................................................... 70 viii University of Ghana http://ugspace.ug.edu.gh ABSTRACT One of the exciting ways of telling stories about brands is by leveraging celebrity fame, admiration and love to enhance effective communication. This strategy is called celebrity endorsement. It has become a reliable formula with which brand communicators break through the clutter of many advertisements and other similar promotional communications to reach their targets. This study set out to investigate implications of gender in celebrity endorsed advertisements. Based on assumptions of the source credibility model and the match-up hypothesis model, a 2×3 factorial within-subject experimental design was adopted to test participants’ perception of gender effects in celebrity endorsed advertisements. The following findings were revealed: Gender of a celebrity endorser has consequences for advertisement believability depending on the nature and gender of the advertised product. In addition, the study saw that respondents’ perception of a match between celebrity endorser gender and product gender increased advertisement believability. Moreover, the researcher discovered that even when the gender of the endorser and the product were inconsistent, participants still perceived a good fit and this resulted in a favorable evaluation of the advertisements. Informed by these findings, the study concluded that a negative match between an endorser’s gender and a product’s gender does not necessarily lead to a negative or positive evaluation of the endorsement. Actually, what matters is whether audiences perceive a fit. Thus, audiences’ perception of endorser gender and product gender fit directly relates to a favorable assessment of the endorsed advertisement. Based on these conclusions, the researcher suggested that future studies should look more into factors which inspire audience members’ definitions of endorser gender and product gender fit. ix University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION 1.0 Background to the Study Advertising is an element in the promotional mix which brand communicators use to communicate to their targets (Samarashinghe, 2018). As the media get increasingly cluttered with countless advertisements, the need for brand communicators and marketers to make their brands stand out has become crucial (Klaus & Bailey, 2008). To do this, advertising practitioners resort to a number of creative strategies. These strategies include: Dramatisation (where the benefits and attributes of a brand are presented in a short story format), slice of life (where a real life problem is shown in a drama and the advertised product represents the solution to the problem), testimonials (where existing brand users are made to share their brand experiences and satisfaction to persuade others to also use the brand) and comparisons (where the focal brand is compared with its strongest competitor). Of all the creative execution options, one that is integral to modern-day advertising is celebrity endorsement (Magaji, 2012). Celebrity endorsement (CE) allows brand communicators to make their brands shout over the deafening clutter of other brands (Samarasinghe, 2018; Gupta, 2017). There are two forms of endorsements: celebrity endorsement and non-celebrity endorsement (Parmar, 2017). Celebrity endorsement is a kind of advertisement in which a celebrity is used as spokesperson for a product whereas non-celebrity endorsement denotes a kind of advertisement in which a person who is not a celebrity, is made to speak on behalf of a product. Over time, celebrity endorsement has become popular because of its proven capability of securing visibility for brands. Thus, the phenomenon has gotten a lot of research attention in various settings. 1 University of Ghana http://ugspace.ug.edu.gh According to McCracken (1986) a celebrity endorser refers to anyone who is favorably recognized by the public and who leverages this public approval by appearing with a product in an advertisement. The idea is for the celebrity to transfer his or her reputation as being for example strong, dependable, credible etc. onto the advertised product. It is again assumed that through the endorsement process, audiences will perceive the endorsed product the same way they perceive the celebrity in question. The term celebrity endorsement denotes any advertising communication such as a vocal statement or depictions of the name or signature of a celebrity which audiences believe reflect the opinion, verdict or knowledge of the celebrity instead of the sponsoring advertiser (Samarashinghe, 2018). Mittal (2018) also defines celebrity endorsement as a type of advertising tool where a celebrity declares support, approval and patronage in favor of a particular product and thus, influences the behavior of the masses so that they buy the products. Celebrity endorsement may manifest in diverse forms including explicit (I endorse this detergent); implicit (I like this drink); imperative (you should listen to Adom Praise every Sunday evening) or co- presentational where the celebrity simply appears with the product (Seno & Lukas, 2007). Studies have shown that when used carefully, celebrity endorsement positively affects consumer attitudes; buying decisions; brand awareness and ad recall (Lartey, 2017; Kumar, 2010; Shimps, 2010; Khattri, 2006; Erdogan, 2001; Amos et al 2000). Accordingly, Lazar (2020) argues that celebrity endorsers provide a set of characteristics that supports consumers in evaluating the presented brand. 2 University of Ghana http://ugspace.ug.edu.gh In Ghana, celebrity endorsement advertising has become very common and is thought to be a successful way of communicating about brands (Nyarko et al., 2015). In fact, it is very easy to see both male and female Ghanaian celebrities endorse different products in TV commercials, on radio and billboards across major streets in the country. It is also very common to find celebrities endorse brands that either target a single gender audience or a general audience. For example, renowned movie stars, Akwasi Boadi (popularly known as Akorobeto) and David Dontor endorsed ‘Proper sanitary pad’, a product which targets women. Likewise, Nana Ama McBrown, a Kumawood- Ghallywood actress, endorsed television sets and speakers from Hisense (These products targeted men. The ad specifically appealed to men to be smart to buy the products so that they could enjoy the 2019 Africa Cup of Nations). This notwithstanding, scholars have found that not every celebrity can endorse any product. Indeed the literature suggests a need for an endorser and product fit for audiences to perceive the message as believable (Lien, 2012). Indeed, existing literature stresses the importance of gender (of both endorser and product) to effective celebrity endorsement (Wolin, 2003). For example, Klaus and Bailey (2008) have argued that gender is very important to celebrity endorsement because traditionally, the basis of segmentation and targeting in brand communication has been gender. Moreover, Dusenburg et al. (2016) have articulated that gender is one of the most salient factors that determine change of attitude regarding effects of communication messages. In the views of Peetz et al. (2004), gender plays a crucial part in the transfer of meaning from celebrity endorsers to endorsed products especially for members of generation Y (born after 1970) market segment. 3 University of Ghana http://ugspace.ug.edu.gh The above arguments highlight endorser-product fit as a key consideration by creative directors in selecting celebrity endorsers. In other words, creative directors would select celebrities who would be trusted and liked by their target audience. Mishra et al. (2001), contend that gender influences audiences’ perceptions about male and female celebrity endorsers as either likable or credible. Support for such arguments has been seen in sports brand communication where the gender of celebrity endorsers is found to influence the perceived quality of an endorsed product which in turn influences target audience’s intention to purchase (Dusenburg et al., 2016). Interestingly, conclusions from studies which specifically investigated effect of gender in the endorsement process appear to be mixed. Whereas one school of thought agrees that the construct gender has some kind of influence on the success or otherwise of celebrity endorsement, another group of scholars contend that gender has no influence in the endorsement process (Lazar, 2020; Klaus & Bailey, 2008; Phan & De Run, 2007). Even among the group that believes gender has a role to play, the particular direction (whether from the celebrity’s gender, the audience member’s gender or both sides) that the impact emanates is also mixed. For example, Ohanain (1991) concluded that there is no correlation between the gender of celebrity spokespersons and audiences’ perception of the endorsed brand. Phan and De Run (2007) also concluded that women had no reaction toward an ad with or without a celebrity endorser even when the product involved is supposed to have a feminine gendered image. Carsky and Zuckerman (1991) also concluded that variations in advertising trustworthiness among participants were not linked with endorser gender. Lazar (2020) and Klause and Bailey (2008) on the other hand, claim that gender of the endorser significantly affects consumers’ attitude towards the brand. 4 University of Ghana http://ugspace.ug.edu.gh Yet, another strand of research has reported that celebrity endorsement becomes powerful when the gender of the endorser blends perfectly with the gender image of the endorsed product. For instance, Hudston et al. (2005) and Lien (2012) have demonstrated that a match between the endorser gender and the advertised product culminates in ad recall by audiences. Some products for example, have a feminine gender image and can be effectively endorsed by female endorsers (Lien, 2012). Furthermore, products derive their gender from the gender of their target users. Such discoveries clearly indicate that there is no one-size-fits-all in the selection of celebrities to appear with products. This is especially so when the selection of celebrities for endorsement when done haphazardly, could have huge cost consequence (Boyd & Shank, 2004). 1.1 Problem Statement There is an increasing body of knowledge on celebrity endorsement which has examined its influence on brands and audience perceptions and attitudes (Samarashinghe, 2018; Dusenburg et al., 2016; Lien, 2012). In Ghana, this interest has led researchers to explore links between celebrity endorsement (of indigenous brands) and Consumer Response (Lartey, 2017); product competitiveness of telecom brands (Agyeman Duah, 2016); youth buying behavior (Nyarko et al., 2015); periurban consumer behaviour (Owusu-Mensah, 201); purchase intension towards electrical appliances (Ofosu-Boateng, 2020); young voters’ choice of political candidates (Asomaning & Ababio, 2020) and promotion of banking services (Mensah, 2020) among others. 5 University of Ghana http://ugspace.ug.edu.gh Both global and Ghanaian literature agree that celebrity endorsement is effective in securing benefits for brands. However, there are disagreements as to the nature of its relationship with other variables including, celebrity endorser gender and product gender (Wolin, 2001). In addition, it appears celebrity endorsement has not been looked at from the perspective of how endorser gender and product gender fit influence perception of the endorsement itself. Meanwhile, there have been quite a number of researches on gender in advertising in Ghana, most of which have centered on representations of men and women in Ghanaian advertisements (Diabah, 2018) and how these representations reinforce traditional gender stereotypes (Adzorvie et al., 2019). However, from the literature (on advertising generally and celebrity endorsement particularly), it appears gender has not been explored from the perspective of how it influences audience perceptions of advertisements. Even though such studies have been conducted elsewhere, it is important to note that gender, which is the cut across theme in these researches, is a cultural phenomenon. This means, the understanding and definition of gender varies from countries. This therefore suggests that, it will be quite problematic to use the gender situation of one society to make claims in another society. Hence the need for the current study to understand how the unique gender constructions and perceptions in Ghana shape appreciation of celebrity endorsed advertisements. It would appear therefore that part of the missing link in the literature is whether fit between a celebrity endorser’s gender and the advertised product makes audiences perceive celebrity endorsement differently. Moreover, the high power distance culture of the Ghanaian society (Ghana* Hofstede Insights, 2020) suggests that majority of Ghanaians respect ranks and power differences among individuals. 6 University of Ghana http://ugspace.ug.edu.gh For this reason, Ghanaians may be described as having a higher inclination to respect opinions of celebrities given the fact that most of these celebrities occupy a relatively high status in the country. Most of these celebrities are even imitated in terms of their dressing, speaking among others. It would therefore be interesting to test how this relationship (between celebrities and audiences) may determine the way celebrity endorsement is appreciated by the Ghanaian advertising audience (Osei-Frimpong et al. 2019). For these reasons, this study seeks to contribute to celebrity endorsement literature in Ghana by examining how gender of a celebrity endorser and the perceived gender of the advertised product affect audiences’ perception of celebrity endorsed advertisements. 1.2 Rationale of the study The question with which this study was begun was whether Ghanaian advertising audiences are influenced by celebrity endorsement advertisements. After engaging literature, it was realized this question has been answered. However, the issue of whether the gender of the celebrity doing the endorsement, makes the audience perceive him or her as the right person to talk about that product appears to be missing. This study was therefore conducted to help provide some answers to this issue and also contribute to the celebrity endorsement literature in Ghana. 1.3 Objectives of the Study The goal of this thesis was to investigate implications of gender on believability of celebrity endorsed advertisements. Specifically, the following objectives were set: § To examine the nature of the relationship between endorser gender and advertisement believability. By this, the study investigated: 7 University of Ghana http://ugspace.ug.edu.gh A. Whether endorsement of feminine products by female endorsers enhanced ad believability B. Whether endorsement of a masculine product by male endorsers enhanced ad believability C. Whether endorsement of a feminine product by male endorsers enhanced ad believability D. Whether endorsement of a masculine product by female endorsers enhanced ad believability E. Whether endorsement of a neutral product by female endorsers enhanced ad believability F. Whether endorsement of a neutral product by a male endorser enhanced ad believability § To discover whether audiences’ perceptions of fit between celebrity endorser gender and product gender influence their believability of the advertisement. 1.4 Research Questions To adequately investigate the problem understudy, these questions were designed to direct the researcher: } What is the nature of the relationship between gender of celebrity endorsers and audiences’ believability of an ad? In other words: A. How will endorsement of a feminine product by female endorsers enhance advertisement believability? 8 University of Ghana http://ugspace.ug.edu.gh B. How will endorsement of a masculine product by male endorsers enhance advertisement believability? C. How will endorsement of a feminine product by male endorsers enhance advertisement believability? D. How will endorsement of a masculine product by female endorsers enhance advertisement believability? E. How will endorsement of a neutral product by female endorsers enhance advertisement believability? F. How will endorsement of a neutral product by male endorsers enhance advertisement believability? } How does the perceived fit between a celebrity endorser’s gender and the gender of the advertised product shape audience’ advertisement believability? 1.5 Significance of Study Conclusions from existing literature regarding implications of gender in celebrity endorsement are equivocal (Peetz et al., 2001; Phan & De Run, 2007). Even the assertion that celebrity endorser gender-product gender match is a key determinant of audience perceptions about celebrity endorsed advertisements is not without controversy. The current study therefore, will help discover a Ghanaian perspective with respect to these controversies in literature. Also, this study is important because it will contribute to the building of literature on celebrity advertising in Ghana. Moreover, findings of the study will provide useful information to brand communicators and advertisers in Ghana in making the best out of the endorsement process by selecting suitable celebrity endorsers. Also, the source credibility model and the match-up hypothesis model which 9 University of Ghana http://ugspace.ug.edu.gh form the theoretical foundations of the study have gained massive support in celebrity endorsement literature; the current study will thus test whether assumptions of these models would be validated in the Ghanaian context as well. 1.6 Organization of Study This thesis is presented in seven chapters. Chapter one introduces the research. It also states the problem, the research objectives and questions as well as the significance of the work. The second chapter discusses the theoretical framework of the work whereas the third section presents a review of related studies, as well as the research hypotheses and the conceptual model of the study. In the fourth chapter, the methodology of the study is explained and justified. The fifth chapter contains analyses and findings of the study while the sixth chapter presents a discussion around the analysed data. In the seventh and final chapter, relevant conclusions are drawn from the outcomes of the discussion. Moreover, suggestions of what could be improved in the future are outlined in the seventh chapter. 10 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO THEORETICAL FRAMEWORK 2.0 Introduction This part of the work attempts an explanation for the models used and explicates their relevance to the work. The study has combined two theoretical concepts: Source credibility model (Hovland et al., 1953) and the Match-up hypothesis model (Kahle & Homer, 1985). 2.1 Source Credibility Model The source credibility model is a popular communication model that provides a framework with which people evaluate a storyteller’s eligibility (Kang et al., 2019). The model, which could be likened to the Aristotelian principle of Ethos, argues that who speaks matters and the message is likely to be accepted if the motives and intentions of the speaker is trusted (Hocever, Metzger & Flanagin, 2017; Erdogan 2001). Basically, the model describes how audiences perceive celebrity endorsers to have relevant knowledge, skills and experience and trust the celebrities to provide truthful information about the products they endorse. Source credibility is a term usually used to denote a speaker’s positive features which influence a receiver’s approval of a message (Ohanian, 1990). It is important to highlight the point that in celebrity endorsement, the celebrity endorser assumes the place of the source of the message. Thus, even though the message is crafted by a sponsoring advertiser, audiences perceive that the message is from the celebrity. In the case of source credibility model, the celebrity endorser is referred to as the source of the message. The source credibility model came out of a landmark study by Hovland, Janis and Kelley (1953). In their study, Hoveland and his companions explored the factors affecting perceived credibility of the communicator. They concluded that expertise and trustworthiness highlight source credibility. 11 University of Ghana http://ugspace.ug.edu.gh Regarding this thesis, the model would be useful in explaining how the gender of celebrity endorsers inform audience perception of whether or not these celebrities possess relevant characteristics to make valid assertions about the product they are advertising. The model claims that the success of a message depends on the supposed “expertise” and “trustworthiness” of the source of that message. According to Ohanian (1990) and Erdogan (2001) expertise refers to the point at which a communicator is seen as a source of valid assertions, whereas trustworthiness refers to the listener’s degree of confidence in, and level of acceptance of the speaker and the message. Similarly, McCracken (1985) argued that trustworthiness looks at the level of trust, confidence and believability that audiences attach to the celebrity endorser because he or she is perceived to be honest, sincere, and dependable. It denotes the supposed willingness of the celebrity endorser to make valid assertions. Expertise, however, refers to consumers’ perception of how knowledgeable the endorser is with respect to the product he or she is endorsing. It is the perceived ability of the endorser to make valid assertions. According to Gupta (2017), audiences dispute unreliable celebrity endorsers notwithstanding their other qualities like fame, likability and attractiveness. The source credibility model also suggests that information from a reliable source influences the opinion, attitude and behavior of consumers who internalize this information because they believe the source is providing them with accurate information. Thus consumers tend to be persuaded more by a source they perceive is credible. Other studies (Cunha & Barbosa, 2016; Mishra, 2015 and Martey, 2014) have shown that expert sources inspire perceptions of the product’s worth. For example, Stanley et al. (2011) have argued that when a male and a female celebrity endorse the 12 University of Ghana http://ugspace.ug.edu.gh same product, audiences perceive the message differently because of the endorser’s gender. According to the model, perceptions of endorser credibility are formed on the basis of some characteristics of the endorser that the audiences believe qualifies him or her (endorser) to be able to make helpful and truthful assertions about the product in question. In the case of this study, a celebrity endorser’s gender could be a basis for him or her to be perceived as credible. Evidence from existing literature shows extensive support for the source credibility model. Interestingly however, the evidence tells that a trustworthy communicator is seen to be more persuasive at changing audiences’ opinion even when he or she is perceived not to be an expert. Notwithstanding, information from a credible source (trustworthy and expert) has been found to influence audiences’ principles, views, attitudes and/or performance (Erdogan, 1999). It is obvious from the forgoing discussions, that the source credibility model alone is inadequate in addressing all the aspects of the study. For instance, the model does not provide any extensive explanation for endorser-product congruence, which is an important variable that this thesis seeks to examine. Therefore, to be to have sufficient theoretical understanding of endorser-product congruence, the researcher has added to the source credibility model, another model, called, the match-up hypothesis model. This model is discussed in the preceding paragraphs. 2.2 Match-Up Hypothesis Model This model demonstrates a direct correlation between celebrity and product match and favourable audience perceptions (Elandsson et al., 2013). Simply put, the match-up hypothesis model claims that positive impact of the endorsement process is dependent on consistent characteristics between an endorser and a product. Earlier articulations (Kotler, 1999; Kamins and Gupta, 1994 and Misra and Beaty, 1990) on this model reveal that the match between celebrity and brand depends on the 13 University of Ghana http://ugspace.ug.edu.gh degree of perceived 'fit' (from the researchers’ perspective) between brand and celebrity image. It is also suggested in the match-up hypothesis model that when endorser image matches the image of the product, it leads to higher celebrity endorser believability (Erdogan, 1990). Sometimes, the endorser and product match is perceived as fit between the celebrity’s personality and the attributes with which the product has been positioned (Erlandsson, 2013; Premeaux, 2009). For instance, this type of match could be achieved when a celebrity who has an elegant personality is chosen to endorse a product which has been positioned as elegant. At other times, the fit is professed to be between the gender image of the endorser as well as the perceived gender image of the product (Petrevu, 2001). For instance, Petrevu (2001) argues that when a female celebrity endorses feminine gendered products (such as a bra, cosmetics etc.), audiences believe her more because she is a woman and can personally experience the product. Consequently, this moderates audiences’ perception of the advertisement. For the purposes of this study, endorser image-product image fit is looked at from the perspective of gender. The model is therefore relevant in the sense that it would help explain the second research question of the study which seeks to investigate the extent to which perceived fit of the advertised product’s gender and a celebrity endorser’s gender influence audiences believability of the advertisement. 2.3 Chapter Summary This chapter discussed the two models (Source credibility model, Match-up hypothesis model) used as the theoretical basis of the study. Attempts were made to draw out the link between the models and the study. The Source credibility model for example is important for this study because it provides a foundation for understanding how certain attributes of a message source make the 14 University of Ghana http://ugspace.ug.edu.gh source be regarded appropriate to communicate the message in question. On the other hand, the match-up hypothesis model aptly supports this thesis in hypothesising that perceived gender match between a celebrity endorser and the product he or she is endorsing increases believability of the advertisement. 15 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE REVIEW OF RELATED STUDIES 3.0 Introduction Increasing competition among brands for consumer attention is a fact in today’s world (Lartey, 2017). Advertisers and brand communicators have resorted to numerous creative strategies to make their brands stand out (Hussain, 2019; Belch & Belch, 2001). Celebrity endorsement is one of such strategies which is most frequently used to increase consumer purchasing motivation (Garud, 2013). At present, the strategy is seen to be an integral part of any marketing communication strategy (Fathi & Kheiri, 2015). For this reason, the phenomenon has attracted a wide scholarly attention across the globe. Demographic variables, particularly gender (of the celebrity endorser and the advertised product) have also been found as greatly influencing the strategy. Generally, celebrity endorsement research is categorized into two dimensions: source studies and congruence studies. The source studies advocate that endorser peculiarities influence efficacy of the endorsement whereas congruence studies state that celebrities’ match with the brand and sometimes, the audience shape the efficiency of the endorsement. Lately, celebrity endorsement studies (including the current study) have started to combine source and congruence related phenomena to fully explore a celebrity’s effectiveness. A key purpose for this development is the recognition of the role celebrity characteristics (such as gender) play in generating endorser-product correspondence and how this correspondence inspires celebrity effectiveness (Thomas & Johnson, 2019). 16 University of Ghana http://ugspace.ug.edu.gh This chapter reviews studies related to gender in celebrity endorsement and advertising believability. The review is organized around five distinct themes derived from the research topic. These are: 1. Celebrity endorsement as a brand communication strategy 2. Celebrity endorsement in Ghana 3. Gender implications in celebrity endorsement 4. Endorser and product congruence 5. Advertisement believability One of the interesting discoveries made from the analysis of the related studies is that even though celebrity endorsement is indeed a useful tool for clutter breaking advertising, the strategy is not necessarily foolproof. 3.1 Celebrity Endorsements as a Brand Communication Strategy As a promotional device, celebrity endorsement allows advertisers to use celebrities as spokespersons for their products. The concept has also been investigated in countless geographical boundaries as well as different fields such as sports, entertainment, technology and fashion, among others. For instance in Nigeria, MacKingsley (2016) investigated how celebrity endorsement impacts brand loyalty and found a strong positive association between celebrity endorsement and brand loyalty. The study revealed that attractive, trustworthy and expert celebrities have stronger persuasive influence on consumers and this enhances their loyalty to the brand. This finding confirms Magaji’s (2012) argument that celebrity endorsement enables brand communicators to make their brands shout over the deafening clutter of other brands. Additionally, Sridevi (2012) made discoveries on the phenomenon in India which validated McKingsley’s (2012) findings. He 17 University of Ghana http://ugspace.ug.edu.gh found that celebrity endorsement enhances product information and influences purchase attitude change of consumers. Similarly, Soni (2006) reported from India that celebrity endorsed advertisements capture audiences’ attention better than non-celebrity endorsed ads. Interestingly, the study revealed that both male and female respondents believe that celebrity endorsed ads inform them better. This finding confirms Erlandsson, Handson and Mokhatri’s (2013) argument that celebrity endorsement exerts high persuasive impact on audiences. In their study also, Fink et al. (2012) assigned a 2×2 full factorial experimental design to 279 undergraduate Sport Management students from four universities across the United States. They argued that celebrity endorsement helps advertisers to pitch their products to consumers. This study by Fink et al. (2012) also confirms the studies by MacKinsley (2012) and Soni (2006) regarding impact of celebrity endorsement as a promotional tool. Some scholars have however countered evidence regarding the massive effectiveness of celebrity endorsement in product promotion (Hussain & Khan, 2019; Ansu-Mensah et al. 2013; Phang & de Run, 2012). These scholars have argued that the benefits of celebrity endorsement have been hyperbolized. According to them, celebrity endorsement does not automatically lead to purchase of product. They have further argued that other ‘exogenous’ factors such as brand name, price, packaging, among others often add weight to the celebrity endorsement strategy to influence purchase. In their study, Phang and de Run (2012) even found that celebrity endorsement, unlike what has been found in previous studies, did not necessarily influence positive advertisement ratings and product evaluation. Moreover, the researchers realised that among male respondents, 18 University of Ghana http://ugspace.ug.edu.gh there was no variance in their reaction towards an advertisement with or without a celebrity. Similarly, Soni (2006) reported that male respondents did not see celebrity endorsement to be as impactful as was recorded about their female counterparts. Furthetmore, studies have shown that celebrity endorsement does not work effectively in every setting. Silvera and Austad (2003) argued that cultural differences exist which moderate the impact of celebrity endorsement. They argued that Americans strongly esteem celebrities and are therefore keener to accept and even adopt endorsement messages. Also, other countries such as China and India are often referred to as celebrity obsessed nations (Kasana 2014). On the contrary, Norwegian cultural standards are hostile toward celebrities. As such, celebrities in Norway earn certain amounts of “grudging respect” for their achievements but they are also distrusted because of the assumption that they pride themselves over the common man (Silvera & Austad, 2003). Moreover, it has been argued that there are certain factors that pose a challenge to celebrity endorsement. According to Haefele (2004), aligning a celebrity’s brand with a company’s own brand could be dangerous. Also, Gupta (2017) asserted that celebrity endorsement may ricochet if there is mismatch between the celebrity and the product’s image. In addition, research evidence describes the strategy as tricky sometimes. This is because a celebrity’s image can overshadow the brand. This situation is often referred to as a ‘vampire effect’ (Erfgen, Zenker, & Sattler, 2015). A study by Garud (2013) revealed instances where respondents remembered a celebrity and forgot about the endorsed product. Furthermore, celebrity endorsement could tarnish the image of a brand especially when the celebrity chosen gets involved in a scandal. In their study on how gender control perception of negative information related to celebrity endorsement, Edward and La Ferle 19 University of Ghana http://ugspace.ug.edu.gh (2009) argued that adverse and controversial conduct of a celebrity could reflect badly on the product they endorse. According to Kasana (2014), despite upsides of the celebrity endorsement strategy, the phenomenon increases a marketer‘s risk of cost and image problems. She argued that advertisers need to be cautious when employing celebrities to ensure promise believability and delivery of the intended effect. This presupposes that the presence of celebrities can influence how audience members process feasibility of the promises given regarding the endorsed product or service in a celebrity endorsed advertisement. 3.2 Celebrity Endorsement in Ghana In Ghana, it appears difficult to find evidence on how a match between the gender of the celebrity endorser and the product which is being advertised correlates with advertisement effectiveness indicators such as ad recall, brand recall, purchase intention and ad believability. Nonetheless, there is agreement that celebrity endorsement is an effective communication strategy in making brands more attractive to audiences. According to Lartey (2017), Ghanaian consumers have an interest in celebrities and this influences their engagements with celebrity endorsed ads even though this does not necessarily impact their purchase of the endorsed products. On the contrary, Osei-Frimpong et al. (2019) have reported that celebrity endorsers who are associated with source characteristics such as attractiveness, trustworthiness and familiarity have a positive effect on audiences’ insight of product quality, purchase intention and brand recall. Likewise, Nyarko et al. (2015) conducted a survey among tertiary students on the influence of celebrity endorsement. The study’s findings replicated what has been found in other places. It revealed that Ghanaian youth perceive celebrity endorsement as more attractive and impactful as 20 University of Ghana http://ugspace.ug.edu.gh compared to non-celebrity endorsements. Zakari et al. (2019) have argued that source characteristics like attractiveness, likeability and trustworthiness of celebrity endorsers positively relate with effectiveness of the endorsement as well as the reputation of the organization to which the endorsed brand belong. Similarly, Entsiful et al. (2013) also discovered that student consumers are motivated by celebrity endorsement to make purchases. They found that students attach class and quality to celebrity endorsed brands and products. Also, the researchers saw that celebrity influences brand choice among students. Interestingly, the study revealed that inasmuch as celebrity endorsement influences student consumers’ purchase decision, most students would buy a brand regardless of whether a celebrity endorsed it or not. Likewise, Eli-Zafoe (2013) investigated the effect of celebrity endorsements on buying behaviour in the telecommunications industry in Ghana. The researcher found that the telecommunication companies decide on about which celebrity to choose on the basis of the following factors: celebrities who are of good behaviour, and celebrities who are famous and are loved by Ghanaians and celebrities whose personalities can be aligned with the companies’ objective. These findings clearly confirm assumptions of the source credibility model as well as the match-up hypothesis model. In relation to the above, Dzisah and Ocloo (2013) examined the factors considered in selecting a celebrity endorser. They discovered something similar to Eli- Zafoe’s (2013) results. Their results indicated that organisations choose celebrities based on credibility, brand match and attractiveness whiles consumers’ are influenced by price, celebrity 21 University of Ghana http://ugspace.ug.edu.gh endorsement, quality and packaging. The researchers argued that there is a positive link between celebrity endorsement and consumer purchase behaviour in Ghana. 3.3 Implications of Gender in Celebrity Endorsement Different views exist regarding the effects of gender in celebrity endorsement. For instance, using a pre-post test experimental design, Edward and La Ferle (2009) randomly assigned 135 male and female college students in the United States to information about either a female or a male celebrity. The researchers concluded that an endorser’s gender affected how audience members evaluated the advertisement. Accordingly, Klause and Bailey (2008) assessed the impact of the gender of the celebrity and the gender of the audience of a celebrity endorsed advertisement on celebrity endorsement evaluation. Four hypotheses were tested in an experimental study all of which were supported. The findings clearly indicated that there is a differential response to adverts featuring male and female celebrities. The researchers also reported that female celebrity endorsers are rated more favorably than male celebrity endorsers. Other scholars such as Lazar (2020), Dusenberg et al. (2016) and Mishra et al. (2001) have reported that men and women differ in their assessment of celebrity endorsements. For instance, Mishra et al. (2001) performed a quasi-experimental procedure on 200 male and female adolescents from seven cities in India. At the end of the study, the researchers saw that the male and female participants had different perceptions about the advertisements they were presented with. On the contrary, some scholars including Phan and De Run (2007) argue that gender does not affect 22 University of Ghana http://ugspace.ug.edu.gh celebrity endorsement in any way. However, considering Silvera and Austad’s (2003) argument that cultural differences moderate the impact of celebrity endorsement, one would say these different reports with respect to gender’s role in celebrity endorsement, may well be as a result of differences in perspectives about gender within the different research settings. 3.4 Endorser and Product Congruence A crucial element in the celebrity endorsement process is the ‘fit’ between the celebrity and the advertised product (Levert, 2018). This fit is achieved when “highly relevant characteristics of the spokesperson are consistent with highly relevant attributes of the product or brand” (Misra & Beatty, 1990, p160). Existing literature suggest that celebrity endorsement works best when celebrity endorsers are selected to appear with particular products. According to Fleck- Dousteyssier and Korchia (2006) a brand/celebrity pair is important if it is meaningful and contributes to clear communication of a message about the brand. The researchers applied a 2 × 4 × 2 mixed factorial design, with a separate control group to test the levels of congruence between celebrities and products. The study which was conducted in France, also revealed that the pair between a celebrity and a product is expected if it agrees to a predefined expectation about the brand. As explained by Choi and Rifon, (2012) consumers tend to believe that the image of a celebrity endorser must blend perfectly with the image of the endorsed product. For example, sporting goods are best endorsed by sportsmen. Choi and Rifon (2012) conducted an experiment on participants sampled from Midwestern University in the United States to examine congruence effects of audience and celebrity as well as product and celebrity. The researchers concluded that congruence 23 University of Ghana http://ugspace.ug.edu.gh between the celebrity and the product has an impact on participants’ attitude toward the advertisement. In their study, Davies and Slater (2015) relied on the match-up hypothesis model to test endorsement effectiveness through sponsorship events. The researchers hypothesized that a high sponsor-event fit would generate a favourable response to the sponsorship. With the help of a case study method and a survey, Davies and Slater (2015) concluded that a perfect celebrity- product permutation makes it easy for the cultural meanings residing in a celebrity’s brand image to be transferred onto the endorsed product for a maximum effect. Similar to Choi and Rifon’s (2012) conclusion, Davies and Slater (2015) also reported that celebrity endorsement is enhanced when there is a fit between the endorser and the product being endorsed. Mishra, Roy and Bailey (2015) have argued cogently that improper pairing of a celebrity with a brand could have negative consequences for the brand. This argument was based on a study they conducted to test brand personality–celebrity endorser personality congruence and its impact on consumers’ attitudes and intentions. The study drew on social adaptation and attribution theories to develop its hypotheses. These hypotheses were tested based on data acquired from a sample of participants in India. Results from the analyses indicated a positive and significant impact of personality-based congruence between a celebrity endorser and the endorsed brand on endorser credibility and suitability. According to Mishra, Roy and Bailey (2015), this significant relationship also has impact on audiences’ ad believability. One may conclude from the discussion above that the concept of advertising a product via a celebrity who has relatively high product congruent image has come to occupy a central position in advertising scholarship and practice essentially because it has been found as generating greater advertisement believability (Erdogan, 1999) as well as reducing vampire effect, a situation where 24 University of Ghana http://ugspace.ug.edu.gh the celebrity appearing with a product grabs the attention from the brand and the ad message to itself thereby preventing the audiences from remembering the message (Erfgen et al. 2015). According to Thomas and Johnson (2019) a better congruence between a celebrity and a brand creates a better attitude towards the advertisement and brand. In view of this, a number of endorser- product fit dimensions have been discussed (Lee & Thorson, 2008). These fit dimensions or congruence metrics include; personality, stage persona and gender. Personality based congruence between a celebrity and a product is created when the human attributes with which a product has been positioned match the personality attributes of the endorser (Thomas & Johnson, 2019). For example within this fit dimension, a product positioned as sophisticated is best endorsed by a celebrity with an air of sophistication; in other words, a high- end celebrity rather than a low-end celebrity (Khan, Memon & Kumar, 2019; Mishra, Roy & Bailey, 2015). In an attempt to bring this concept home to Ghana one would cite an example of a celebrity like Ivonne Okoro (a high-end Ghallywood actor) as being be more suitable to endorse a brand like the Accra Marriot hotel, the first of the top ten luxurious and sophisticated hotels in Accra, other than a celebrity like Rose Mensah (Kyeiwaa, a low-end Kumawood actor). At other times, celebrities belonging to certain professions such as acting, journalism, sports among others are selected to endorse products relating to their profession so as to achieve a fit. This fit category is termed as stage persona (Thomas & Johnson, 2019). Examples could be Pastor Mensah Otabil (a renowned preacher) endorsing an anointing oil or Nana Aba Anamoah (a respected journalist) endorsing a journalism school. As well, gender based congruence is achieved when celebrities are made to endorse products which relate to their gender. Among these congruence metrics however, 25 University of Ghana http://ugspace.ug.edu.gh gender appears to be one that has sparked a lot of controversies in literature (Thomas & Johnson, 2019; Samarashinghe, 2018). While it sounds logical that high celebrity-product congruence leads to better ad evaluation, some scholars (Levert, 2018; Lee, 2013; Lee & Thompson 2008) have reported that this positive fit is not always necessary. According to these researchers, a relatively poor fit or mismatch between a celebrity endorser and a product does not necessarily lead to negative evaluation of the advertisement. On the contrary, it stimulates audiences to pay more attention to the message. In his study, Lee (2013) proposed an additional method for manipulating ad message involvement (AMI). Performing a 2×2 between-subject experimental procedures on a sample drawn from 80 undergraduate students from the national university of Singapore, Lee (2013) reported that unexpected information (for example a mismatch between an endorser and a product) creates relatively persistent and favorable attitudes among audiences. Likewise, Lee and Thorson (2008) conducted experiment on 318 adults from different states in America to examine how different degrees of celebrity-product incongruence influence the persuasiveness of an advertisement. The researchers found that a moderate mismatch and not a complete match between a celebrity’s image and a product’s image produced favorable responses to advertisements. Nevertheless, Meyers- Levy and Tybout (1989) have reported that the problem with these ‘abnormal’ celebrity-product fit permutations is that they must be perceived by audience members as interesting and positive before they work. The researchers conducted three separate experiments on participants ranging from an upper middle class suburb of Chicago to middle-aged managers and graduate students in America. The purpose of their study was to test whether endorser-product incongruence has a positive impact on advertisement evaluation. In D’Astous and Bitz, (1996) affirms the argument 26 University of Ghana http://ugspace.ug.edu.gh by Meyers-Levy and Tybout (1989), by stating that if the incongruence between an endorser and a product is so great, that consumers have to alter their intellectual constructions to understand it, they have a tendency to demonstrate negative reactions, which leads to negative cognitive elaboration and frustration In a study which investigated match-up effect between a spokesperson’s gender type and a product’s image, Lien (2012) found that, for totally feminine or masculine gendered products for example, the spokespersons’ sex has a substantial advertising effect. That is if the target audience is male, a male spokesperson is better but if the target is female, a female spokesperson is better Boyd and Shank (2004) support this assertion in their argument that an ad message is more effective when there is a match-up between endorser’s gender and the gender image of the endorsed product. Misra and Beatty (1990) also examined the effect of endorser/brand image congruity on brand recall and found that congruity of endorser and brand image significantly impact audience brand recall. This argument was confirmed by Boyd and Shank (2004). However, their study revealed that irrespective of product type, subjects rate the endorsers of the same gender as more trustworthy- a revelation which supports the findings of Edward and La Ferle (2009). Moreover, Kumar and Singh (2015) have argued that non-gender products (neutral gender products) should be sold by male endorsers. 3.5 Advertisement Believability Believability is an essential effectiveness indicator of every communication (Bertramini, 2006). According to Rajaobelina et al. (2019), Advertising believability refers to the degree to which the consumer sees claims made about the brand in an ad to be truthful and believable. Research 27 University of Ghana http://ugspace.ug.edu.gh evidence suggests that there is a significant relationship between advertisement believability and purchase behavior (Mingioneb et al., 2019). Hoyer and MacInnis (2004) have argued that advertising believability is essential for successful consumer persuasion. For example from their study which examined differences between advertisements with celebrities and non-celebrities endorsers, Menon, Boone and Rogers (2001) reported that purchase intentions seem to be correlated to advertising believability. Data was collected from Southeastern university students in America through a survey. Results of the analysis emphasised that believability and purchase intentions are vital dependent variables for advertisers. From the forgoing discussion, one might not be wrong to say that the bottom line goal of every advertisement is to get the audience member to act in a particular way. Therefore, creative directors make sure ads contain enough plausible motivations to drive audiences’ behavior. This is what influences a person’s desire to purchase a product or buy into an idea. According to theories of audience psychology, behavior is almost always preceded by intention which is also directly influenced by some cognitive processes such as belief, often defined as the mental acceptance of a claim as true. Believability therefore is a crucial determinant of purchase behavior among advertising audiences (Michela et al., 2019; Petrescu et al., 2019). In fact, advertisement believability is not a new phenomenon in advertising literature. As far back as 1963, Maloney specified that advertising believability epitomizes the effect that advertising has on the minds of audiences, in the form of attitudes and intentions towards the product (Michela et al., 2019). Even Petty and Cacioppo (1984) noted that different types of advertising message content might be feeble because consumers interpret the content as unbelievable and implausible 28 University of Ghana http://ugspace.ug.edu.gh in the message context. Accordingly Karnins (1989) argued simply that, if an advertisement is not believed, then its effectiveness is restricted. While consumers might recall an ad or understand its central message, for it to have a positive influence, individuals need to believe the information contained in an ad (Beltramini 2006). When consumers fail to trust what is being advertised, the probability of positive consumer attitudes and behavioral objectives is small (Atkin and Beltramini 2007). However, Soh et al. (2009) contend that when consumers consider an ad as believable, positive effects on their attitudes towards the advertised product or service emerge. Similarly, Michela et al., (2019) found from their study that advertisement believability enhance sales of the advertised product. This conclusion suggests that advertisement believability has a direct connection to sales thereby making the phenomenon crucial to brand communicators and advertisers. Even the Nielsen survey, an international institution responsible for carrying out periodic global survey on ad trust among individuals across 58 counties, finds believability as key to Advertising Effectiveness (Consoli 2018). Findings from the survey indicate that, beyond trust in advertising leading to consumer goodwill toward a company, it also motivates consumers to go out and buy products they see in that company's ads, with the percentages pretty much bearing that out. Scholars have also reported that the believability of advertising information depends on the type of ad appeal, design, structure and style of the ad (Lin et al., 2012; Petrescu et al., 2015). According to these researchers audiences tend to consider an ad more believable if they found it is well- written and well-designed (Lin et al., 2012). From the forgoing, it appears strategic selection of celebrities and products influence audiences’ perception of endorser credibility which in turn 29 University of Ghana http://ugspace.ug.edu.gh predicts perception about the ad in general. One would therefore not be wrong to assert that certain endorser and product permutations are likely to be perceived as well-designed and thus capable of creating trust or believability in audiences. 3.6 Research Hypotheses A research hypothesis is a realistic guess or an educated conjecture that provides a tentative explanation for a phenomenon under investigation (Leedy and Ormrod, 2001). The current study is grounded on four hypotheses whose formulations are informed by mixed viewpoints from the above literature review as well as the underpinning models. The study predicts that there is a relationship among its independent variables (perceived fit between celebrity endorser gender and product gender) and dependent variable (advertisement believability). As succinctly argued in the match up hypothesis model, audiences are likely to favourably evaluate celebrity endorsed ads in which the endorser’s gender aligns with the perceived gender image of the advertised product (Putrevu, 2001). Additionally, Lien (2012) has contended that advertising effect is more positive with a female spokesperson than with a male spokesperson when it comes to endorsement of feminine-gendered products. Even though it has been reported that sometimes endorser/product incongruity creates some sort of a deviation from the norm which excites audiences and consequently, increases the effectiveness of the endorsement (Lee & Thornson, 2008), it is important to note that such pairings are sometimes problematic since audiences become frustrated while they try to figure out links between the endorser and product (D’Astous and Bitz, 1996). Besides, these pairings have been found to be successful among participants with higher product involvement than those with lower product involvement. 30 University of Ghana http://ugspace.ug.edu.gh Moreover, studies (Fink et al., 2004; Erlandson et al., 2013) have documented fit between celebrity endorser gender and product gender as predicting endorsement effectiveness. Also, the source credibility model explains that endorsers who are perceived as trustworthy or experts are to this extent considered as believable (Ohanian, 1990). Accordingly, it can be argued that certain characteristics of sources influence audiences’ perception of their (sources) credibility. Gender has been found to be a variable of these characteristics which influence perception of source credibility (Glenn et al., 2019). However, evidence from literature suggests that the gendered characteristics of the endorsed product mostly determines the suitability or otherwise of the endorser and consequently, the perception of his or her (endorser’s) credibility (Park & Lin, 2020; Lien, 2001). Also, Ohanian (1990) argued that perception of a source’s credibility directly relates to acceptance of his or her (source’s) message. Thus, this study foresees that perceived fit between celebrity endorser gender and perceived gender image of the endorsed product will positively relate to audiences’ believability of the endorsed ad. Specifically, the study assumes that: } H1: Perceived match between a female endorser and a feminine product will enhance believability of the endorsed advertisement. } H0: Perceived match between a female celebrity endorser and a feminine product will not enhance believability of the endorsed advertisement. } H2: Perceived match between a male celebrity endorser and a masculine product will enhance believability of the endorsed advertisement. } H0: Perceived match between a male celebrity endorser and a masculine product will not enhance believability of the endorsed advertisement 31 University of Ghana http://ugspace.ug.edu.gh } H3: Perceived match between a female celebrity endorser and a masculine product will not enhance believability of the endorsed advertisement. } H0: Perceived match between a female celebrity endorser and a masculine product will enhance believability of the endorsed advertisement. } H4: Perceived match between a male celebrity endorser and a feminine product will not enhance believability of the endorsed advertisement. } H0: Perceived match between a male celebrity endorser and a feminine product will enhance believability of the endorsed advertisement. Figure 3.1: The researcher’s Conceptual model describing the relationship between the variables under investigation Celebrity endorser Gender Advertisement Believability Product Gender The model above represents a hypothetical relationship between the independent and dependent variables of the current study. The study expects that gender matching of a celebrity endorser and the endorsed product will influence audiences’ advertisement believability. 32 University of Ghana http://ugspace.ug.edu.gh 3.7 Chapter Summary This chapter contained the review of related literature, the hypotheses and conceptual model of the study. From the literature, it seems celebrity endorsement is esteemed highly than other advertising strategies especially in terms of its ability to promote brands faster. Also, it is evident that the strategy has received some research attention in Ghana although little is known about how a fit between a celebrity’s gender and the endorsed product’s gender affects audience perceptions. The review again revealed that a positive match between an endorser and a product increases the tendency for a favorable evaluation of the advertisement. Based on evidence from the review and assumptions from the models, the researcher developed four hypotheses. The researcher also developed a model to demonstrate a hypothetical relationship among the main variables which the study is investigating. The four hypotheses are: 1. Perceived match between a female endorser and a feminine product will enhance believability of the endorsed advertisement. 2. Perceived match between a male celebrity endorser and a masculine product will enhance believability of the endorsed advertisement. 3. Perceived match between a female celebrity endorser and a masculine product will not enhance believability of the endorsed advertisement. 4. Perceived match between a male celebrity endorser and a feminine product will not enhance believability of the endorsed advertisement. 33 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR METHODOLOGY 4.0 Introduction This section presents procedures which were adopted in doing the study. It discusses the research paradigm, empirical setting and sample, sampling procedure, the data collection processes, the data analysis procedure as well as ethical considerations made. 4.1 Research Paradigm This research adopted the deductive paradigm of enquiry. Its core agenda was to test existing theories and concepts through thorough investigation of hypotheses. The study thus adopted the quantitative research approach which is based on scrutinising relationships among variables where data is collected in numeric form and analysed using statistical tools (Saunders & Lewis, 2012). Experiment was the chosen method for data collection. The main objective of the study was to discover the effect of celebrity endorser gender and product gender on advertisement believability. The experiment method is therefore appropriate as it has been found to be the most useful procedure in testing causal relationships among variables (Ross & Morrison, 2014). 4.2 Research Design As explained earlier, this research adopted the experiment method. Specifically, a 2 (Endorser gender: male vs female) × 3 (product gender: masculine vs. feminine vs neutral) full factorial within-subjects design was chosen. This design is a variation of a true experimental design which allows for the situation where all treatments are shown to all subjects (Rogers & Revesz, 2020; Gravetter & Forzano, 2018; Rogers et al., 2017). Thus, each person serves as his or her own control 34 University of Ghana http://ugspace.ug.edu.gh and is tested or “observed” (Ross & Morrison, 2014). This means that within-subject designs enable researchers to compare responses of a participant across all the various levels of the experimental conditions. The study’s design can be defined from two major angles. First, it is full factorial because it involves exploitation of variables which have more than one factor or face (e.g. Gender= male & female). Also, it is a within-subjects design because it offers the researcher the opportunity to investigate research effects while allowing all subjects to experience all treatments (Gravetter & Forzano, 2018; Abbuhl & Mackey, 2017). In a within-subjects design, participants are exposed to all levels of the independent variables. Within-subject designs differ from between-subject designs in the sense that between-subject designs allow for portions of the treatment conditions to be assigned to different groups of participants. In other words, different participants are tested on the various levels of the independent variable (Ross & Morrison, 2014). However, a within-subjects design helps to examine how each of the subjects behaves under every treatment. In other words, it helps to determine the effects of all the manipulations on all the subjects. It also helps to reduce blunders associated with individual differences which have the tendency to confound the results in a between-subjects design. In a between-subjects design where individuals are randomly assigned to portions of the treatment, there is a possibility that there may be fundamental individual differences between the groups which might influence the results. This error is somehow controlled in a within-subject design where every subject receives every treatment. Even though the risk of sensitisation or what is technically called carryover effect (a situation where the effect of one manipulation may be present when the next manipulation is introduced) 35 University of Ghana http://ugspace.ug.edu.gh exists in the chosen design, there is a formula that has widely been used to at least, balance out the situation. This formula constitutes varying the order in which the treatments are presented among the participants and the researcher adopted this strategy to mitigate the tendency of sensitisation. 4.3 Empirical Setting and Sample The study was conducted among students in only the Legon (main) campus of the University of Ghana. The University of Ghana is the premier and largest university in the country and has multiple campuses across Ghana. It attracts people from all parts of the country giving researchers a good mix of different demographics. As at 2018, student population size of the Legon campus was 37,940. Out of this number, 22,813 are undergraduates (University of Ghana Institutional Research and Planning office, 2018). This figure clearly suggests that the undergraduate students constitute the bulk of students in the Legon campus thus there is the tendency that they will better reflect the dynamics within the students’ population of this particular campus of the University. Undergraduate students in the Legon (main) campus of the University of Ghana were the focus of the study. Students in Legon were considered ideal for this study because, according to Kim et al. (2008), young adults are particularly appropriate to the investigation of celebrity-related phenomena because they are usually a target market for advertisers who use celebrities and easily form emotional attachment to celebrities. 4.4 Sampling Procedure In experimental designs, the sampling method which is frequently used among most mass communication researchers is the nonprobability sampling approach (Erba et al., 2018). Even though Moy and Murphy (2016) contend that data obtained from non-probability samples may not 36 University of Ghana http://ugspace.ug.edu.gh be representative of the population of interest thereby making generalisation of such data to a larger population unreasonable, Courtright (1996) argues that generalisation to a population is not the goal of experiments. Rather, the purpose of experiments is to examine hypothesised causal relationships between independent and dependent variables (pg. 414). Courtright further asserts that: “Researchers will almost certainly continue to investigate certain important relationships in mass communication using experimental methods and convenience samples composed of college-aged students. Whether these experiments are labeled ‘scientific’ or ‘prescientific’ will not change this basic fact nor will such labels affect the quality and theoretic utility of those studies” (pg.420). Similarly, Shapiro (2002) argues that for communication scholars, the relationship between a sample and a population is not the only or the most powerful way of thinking about generalisation but the ability to make specific predictions about when particular causal relationships will hold and when they will not hold. Besides, what is common among researchers in the area of celebrity endorsement, particularly, researchers (Edward & La Ferle, 2009; Klause & Bailey, 2008; Boyed & Shank, 2004) who use experiments is that they adopt a convenience sampling technique. The researcher solicited for participation from the various class WhatsApp pages of University of Ghana undergraduate students on Legon campus. The study therefore conveniently sampled 120 undergraduate UG students. Information was sent to them via their class WhatsApp pages. The link to the research WhatsApp group page was provided as well. Students who showed interest were asked to use the link to join the page created for the research. 37 University of Ghana http://ugspace.ug.edu.gh 4.5 Data Collection Procedure As noted earlier, there was only one group of subjects for this experiment. Subjects were presented with colored print advertisements featuring a male (Mikki Osei Berko, popularly known as Master Richard) and a female (Nana Ama Mcbrown) celebrity endorser appearing with three categories of products (Feminine, masculine and neutral gendered products). The specific products were: sanitary pad for a feminine product, shaving cream for the masculine product and bottled water as the neutral product. Thus, there were six treatments in all (i.e. each endorser appearing with the three products). The advertisements together with the brands were fictitious. This decision was made in order to control for the possibility of a situation where participants might bring their prior impressions of already existing advertisements and brands to the experiment which could confound the results. For an ad of a particular product category, for example, feminine product (sanitary pad), the ad message, shape and colour of the product were constant. The only thing that changed was the gender of the endorser. This was done to manipulate endorser gender as the only possible explanation for audience believability of the endorsed ad. After exposure to each treatment, subjects’ responses were observed using a questionnaire. Treatments for the experiment together with their corresponding questionnaires were arranged unsystematically for the subjects. This decision was taken as a way of eliminating any sequence effect which could alter the results. The researcher specifically adopted this strategy to mitigate sensitization (carryover effect) which is common in within-subjects experiments (Rogers & Revesz, 2020; Gravetter & Forzano, 2018). The treatments were in the following forms: 38 University of Ghana http://ugspace.ug.edu.gh 1. A picture of a female celebrity appearing with shaving cream 2. A picture of a male celebrity appearing with bottled water 3. A picture of a female celebrity appearing with sanitary pad 4. A picture of a female celebrity appearing with bottled water 5. A picture of a male celebrity appearing with sanitary pad 6. A picture of a male celebrity appearing with shaving cream 4.5.1 Pre-Test A pre-test was conducted on hundred (fifty males and fifty females) undergraduate students from the department of information studies. Participants were put on two separate WhatsApp platforms; one for the males and one for the females. A day was agreed for the exercise on the males’ page first. At the set time, the link to the first part of the study was put on the page for the participants to fill. About 25 participants filled it. The remaining participants had challenges accessing the link. This became a major problem and made it impossible to administer the second part of the study. The researcher asked for the study to be postponed to another day so the link could be worked on. At the agreed day, the researcher put the link back on the page and yet about half of the participants complained of their inability to access the link. Ten participants who were able to access the link were asked to meet on Zoom where the second part of the study was to be conducted. These participants were shown the treatments and were asked the subsequent questions. At the end of the exercise however, only three participants held on to present their answers. This informed the researcher to make some adjustments to the study with the females. So, for the females, both parts 39 University of Ghana http://ugspace.ug.edu.gh of the study were conducted on WhatsApp. At the agreed time, the link to the first part was put on the page. Twenty out of the 50 participants filled it. After the agreed time for the first part was up, the second part of the instrument was administered on the same page. Ten out of the 20 participants who were partakers of the first part of the experimental process answered it. Participation was thus very low. The challenges in the pretest greatly informed decisions in the main study. For instance, both the two parts of the study were changed into a Google form. A key reason for the decision to use Google form was the need to avoid the situation where the treatments got flooded in other messages on the WhatsApp page and participants repeatedly asked for a treatment which had already been sent on the page to be resent. Also, the researcher decided to change the experimental setting from Zoom to WatsApp because participants, in the pre-test, complained of network problems. Moreover, the study was not organized as a one-off exercise where all the participants were assembled as initially planned. Instead, participants were grouped in small batches in accordance with their own chosen convenient times. In all, there were fifteen different WatsApp groups created. The groups were made up of at least two participants and at most about forty participants. 4.5.2 Main Study The experiment was conducted online in two parts. There were two separate links leading to two different Google forms. In the first part, participants were asked to provide brief information about themselves, their perceptions about celebrity endorsement, male and female celebrity endorsers among others on the Google form. After completion of the first section, participants were asked to fill another Google form which contained the six experimental treatments. Each treatment was 40 University of Ghana http://ugspace.ug.edu.gh followed by a set of questions. The order of presentation of the treatments was jumbled as explained earlier to reduce the effect of sensitization. Treatment A contained the picture of Mikki Osei Berko (Master Richard) and a sanitary pad packaged in a pink wrapper with the inscription “sleek lady sanitary pad” on it. Other inscriptions placed beneath the endorser and the product read: “I love to make things easy for my woman at all times, so I make sure I get her sleek lady sanitary pad…and things never go wrong”, beneath the endorser and the product. Participants were then asked to indicate their perception of endorser and product fit in the ad. They were also made to demonstrate their believability or otherwise of the advertisement on the basis of the endorser’s gender. In treatment B, Nana Ama Mcbrown is seen with the same sanitary pad product and similar inscription. Participants were again asked to indicate perception of fit between the endorser and the product as well as their believability of the ad based on the endorser’s gender. Treatment C contained the picture of Mikki Osei Berko and a small black and blue bottle on which were the writings “Shave cream…Kings” There was also boldly written beneath the endorser and the product, “I love to always look like the King that I am so I use Kings shaving cream. It works magic! Get yours now.” After this ad, participants were asked again to determine fit or unfit between the endorser and the product as well as whether they believe the ad because of the endorser’s gender. Treatment D also had the pictures of Nana Ama Mcbrown and the same shaving cream and similar inscriptions in treatment C. Similar set of questions which followed treatment C followed treatment D as well. In treatment E, there was the picture of Mikki Osei Berko and a bottled water brand captioned “Aqua bliss”. Beneath the endorser and the logo were the writings, “Every gulp is a bliss.” In the subsequent questions, participants’ perception of endorser and product fit as well as their believability of the ad on the 41 University of Ghana http://ugspace.ug.edu.gh basis of the endorser’s gender were asked. Treatment F had the picture of Nana Ama Mcbrown appearing with the same product and message in treatment E. Again, similar questions which were asked after treatment E followed treatment F. After completion of all the experiments, participants were thanked for their time and efforts. 4.6 Construct Measures According to Rajaobelina et al. (2019), Advertising believability refers to the “extent to which the consumer perceives claims made about the brand in an ad to be truthful and believable. A commonly used measure of ad believability is Obermiller and Spangenberg’s (1998) ad skepticism (SKEP) scale. It measures general advertising credibility (believability) from the opposite direction using nine items to tap whether participants believe an ad or not. The specific items can be found in Appendix one. Ohanian’s (1990) source credibility scale was adopted to measure perception about male and female celebrity endorsers. According to Ohanian (1990), source credibility implies a communicator’s perceived positive characteristics which affect the receiver’s acceptance of a message. The scale consists of ten items which are: dependable, honest, reliable, sincere, trustworthy, expert, knowledgeable, experienced, qualified and skilled. To assess perceived fit between celebrity endorser gender and product gender, Till and Busler’s (2008) one item scale (which asks respondents to indicate fit between an endorser and the endorsed product) for measuring fit between celebrity endorser and product was adapted. This scale could also be found in Appendix one. Also, demographic data were collected with items including age, gender and level of education. Moreover, participants’ celebrity involvement levels were measured with items asking them to indicate their interests in celebrities. Likewise, participants’ ad literacy 42 University of Ghana http://ugspace.ug.edu.gh information were gathered using the CALS scale by Rozendaal et al. (2016). The scale contains six items. These items are: I think adverts are meant to make me think positively about the advertised product. 2. I think adverts always tell the truth 3. I think adverts are targeted at a specific group of people at all times 4. Celebrities are often used in adverts to encourage purchase 5. I think adverts tell half truth 6. I think adverts are paid for by someone to influence people In measuring participants’ celebrity involvement, Yen and Teng’s (2015) celebrity involvement scale was borrowed. As well, a six item measure adapted from Pullay and Mittal’s (1993) opinion towards ad scale was used to collect participant’s attitude towards celebrity endorsed advertisements. The scale includes items like: “I learn about fashions and about what to wear to impress others from celebrity endorsed ads”; “Celebrity endorsed ads are a valuable source of information about local sales”; “celebrity endorsed ads tell me what people with lifestyle similar to mine are buying and using”; “Celebrity endorsed ads tell me which brands have the features I am looking for” “Celebrity endorsed ads present a true picture of the product advertised” and “Celebrity endorsed ads help me keep up to date about product/services available in the marketplace.” As stated earlier, all these scales could be found in Appendix one. 4.7 Data Analysis Procedure Both descriptive and inferential statistics were run to measure the relationship among the variables. The researcher processed and managed the gathered data by first cleaning them to identify missing 43 University of Ghana http://ugspace.ug.edu.gh cases and values as well as remove all forms of errors which might have gone unnoticed. Descriptive statistics was employed to analyze the background information of participants. Data was also analysed inferentially. For instance, in testing all the four hypotheses, linear regression was used to assess effects of the independent variables on the dependent variables. 4.8 Ethical Consideration All required procedures for attaining ethical clearance from the Ethics Committee for the Humanities (ECH) in the University of Ghana were met. The researcher was thus able to attain ethical clearance for the work. The researcher also satisfied important ethical concerns. For instance, assurance of confidentiality of responses and anonymity of participants were included in the study. Also, the researcher explained participants’ right to withdraw from the study at any time they feel they no longer want to participate. Also, numbers and email addresses of the researcher were made available to participants for answers to any questions they might have about the research. 4.9 Chapter Summary This chapter contained the methodological processes the researcher adopted in doing the study. Some of these procedures include: Research design, empirical setting and sampling procedures, data collection and data analysis procedures among others. Also, the various scales used to measure the different constructs of the study were outlined. Moreover, the researcher discussed the ethical decisions taken. In the chapter, the researcher specifically explained the study’s design, which is a 2 (endorser gender: male and female) ×3 (Product gender: feminine, masculine and neutral) full factorial experimental design. This design enabled the researcher to show all the 44 University of Ghana http://ugspace.ug.edu.gh treatments to all the subjects. The treatments were six colored print advertisements in which two celebrities (Mikki Osei Berko and Nana Ama McBrown) appeared with the three products categories, being: Sanitary pad, shave cream and bottle water. To prevent the risk of sensitisation, the researcher jumbled the treatments up to distort any clear order which could influence respondent’s answers. A pre-test was done prior to the main study. Lessons learnt from the pre- test greatly shaped the organisation of the main study. In all, 108 undergraduate students from the University of Ghana participated in the experiment. The experiment was conducted online. 45 University of Ghana http://ugspace.ug.edu.gh CHAPTER FIVE FINDINGS 5.0 Introduction The current study examined implications of gender in celebrity endorsed advertisements. The aim was to contribute to the body of knowledge on celebrity endorsements in Ghana. This chapter presents analysis and findings of the study. 5.1 Sample Description The study sampled one hundred and fourteen (114) undergraduate students from the University of Ghana, however, 108 (fifty-four males and fifty-four females) responses were used after data cleaning. The level of education of participants ranged from level 100 to 400. Out of the 108 participants, 44.4% were in level 100, 7.4% were in level 200, 17.6% were in level 300 whereas 30.6% of them happened to be in their final year. Participants fell within three main age categories: 18-23, 24-29, 30-36. Out of the total number of participants, 89.8% of them fell within the age range of 18-23 whereas 9.3% were within the age category two which is 24-29. The remaining 0.9% fell within the age range of 30-36. 46 University of Ghana http://ugspace.ug.edu.gh Figure 5.1: Age 120 100 80 60 40 20 0 18-23 24-29 30-36 Figure 5.2: Level of Education 50.0 44.4 45.0 40.0 35.0 30.6 30.0 25.0 20.0 18.5 15.0 10.0 6.5 5.0 0.0 100 200 300 400 47 University of Ghana http://ugspace.ug.edu.gh Figure 5.3: Participants’ Gender Male 50% 50% Female 5.1.1 Product Gender To be able to measure the percieved gender image of the product cateogories used in the experimental treatments, participants were asked to indicate the gender of three different kinds of products namely, sanitary pad, shave cream and bottled water. Majority of the participants rated sanitary pad as a feminine product, shave cream as a masculine product and botlled water as a neutral product. 48 University of Ghana http://ugspace.ug.edu.gh Table 5.1: Perceived product gender PRODUCT PERCIEVED GENDER Feminine Masculine Neutral Shave cream 3% 89% 8% Sanitary pad 98% 2% 0% Bottled water 5.9% 1% 93% 5.2 Analysis of Construct Measures The study used a number of multi-item measured constructs. Thus in order to reduce the large number of items into a manageable set, an initial exploratory factor analysis was done. After the initial factor analysis, problematic items including those which did not load above the threshold mark (0.5) were excluded from further analysis. A principal component analysis employing rotation (Promax 4) and extraction (Unweighted least squares) was run again, all the items which loaded above the threshold mark of 0.5 were maintained. Factors in the initial solution were Ad Believability for Treatment 1 (T1:ADB = 5 items), Ad Believability for Treatment 2 (T2:ADB = 5 items), Ad Believability for Treatment 3 (T3:ADB = 5 items), Ad Believability for Treatment 4 (T4:ADB = 5 items), Ad Believability for Treatment 5 (T5:ADB = 5 items), Ad Believability for Treatment 6 (T6:ADB = 5 items), Attitude towards Celebrity Endorsements (CEATT= 6 items), Source Credibility (SOURCECRED = 10 items), Advertising Literacy (ADLIT= 5 items) and Celebrity Involvement (CELEBINVO = 6 items). The initial model estimated met the assumptions of the EFA. The KMO test statistic returned was 0.625 49 University of Ghana http://ugspace.ug.edu.gh indicating that the sample was adequate for estimating the factor model. Also, the data was found as fit for reduction since the Bartlett’s test of sphericity was significant. The EFA returned an eleven factor model (instead of the expected ten factor model). Generally, these eleven factors extracted explained 75% of the cumulative variance in the model. Besides the ten factors expected to be extracted, one item each of Advertising Literacy (ADLIT5) and Attitude towards celebrity endorsement (CEATT 5) created a surplus factor. This suggests that the items in question do not share any properties with their expected underlying factor. Moreover, one item of Attitude towards Celebrity Endorsement (CEATT4) cross-loaded significantly on Celebrity Involvement. In addition, one item of Ad Believability for Treatment 2 (T2:ADB5), Source Credibility (SOURCECRED 5) as well as Attitude towards Celebrity Endorsement (CEATT6) returned a loading below the set criteria of .50. Thus, these items were excluded from further analysis. More importantly, Ad Believability for Treatments 1 and 3 cross-loaded significantly on each other. The situation persisted even after different extraction and rotation methods were explored. The possible option was exclusion, yet, each variable is core to the study’s objective. Besides, they all returned very high communality indexes. This made it even tougher to delete them given the fact that all the items loaded significantly. Dang (2020) citing Hair, Black and Babin (2009) advised that in such a situation, a researcher can ignore those problematic variables. Standing on these facts, the researcher decided to ignore the variables. 50 University of Ghana http://ugspace.ug.edu.gh Table 5.2: Initial Factor Solution ITEMS FACTORS 1 2 3 4 5 6 7 8 9 10 T1:ADB1 .930 .809 T1:ADB2 .908 .786 T1:ADB3 .893 .775 T1:ADB4 .880 .700 T1:ADB5 .753 .677 T2:ADB1 .873 T2:ADB2 .684 T2:ADB3 .634 T2:ADB4 .534 T2:ADB5 T3:ADB1 .912 T3:ADB2 .885 T3:ADB3 .807 T3:ADB4 .769 T3:ADB5 .681 T4:ADB1 .857 T4:ADB2 .840 T4:ADB3 .828 T4:ADB4 .782 51 University of Ghana http://ugspace.ug.edu.gh T4:ADB5 .772 T6:ADB1 .946 T6:ADB2 .905 T6:ADB3 .884 T6:ADB4 .831 T6:ADB5 .795 T6:ADB5 SOURCECRED1 .969 SOURCECRED2 .906 SOURCECRED3 .902 SOURCECRED4 .881 SOURCECRED5 SOURCECRED6 .856 SOURCECRED7 .851 SOURCECRED8 .819 SOURCECRED9 .798 SOURCECRED10 .780 ADLIT1 .735 ADLIT2 .647 ADLIT3 .640 ADLIT4 .527 ADLIT5 .719 52 University of Ghana http://ugspace.ug.edu.gh CELEBINVO1 .844 CELEBINVO2 .822 CELEBINVO3 .771 CELEBINVO4 .732 CELEBINVO5 .712 CELEBINVO6 .650 CELEBINVO7 .605 CELEBINVO8 .547 CEATT1 .798 CEATT2 .690 CEATT3 .532 CEATT4 .529 CEATT5 .581 CEATT6 Subsequently, the researcher estimated a second EFA solution comprising 53 items (T1: ADB= 5 items; T2: ADB= 4 items; T3: ADB= 5 items; T4: ADB= 5 items; T5: ADB= 5 items; T6: ADB= 5 items; CELEBINVO = 8 items; SOURCECRED = 9 items; CEATT = 3 items and ADLIT=4 items). The same methods of extraction and rotation were adopted in estimating this second EFA solution. Interestingly, Ad Believability for Treatments 1 and 3 loaded significantly on separate variables. All items which loaded below the threshold value (0.5) were excluded from further analysis. 53 University of Ghana http://ugspace.ug.edu.gh Table 5.3: Result of Final Factor Solution ITEMS FACTOR LOADING 1 2 3 4 5 6 7 8 9 10 T3:ADB1 .893 T3:ADB2 .825 T3:ADB3 .778 T3:ADB4 .769 T3:ADB5 .555 SOURCECRED1 .942 SOURCECRED2 .914 SOURCECRED3 .892 SOURCECRED4 .846 SOURCECRED6 .805 SOURCECRED7 .942 SOURCECRED8 .914 SOURCECRED9 .892 SOURCECRED10 .846 T4:ADB1 .842 T4:ADB2 .829 T4:ADB3 .825 T4:ADB4 .818 T4:ADB5 .773 CELEBINVO1 .754 CELEBINVO2 .748 CELEBINVO3 .727 CELEBINVO4 .686 CELEBINVO5 .635 CELEBINVO6 .627 T5:ADB1 -.938 T5:ADB2 -.902 T5:ADB3 -.890 T5:ADB4 -.826 T5:ADB5 -.821 ADLIT1 .718 ADLIT2 .646 ADLIT3 .639 54 University of Ghana http://ugspace.ug.edu.gh ADLIT4 .520 T1:ADB1 -.813 TI:ADB2 -.768 T1:ADB3 -.751 TI:ADB4 -.687 T1:ADB5 -.676 CEATT1 .812 CEATT2 .668 CEATT3 . 635 T6:ADB1 .742 T6:ADB2 .655 T6:ADB3 .617 T6:ADB4 .597 T2:ADB1 .-653 T2:ADB2 .-582 T2:ADB3 .-573 5.2.1 Validity and Reliability Tests All items were tested for convergent validity and were seen to be valid since they all loaded significantly above 0.5. Cronbach’s Alpha was used to test for reliability. The test showed that item sets for all constructs are reliable as they exceeded the recommended threshold of 0.5 (Hinton, 2014). Ad Believability for Treatment 5 recorded the highest Cronbach’s Alpha of .972 whereas Celebrity Involvement had the lowest Alpha of .664. 55 University of Ghana http://ugspace.ug.edu.gh Table 5.4: Result of Reliability Tests ITEMS CRONBACH’S ALPHA AD BELIEVABILITY FOR TREATMENT .930 3 T3:ADB1 T3:ADB2 T3:ADB3 T3:ADB4 T3:ADB5 SOURCE CREDIBILITY .961 SOURCECRED1 SOURCECRED2 SOURCECRED3 SOURCECRED4 SOURCECRED5 SOURCECRED6 SOURCECRED7 SOURCECRED8 SOURCECRED9 CELEBRITY INVOLVEMENT .664 CELEBINVO1 CELEBINVO2 CELEBINVO4 56 University of Ghana http://ugspace.ug.edu.gh CELEBINVO5 CELEBINVO6 CELEBINVO8 AD BELIEVABILITY FOR TREATMENT .972 5 T5: ADB1 T5: ADB2 T5: ADB3 T5: ADB4 T5: ADB5 ADVERTISING LITERACY .675 ADLIT1 ADLIT2 ADLIT3 ADLIT4 AD BELIEVABILITY FOR TREATMENT .912 1 T1:ADB1 T1:ADB2 T1:ADB3 T1:ADB4 T1:ADB5 57 University of Ghana http://ugspace.ug.edu.gh ATTITUDE TOWARDS CELEBRITY .836 ENDORSEMENT CEATT1 CEATT2 CEATT3 AD BELIEVABILITY FOR TREATMENT .960 6 T6:ADB1 T6:ADB2 T6:ADB3 T6:ADB4 AD BELEIVABILITY FOR TREATMENT .823 2 T2:ADB1 T2:ADB2 T2:ADB3 AD BELEIVABILITY FOR TREATMENT .953 4 T4:ADB1 T4:ADB2 T4:ADB3 T4:ADB4 58 University of Ghana http://ugspace.ug.edu.gh T4:ADB5 5.2.2 Discriminant Validity Test The researcher also conducted a discriminant validity test to make sure measures that should not be correlated are actually not related. Discriminant validity ensures that a construct measure is empirically unique and represents phenomena of interest that other measures in a model do not capture (Hair et al., 2010). Discriminant validity requires that “a test not correlate too highly with measures from which it is supposed to differ” (Campbell, 1960, p. 548 as cited by Henseler et al., 2015). The result shows sufficient discriminant validity among constructs. The researcher thus moved on to use the composites of the constructs for further analysis. Table 5.5: Result of Discriminant Validity Test Correlations T3:ADB SC T4:ADB CE T5:ADB ADLIT T1:ADB CEATT T6:ADB T2:ADB T3:ADB 1.000 .052 -.111 -.002 .514** .106 .036 .149 .747** .494** SC .052 1.000 .146 .331** .151 .093 .111 .483** .071 .066 T4:ADB -.111 .146 1.000 -.019 .146 -.052 .602** .152 -.140 .278** CE -.002 .331** -.019 1.000 .102 .243* -.064 .577** .124 -.060 T5:ADB .514** .151 .146 .102 1.000 .128 .151 .271** .534** .507** .106 .093 -.052 .243* .128 1.000 .023 .198* .109 .086 59 University of Ghana http://ugspace.ug.edu.gh ADLIT T1:ADB .036 .111 .602** -.064 .151 .023 1.000 .104 .020 .359** CEATT .149 .483** .152 .577** .271** .198* .104 1.000 .189 .125 T6:ADB .747** .071 -.140 .124 .534** .109 .020 .189 1.000 .464** T2:ADB .494** .066 .278** -.060 .507** .086 .359** .125 .464** 1.000 Table 5.6: Participant’s Response to Relevant Constructs Construct N Mean ATTITUDE TOWARDS CELEBRITY ENDORSEMENT 108 3.24 SOURCE CREDIBILITY 108 2.72 ADVERTISING LITERACY 108 3.99 CELEBRITY INVOLVEMENT 108 3.06 ATTITUDE TOWARDS NANA AMA MCBROWN 95 3.62 ATTITUDE TOWARDS MASTER RICHARD 96 2.68 From Table 6, it can be seen that the participants had moderate levels of ad literacy (3.99) celebrity involvement (3.06) attitude towards celebrity endorsement (3.24) and attitude towards Nana Ama Mcbrown (3.621). They also tended to have a relatively low perception (2.72) about male and female endorsers’ credibility in relation to endorsement of certain kinds of products. Moreover, participants had a low level of attitude towards Master Richard (2.68) 60 University of Ghana http://ugspace.ug.edu.gh 5.3 Nature of Relationship between Endorser Gender and Ad Believability The mean scores of participants regarding their belief in the six advertisements exposed to them were computed. Results indicate that participants’ believability of the endorsements differ across a male and a female celebrity endorser. Specifically, endorsement of feminine product by a female celebrity endorser scored a higher believability (3.89) than that of a masculine product endorsed by a female celebrity (2.74). Participants indicated an even lower level of belief (2.46) in the ad of a feminine product endorsed by a male celebrity. Also, participants showed a lower believability towards the ad in which the female celebrity endorsed the masculine product. Believability level was higher (3.87) for endorsement of masculine product by a male celebrity endorser than that by the female celebrity endorser (2.74). Where a neutral product was concerned, participants tended to hover around the same levels of belief as one moved from one endorser gender to the other (Male = 3.26; Female = 3.54). Comparatively, however, endorsement of bottled water (neutral product) by the female celebrity endorser received a slightly higher believability score than that by the male celebrity endorser. From the foregoing, therefore, the study finds that gender of a celebrity endorser does appear to have implications for ad believability depending on the nature and gender of the product advertised. 61 University of Ghana http://ugspace.ug.edu.gh Table 5.7: Relationship between Endorser Gender and Ad Believability Treatments Believability (Mean) Treatment 1 (Nana Ama McBrown and Shave 2.74 Cream) Treatment 2 (Master Richard and Bottled 3.26 water) Treatment 3 (Nana Ama McBrown and 3.89 Sanitary pad) Treatment 4 (Master Richard and Sanitary pad) 2.46 Treatment 5 (Nana Ama McBrown and Bottled 3.54 water) Treatment 6 (Master Richard and Shave 3.87 Cream) 5.4 Hypothesis Testing Linear regression was run to test each of the four hypotheses of the study. Linear regression was chosen because it helps to measure causality among variables. In other words, it tells whether a change in a dependent variable is explained by the independent variable. The threshold mark for a linear regression test score is P=0.5 (which is equivalent to 50%). So, when a test result is 0.5 or lesser, then it means there is causality among the tested variables. The first hypothesis (H1) argued that the perception of a match between a female celebrity endorser and a feminine product will increase believability. The model summary indicated that the predictors explained 15% of the changes in the dependent variable. 62 University of Ghana http://ugspace.ug.edu.gh Table 5.8: Model Summary for hypothesis one Adjusted R Std. Error of Model R R Square Square the Estimate 1 .391a .153 .126 .86049 a. Predictors: (Constant), Perceived endorser-product for Nana and Sanitary pad, Perceived endorser-product for Richard and Sanitary pad The Anova table shows that the model significantly predicted the dependent variable, Believability for Treatment 3, P=0.05 Table 5.9: ANOVA table for hypothesis one Sum of Model Squares Df Mean Square F Sig. 1 Regression 8.664 2 4.332 5.850 .005b Residual 48.129 65 .740 Total 56.792 67 a. Dependent Variable: ADBNanasleek b. Predictors: (Constant), Perceived endorser-product for Nana and Sanitary pad, Perceived endorser-product for Richard and Sanitary pad From the Coefficient table, it can be seen that a unit change in perceived endorser-product fit between Nana Ama McBrown and Sanitary pad was accompanied by a 12% change in believability levels. The findings led to the acceptance of the hypothesis. 63 University of Ghana http://ugspace.ug.edu.gh Table 5.10: TABLE 10: Coefficients table for hypothesis one Unstandardized Standardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.136 .543 7.618 .000 EPM for Nana and .120 .123 .111 .975 .333 Sanitary pad EPM for Richard and Sanitary pad -.285 .088 -.372 -3.253 .002 a. Dependent Variable: ADBNanasleek The study hypothesized in H2 that perceived match between a male celebrity endorser and a masculine product will enhance believability of the endorsed advertisement. The model table shows that a unit increase in the independent variable (perceived fit between Master Richard and Shave Cream) reflected a 9% change in the independent variable (Ad Believability). Table 5.11: Model Summary for hypothesis two Adjusted R Std. Error of Model R R Square Square the Estimate 1 .307a .094 .067 .96707 a. Predictors: (Constant), Perceived endorser-product for Master Richard and Shave Cream, Perceived endorser-product for Nana Ama McBrown and Shave Cream. The Anova table shows that the model significantly predicted the dependent variable, Believability for Treatment 6, (P>.038). 64 University of Ghana http://ugspace.ug.edu.gh Table 5.12: ANOVA table for hypothesis two Sum of Model Squares Df Mean Square F Sig. 1 Regression 6.440 2 3.220 3.443 .038b Residual 61.725 66 .935 Total 68.165 68 a. Dependent Variable: ADBRichardShaveCream b. Predictors: (Constant), Perceived endorser-product for Master Richard and Shave Cream, Perceived endorser-product for Nana Ama McBrown and Shave Cream The coefficient table shows that a unit change in perceived fit between Master Richard and Shave Cream resulted in 32% change in believability. The direction of effect was as well positive (b=.321, p>.018) Findings therefore supported the hypothesis. Table 5.13: Coefficients table for hypothesis two Unstandardized Standardized Coefficients Coefficients Model B Std. Error Beta T Sig. 1 (Constant) 2.871 .605 4.745 .000 EPM for Master Richard .321 .133 .283 2.418 .018 and Shave Cream EPM for Nana Ama McBrown and Shave -.095 .097 -.114 -.976 .333 Cream a. Dependent Variable: ADBRichardShaveCream In the third hypothesis, the study assumed that perceived match between a female endorser and a masculine product will not enhance believability of the endorsed advertisement. The model summary indicates a 33% variance in Ad believability for treatment 1 attributable to perceived match between Nana Ama McBrown and Shave Cream. 65 University of Ghana http://ugspace.ug.edu.gh Table 5.14: Model Summary for hypothesis three Adjusted R Std. Error of Model R R Square Square the Estimate 1 .577a .333 .312 .86220 a. Predictors: (Constant), Perceived endorser-product for Nana and Shave Cream, Perceived endorser-product for Master Richard and Shave Cream The Anova table shows that the model significantly predicted the dependent variable, Believability for Treatment 1. The P-Value was (p>.000). Table 5.15: ANOVA table for hypothesis three Sum of Model Squares Df Mean Square F Sig. 1 Regression 21.828 2 10.914 14.481 .000b Residual 52.003 69 .754 Total 73.831 71 a. Dependent Variable: ADBNanaShaveCream b. Predictors: (Constant), Perceived endorser-product for Nana and Shave Cream, Perceived endorser-product for Nana and Sanitary pad From the Coefficients Table, there is a positive and statistically significant (P>.000) relationship between the independent variable (Perceived endorser-product fit for Nana and shave cream) and the dependent variable (Ad believability for treatment 1) This is such that a unit change in perceived fit for Nana Ama Mcbrown and Shave Cream led to 45% increase in believability. The study expected an insignificant relationship between the variables. In other words, a p-Value greater than 0.5 so as to confirm the assumption that perceived fit between Nana Ama Mcbrown and Shave Cream will not enhance ad believability. On the contrary, the result failed to confirm the claim because the p-Value was significant (P>.000). Therefore, the hypothesis was rejected. 66 University of Ghana http://ugspace.ug.edu.gh Table 5.16: Coefficients table for hypothesis three Unstandardized Standardized Coefficients Coefficients Model B Std. Error Beta T Sig. 1 (Constant) 1.888 .508 3.715 .000 EPM for Nana and .452 .084 .552 5.376 .000 Shave Cream EPM for Nana and Sanitary pad -.083 .120 -.071 -.694 .490 a. Dependent Variable: ADBNanaShaveCream The fourth hypothesis claimed that perceived match between a male celebrity endorser and a feminine product will not enhance believability of the endorsed advertisement. The model summary reveals that the predictors showed 43% of the variance in ad believability for treatment 4 which is attributable to perceived match between Master Richard and Sanitary pad. Table 5.17: Model Summary for hypothesis four Adjusted R Std. Error of Model R R Square Square the Estimate 1 .660a .436 .419 .82632 a. Predictors: (Constant), Perceived endorser-product for Master Richard and Sanitary Pad, Perceived endorser- product for Nana Ama Mcbrown and Sanitary pad. The Anova table shows that the model significantly predicted the dependent variable, Believability for Treatment 4. The P-Value was (p>.000). 67 University of Ghana http://ugspace.ug.edu.gh Table 5.18: ANOVA table for hypothesis four Sum of Model Squares Df Mean Square F Sig. 1 Regression 34.809 2 17.405 25.490 .000b Residual 45.066 66 .683 Total 79.875 68 a. Dependent Variable: ADBRichardSleek b. Predictors: (Constant), Perceived endorser-product match for Master Richard and Sanitary Pad, Perceived endorser-product for Nana Ama Mcbrown and Sanitary pad The coefficient table also showed that a unit change in perceived fit between Master Richard and Sanitary pad led to 56% increase in believability. Here also, the study anticipated an insignificant p-Value (greater than 0.5) to support the researcher’s claim that perceived match between Master Richard and sanitary pad will not increase believability. The result however revealed a significant p-Value (p>.000) indicating a strong causal relationship between perceived fit of Master Richard and sanitary pad and advertisement believability. Subsequently, H4 was rejected by the data. Table 5.19: Coefficients table for hypothesis four Unstandardized Standardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 2.156 .520 4.144 .000 EPM for Master Richard and Sanitary .562 .084 .618 6.673 .000 Pad EPM for Nana Ama Mcbrown and Sanitary -.248 .116 -.198 -2.138 .036 pad a. Dependent Variable: ADBRichardSleek 68 University of Ghana http://ugspace.ug.edu.gh Table 5.20: Summary of the Hypotheses Test Results HYPOTHESES TEST P-VALUE RESULT H1: Perceived match between a female Linear .005 Supported endorser and a feminine product will enhance Regression believability of the endorsed advertisement H2: Perceived match between a male Linear .038 Supported endorser and a masculine product will Regression enhance believability of the endorsed advertisement H3: Perceived match between a female Linear .000 Not endorser and a masculine product will not Regression Supported enhance believability of the endorsed advertisement H4: Perceived match between a male Linear .000 Not endorser and a feminine product will not Regression Supported enhance believability of the endorsed advertisement 69 University of Ghana http://ugspace.ug.edu.gh Figure 5.4: Empirical Model Celebrity endorser Gender Advertisement Believability Product G e n d e r Audience’s definition of celebrity gender – product gender fit The above empirical model shows that the relationship between celebrity endorser gender –product gender match and advertisement believability is moderated by audience members’ definition of fit. This means that it does not matter what kind of pairing (of celebrities and products) audiences may be presented with, if they perceive a fit, they will favorably rate the advertisement. 70 University of Ghana http://ugspace.ug.edu.gh 5.5 Chapter Summary This chapter presented analysis and findings of the study. The researcher performed two factor analyses to be able to form composites of variables which were measured with multi items. The researcher also conducted convergent validity and reliability tests on the variables before they were used for further analysis. All the tests proved that the variables were sufficient to be used for further analyses. Both descriptive and inferential analyses were done. The findings indicated that celebrity endorser gender affects advertisement believability. Also, audience’s definition of fit between endorser gender and product gender positively relates with advertisement believability. 71 University of Ghana http://ugspace.ug.edu.gh CHAPTER SIX DISCUSSION OF FINDINGS 6.0 Introduction The discussion is done in conjunction with the theoretical concepts (Source credibility model and the Match-up hypothesis model) of the study and relevant literature. The study set for itself, two major objectives which were to: § Examine the nature of relationship between celebrity endorser gender and advertisement believability as to whether: § endorsement of feminine product by female celebrity endorsers enhances ad believability § endorsement of a masculine product by male celebrity endorsers enhances ad believability § endorsement of a feminine product by celebrity male endorsers enhances ad believability § endorsement of a masculine product by celebrity female celebrity endorsers enhances ad believability § endorsement of a neutral product by female celebrity endorsers enhances ad believability § endorsement of a neutral product by a male celebrity endorser enhances ad believability § Discover whether perceived match between celebrity endorser gender and product gender influences audiences’ believability of the advertisement. 72 University of Ghana http://ugspace.ug.edu.gh 6.1 Main Findings of the Study 1. Gender of celebrity endorser has consequences for ad believability depending on the nature and gender of the product advertised 2. Perceived match between celebrity endorser gender and product gender enhances believability. Specifically: a. Perceived match between a female celebrity endorser and a feminine product enhances believability of the endorsed advertisement b. Perceived match between a male celebrity endorser and a masculine product enhances believability of the endorsed advertisement c. Perceived match between a female celebrity endorser and a masculine product enhances believability of the endorsed advertisement d. Perceived match between a male celebrity endorser and a feminine product enhances believability of the endorsed advertisement Discussion of the above findings is done in relation to the objectives they fall under. It is interesting to note that some of the researcher’s assumptions grounded on the theories and literature were challenged. 6.2 The Relationship between Endorser Gender and Advertisement Believability A major part of the controversy surrounding the place of gender in celebrity endorsement advertising is whether or not gender affects evaluation of the endorsement process in any way (Erdogan, 2001). Two schools of thought have emerged on this issue. Whereas scholars such as Mishra et al. (2001) and Klaus and Bailey (2008) believe gender plays a major role in assessment 73 University of Ghana http://ugspace.ug.edu.gh of celebrity endorsed advertisements, others, including Phan and De Run (2007), Ohanian (1991), Pierce and MacBride (1991) argue that audiences are insensitive to endorser gender when it comes to their assessment of celebrity endorsements. Aspects of the controversy have been premised on the source credibility model which explains that acceptance or rejection of a communication is largely dependent on the unique characteristics of the communicator. Standing on the assumptions of the source credibility model, the study set for itself as the first objective, to gain a Ghanaian perspective regarding whether or not gender plays a role in celebrity endorsement evaluation. The objective was to examine the nature of relationship between celebrity endorser gender and advertisement believability. In reference to table 7 in the previous chapter, the results showed that celebrity endorser gender does affect ad believability. This result confirms Dusenburg et al. (2016) and Klaus and Bailey’s (2008) arguments that gender matters in celebrity endorsements. The result also supports assumptions of the source credibility model that the characteristics of a communicator matters a lot to reception and acceptance of persuasive messages (Kang et al., 2019). As the findings showed, participants indicated higher believability levels when the female celebrity endorsed the feminine product and the male celebrity endorsed the masculine product. From a source credibility model point of view, it appears that female celebrities endorsing feminine products are viewed as credible. This could be explained in line with the match-up hypothesis model that the product category (feminine products) is something they (female celebrities) may be expected to know about and be able to share reliable testimony about. The reverse is true for the male celebrity endorser. This also reflects the concept of trustworthiness and expertise captured in the source credibility model. It looks like the participants trusted the female celebrity endorser as having the ability to make valid assertions 74 University of Ghana http://ugspace.ug.edu.gh about sanitary pad more than the male celebrity endorser. They also saw the male celebrity endorser as having the capacity to better talk about shave cream than the female celebrity endorser. Interesting though is the fact that across the neutral product, participants scored a slightly higher believability for the female celebrity endorser than the male celebrity endorser. This finding somehow contradicts the argument by Kumar and Singh (2015) that non-gender products (neutral gender products) are best endorsed by male celebrity endorsers. 6.3 Endorser Gender -Product Gender Fit Implications for Believability Part of the difficult and risky areas of celebrity endorsement is choosing the perfect celebrity to go with a product (Till & Busler, 2000). Since the celebrity assumes the position of the originator of the message, there is the need to make sure the pairing looks as real as possible. Accordingly, the second objective of this study looked at whether perceived fit between endorser’s gender and product gender affects advertisement believability. The objective was explored through four hypotheses. Interestingly, findings suggest that when the audiences perceive a good fit between a celebrity endorser’s gender and a product’s gender, ad believability is enhanced. The study saw that it did not matter whether there were consistent or inconsistent characteristics between the celebrity endorser and the product, if participants perceived a fit, their evaluation was favorable. The study found for the first and second hypotheses that perception of fit between a female celebrity endorser and a feminine product enhances believability. Also, perception of fit between a male celebrity endorser and a masculine product enhances believability. These findings clearly support the match-up hypothesis model which holds that when characteristics of the endorser match with the characteristics of the endorsed product, evaluation of the endorsement tends to be 75 University of Ghana http://ugspace.ug.edu.gh positive. The findings also confirms Boyd and Shank (2004) who argued that an ad message is more effective when there is a match-up between an endorser’s gender and the gender image of the endorsed product. Also, the findings reflect Petrevu’s (2001) argument that when a female celebrity endorses feminine gendered products (such as bra, cosmetics etc.), audiences believe her more because she is a woman and can personally experience the product. Furthermore, Misra and Beatty (1990) also examined the effect of endorser/brand image congruity on brand recall and found that congruity of endorser and brand image significantly impact audience brand recall. Contrary to the Match-up hypothesis model and findings of some scholars (Lien et al., 2012; Boyd & Shank, 2004), the study also saw from the third and fourth hypotheses that perceived match between a male celebrity endorser and a feminine product as well as perceived match between a female celebrity endorser and a masculine product enhanced advertisement believability. In other words, even when the gender of the endorser is different from that of the product category with which it is paired, believability is still strong so far as participants perceive a good fit between the individual celebrity endorser and the product. This finding somehow is in consonance with Lee’s (2013) argument that unexpected information (for example a mismatch between a celebrity endorser and a product) creates relatively persistent and favorable attitudes. Likewise, the findings also support Lee and Thompson (2008) who examined how different degrees of celebrity-product incongruence influence the persuasiveness of an advertisement. They found that a moderate mismatch and not a complete match between a celebrity’s image and a product’s image produced favorable responses to advertisements. However, the findings contrast with Rifon’s (2012) evidence that consumers tend to believe that 76 University of Ghana http://ugspace.ug.edu.gh the image of a celebrity endorser must blend perfectly with the image of the endorsed product. It also counteracts Roy and Bailey’s (2015) assertion that improper pairing of a celebrity with a brand could have negative consequences for the brand. The results however affirm what some researchers (e.g. Levert, 2018; Lee, 2013; Lee & Thompson 2008) have reported that a positive fit between an endorser and a product is not always necessary. According to these researchers, a relatively poor fit or mismatch between a celebrity endorser’s gender and a product’s gender does not necessarily lead to negative evaluation of the advertisement. Even though it seems that a perfect blend of celebrity and product characteristics is not the only strict indicator of positive evaluation of the celebrity endorsed advertisement, scholars and practitioners have been cautioned that ‘inconsistent’ celebrity-product permutations must be perceived by audience members as interesting and positive before they work (Meyers-Levy and Tybout 1989). More importantly, D’Astous and Bitz, (1996) have argued that if the incongruence is so great that audiences have to stretch themselves to understand it, they are likely to exhibit negative reactions, leading to negative cognitive elaboration and a feeling of frustration. In the end, they might have a negative evaluation about the adverts. Besides, according to Fleck- Dousteyssier and Korchia (2006), a brand/celebrity pair is relevant if it is meaningful and contributes to clear communication of a message about the brand. They further stated that the pairing of a celebrity and a product is expected if it corresponds to a predefined expectation about the brand. This suggests that inconsistent or ‘abnormal’ endorser and product pairings should not be merely guess works or a trial and error exercise. In fact, the combinations must meet pre- existing expectations about the brand. In the case of this study, participants’ perception of fit between the female celebrity endorser and the masculine product for example, led to their 77 University of Ghana http://ugspace.ug.edu.gh believability of the advert. In Ghana, it is common to find women buying masculine gendered products for their husbands, brothers, sons, uncles etc.; as it probably is in other countries. The reverse may also be true about Ghanaian men. As such, one may not be wrong to argue that the above celebrity- product combination (female celebrity endorser and shave cream) sort of does not really deviate from predefined expectations about the brand as it is possible for both male and female to be seen either purchasing or selling products whose gender image are different from theirs. One may say that this may explain why participants perceived a fit and eventually believed the advertisement. Participants’ new definitions of fit between endorser gender and product gender could be said to be influenced by their educational as well as social media exposure levels. The sample consists of university students who are more likely to have exposure to social media. Presently, discussions on emerging gender interpretations on social media have become more open. It is no longer news for some men to be seen associating with sanitary pad or some women buying shave creams. This varied interpretations people are giving to gender could be said to influence the new perspectives of what should belong to a man or a woman and even who really a man or a woman is. Thus, one would argue that these currents trends about gender interpretations are contributing to new definitions of fit between endorser gender and product gender. This therefore shows that the pairing of celebrity endorsers with products must always be done in the lens of audience members because if they perceive a fit, irrespective of the difference that might be existing between the product and the endorser, they may have a positive assessment of it. 78 University of Ghana http://ugspace.ug.edu.gh In summary, the major findings of the study revealed that a celebrity endorser’s gender based on which product is being advertised, increases believability of the endorsed advertisement. It was also found that perceived fit between a female celebrity endorser and a feminine product enhances believability. Again, perceived fit between a male celebrity endorser and a masculine product enhances believability. Moreover, it was realized that perceived fit between a female celebrity endorser and a masculine product also increases advertisement believability. Finally, it was found that perceived fit between a male celebrity endorser and a masculine product positively affects advertisement believability. 6.4 Chapter Summary This chapter presented a discussion on the findings. The researcher constructed a narrative around the findings based on the theoretical models and evidence from literature. Findings were discussed in relation to the objectives they fall under. For instance, the findings revealed that the gender of a celebrity endorser has consequences for ad believability depending on the nature and gender of the product which is being advertised. Also, perceived match between celebrity endorser gender and product gender enhances believability. Interestingly, the study showed that gender indeed matters in the assessment of celebrity endorsed advertisements. Also, the scope of the match-up hypothesis model was challenged by the findings. Customarily, the match-hypothesis model defines endorser gender-product gender match to be when there is consistency between the gender of the endorser and the product. However, the results of this thesis has proven that even when there are inconsistencies between the gender of an endorser and the product, audiences may still perceive a fit/match and this could significantly affect their responses. 79 University of Ghana http://ugspace.ug.edu.gh CHAPTER SEVEN SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 7.0 Introduction This research sought to investigate implications of gender in celebrity endorsed advertisements. With the aid of a 2×3 full factorial within-subject experimental design, the study tested participants’ perceptions about endorsements of three product categories by a male and a female celebrity. Grounded in the source credibility model and the match-up hypothesis model, the study explored two major objectives. The objectives were to examine the nature of relationship between a celebrity endorser’s gender and advertisement believability and as well test whether perceived endorser gender-product gender fit enhances advertisement believability. This chapter presents the summary, conclusions, limitations and recommendations of the study. 7.1 Summary The study used 108 undergraduate students from the University of Ghana main campus as participants in an online experiment. The first aim of the experiment was to examine the nature of relationship between a celebrity endorser’s gender and advertisement believability. Guided by this objective, the study tested whether: A. Endorsement of a feminine product by female celebrity endorsers enhances ad believability B. Endorsement of a masculine product by male celebrity endorsers enhances ad believability C. Endorsement of a feminine product by male celebrity endorsers enhances ad believability 80 University of Ghana http://ugspace.ug.edu.gh D. Endorsement of a masculine product by female celebrity endorsers enhances ad believability E. Endorsement of a neutral product by female celebrity endorsers enhances ad believability F. Endorsement of a neutral product by a male celebrity endorser enhances ad believability The research also investigated whether perceived celebrity endorser gender-product gender fit enhances advertisement believability. This objective was explored through four hypotheses which were premised on the match-up model and findings from other scholars (Lien, 2012; Klause & Bailey, 2008). The hypotheses were: } H1: Perceived match between a female endorser and a feminine product will enhance believability of the endorsed advertisement. } H2: Perceived match between a male endorser and a masculine product will enhance believability of the endorsed advertisement. } H3: Perceived match between a female endorser and a masculine product will not enhance believability of the endorsed advertisement } H4: Perceived match between a male endorser and a feminine product will not enhance believability of the endorsed advertisement After a descriptive and a linear regression analysis, the major findings were that the gender of a celebrity endorser has an implication on audience believability of the endorsed advertisement 81 University of Ghana http://ugspace.ug.edu.gh depending on the nature and gender of the advertised product. Also, perceived match between a male celebrity endorser and a masculine product enhances ad believability. Moreover, perceived match between a female celebrity endorser and a feminine product enhances ad believability. Also, perceived match between a female celebrity endorser and a masculine product enhances ad believability. Finally, perceived match between a male celebrity endorser and a feminine product also enhances ad believability. 7.2 Conclusions/recommendations This study has made interesting discoveries which support and challenge previous knowledge on gender in celebrity endorsed advertisements. For example, the study confirmed assumptions of the match-up hypothesis model that perceived fit between celebrity endorser and the product which is being advertised, has an impact on positive advertisement evaluations. The study also challenged findings from other scholars which showed that gender does not have any influence on audiences’ assessment of celebrity endorsements. The results of this study show that gender (of a celebrity endorser and the endorsed product), significantly affects advertisement believability. The findings of the study have the following implications for scholarship and practice. First, from the results for objective one, it could be argued that endorser characteristics, depending on the nature of the advertised product influence audience perceptions. From the study, when the female endorser endorsed the sanitary pad and the male celebrity endorser also endorsed the shave cream, participants tended to believe both ads. The reason may be ascribed to Ghanaian traditional standpoints that sanitary pad represents womanhood therefore a woman must be the best person to talk about it. In the same way, shave cream is largely for men therefore a man could be the one 82 University of Ghana http://ugspace.ug.edu.gh who would know how best to talk about it. The above result confirms assumptions of the source credibility model indicating its relevance to the Ghanaian context as well. Furthermore, it could be inferred from these findings and reports by other scholars that gender effects in celebrity endorsements are but reflections of the gender constructions across societies. Support for this assertion can be seen in Silvera and Austad’s (2003) argument that cultural differences among societies moderate how celebrity endorsement works among audiences. This means that ratings of celebrity endorsed advertisements especially in relation to the endorser’s gender and the kind of product he or she can endorse is a matter of the lens from which a particular society views the concept gender and the acceptable way men and women are expected to behave in that society. It is thus recommended that a cross cultural study be conducted to compare gender effects in celebrity endorsements across different settings. Results from such a study could deepen understanding on gender in the celebrity endorsement literature. The study also revealed that a celebrity endorser’s gender depending on the nature of the advertised product was able to influence the level of trust participants had about the endorsements. This somehow reflects assumptions of the match-up hypothesis model and suggests that careful thought should be given to the selection of celebrity endorsers to appear with products. Thus, although the study did not actually involve advertisers and creative directors in Ghana, yet it will not hesitate to make the recommendation that future studies could reach out to them (advertisers and creative directors) to find out the procedures they undertake in selecting celebrities to appear with products. Empirical evidence about processes these advertisers and creative directors go through in choosing celebrities for the endorsement of products will provide additional information for a more holistic understanding about celebrity-product pairing. 83 University of Ghana http://ugspace.ug.edu.gh Moreover, the study revealed that a positive or negative match between an endorser’s gender and a product’s gender does not in itself lead to a positive or negative evaluation of an endorsement. Rather, what is more important is whether audiences perceive a fit. This means practitioners should rather spend time to understand what audience members view as a match rather than making abstract combinations of celebrities and products based on what they themselves think is a match. The findings also imply that Ghanaian advertising audiences are becoming liberal with respect to their definitions of what a match between an endorser’s gender and a product’s gender should be. There is therefore the need for a more open discussion on the part of both scholars and advertising practitioners regarding audience members’ definitions of celebrity endorser gender-product gender congruence. Furthermore, the findings revealed that the Match-up hypothesis model is still very effective in explaining how consistent characteristics of a celebrity endorser and the advertised product induce a favorable evaluation of the endorsement. However, there is the need for an extended conversation on how even a mismatch between an endorser and the advertised product leads to a positive perception of fit and hence a favorable assessment of the advertisement. The theory limits endorser-product fit only to when there is a positive match between an endorser’s characteristics and the characteristics of the endorsed product. Therefore, one may conclude that although the traditional explanation of what is fit (as proposed by the match-up hypothesis model) may work, what audience members see as fit (which might even be a negative match between the endorser and product) also may induce a positive evaluation of the advertisement. This suggests for further enquiries on endorser-product congruity to not over-rely on the match-up hypothesis model as a straitjacket way of explaining endorser-product congruence. 84 University of Ghana http://ugspace.ug.edu.gh Finally, the study realized that when it came to the neutral product (bottled water), the female celebrity endorser drew a slightly higher believability than the male celebrity endorser. This somehow suggests (though weakly) that female celebrities may be chosen over male celebrities to appear with neutral products. Meanwhile, Kumar and Singh (2015) have reported that male celebrities other than female celebrities are the best people to endorse neutral products. It is thus recommended that future researchers could engage a number of different neutral product categories to be able to draw a more solid conclusion regarding which endorser type (male celebrity or female celebrity) is the most appropriate for endorsing such product category and even whether gender of celebrities increases their suitability for endorsing certain types of neutral products. 7.3 Limitations/recommendations The following limitations ought to be noted to shape the direction of future enquiries. The study failed to adequately control for celebrity personality. Even though participants were specifically instructed in the questionnaire to indicate believability exclusively based on the celebrity’s gender, it is assumed that their different attitude towards the celebrities could still influence how they perceived the endorsements. The study recommends that future studies should do a pre-test to choose celebrities with same or very close personality evaluations by audiences so as to rule out the possibility of confounds which could be introduced by differences in the personalities of the selected celebrities. Moreover, although the study used celebrities, the main focus was on their gender thus, from the findings, it is highly possible for the gender of non-celebrity endorsers to also affect believability. The study therefore recommends that future studies could compare gender effects in both celebrity and non-celebrity endorsed advertisements. 85 University of Ghana http://ugspace.ug.edu.gh Furthermore, the study’s design changed due to the COVID 19 pandemic. Initially, the researcher planned to have a physical experiment where all participants would be assembled together and be exposed to the experimental stimulus at the same time. However, an online experiment became the only option. There was therefore the need to make sure participants are gathered online and are made to see the experimental conditions at the same time. The researcher encountered many challenges in conducting the online experiment most of which were beyond control. For example, treatments were shared on WhatsApp pages for participants and the researcher perceived that since most of them were class mates, they could have discussed their responses to the various questions. Also, the experiment could not be done as a one-off exercise since it was very difficult to assemble all participants online at one particular time. The experiment was thus conducted in batches and even though participants left the respective pages after completion, there is the possibility they could share the links with their other colleagues who were yet to do the study to have a prior knowledge about the treatments. All these could influence their responses and hence the outcome of the study. A physical experiment is recommended to future researchers especially when participants are drawn from the same classes, departments, colleges etc. Finally, since the study’s sample is made of university students, it is possible that their educational and media exposure levels could have effects on their responses. Therefore, a non-student sample is recommended for future studies so as to test whether the findings will be replicated. Lastly, it would be interesting to explore how other circumstances such as (attractiveness of an endorser; perception of an expensive or a cheap celebrity endorser; and when the same celebrity endorser endorses competitive brands) affect audiences’ response to celebrity endorsement. Since 86 University of Ghana http://ugspace.ug.edu.gh the scope of this thesis did not allow the researcher to investigate these issues, future studies can look into them in an attempt to build the celebrity endorsement literature in Ghana. Also, this study defined gender strictly as male and female. The findings of the study therefore could not be extended to other gender orientations like transgender, bisexual etc. 7.5 Chapter Summary This section attempted to draw informed conclusions from the discussion. For instance, the study concluded that gender (of a celebrity endorser and the endorsed product), significantly affect believability of celebrity endorsements. The researcher also made some recommendations for future studies and ended with the study’s limitations. One of these recommendations was the need to broaden the scope of the match-up hypothesis model to cater for the situation where even a mismatch between an endorser’s gender and the gender image of the advertised product could lead to a positive perception of fit and hence a favorable assessment of the advertisement. The researcher also made suggestions regarding how the points outlined as limitations of the study could be addressed in future studies. For instance in the case of ruling out all possible confounds, the researcher suggested that a pre-test be conducted in future studies, to select celebrities who are perceived equally in terms of personality. So that, only their gender could be the only factor to explain audiences response to the advertisement they would be featuring in. 87 University of Ghana http://ugspace.ug.edu.gh REFERENCES Ansu-Mensah, P., Asuamah, S. Y., & Amankwah, J. (2013). Consumers’ attitude towards advertisement elements: A survey of marketing students in Sunyani Polytechnic, Ghana, West Africa. International Journal of innovative research in management, 2(4), 13-24. Atkin, J. L., & Beltramini, R. F. (2007). Exploring the perceived Believability of DTC advertising in the US. Journal of Marketing Communications, 13(3), 16 180. https://doi.org/10.1080/13527260701250695 Baah Entsiful, H., Dawson-Ahmoah, E., & Mensah, P. (2013). GLO Ghana’s celebrity endorsements and student purchase (Doctoral dissertation). Belch, G. E., & Belch, M. A. (2013). Advertising and promotion: An integrated marketing communications perspective. New York, NY: McGraw-Hill Education. Beltramini, R. F. (2006). Consumer Believability of information in direct-to-Consumer (DTC) advertising of prescription drugs. Journal of Business Ethics, 63(4), 333-343. https://doi.org/10.1007/s10551-005-4711-2 Bitz, P. (1995). Consumer evaluations of sponsorship programmes. European Journal of Marketing. Boyd, T. C., & Shank, M. D. (2004). Athletes as Product Endorsers: The Effect of Gender and Product Relatedness. Sport Marketing Quarterly, 13(2). Cacioppo, J. T., & Petty, R. E. (1984). The elaboration likelihood model of persuasion. ACR North American Advances. Carsky, M. L., & Zuckerman, M. E. (1991). In search of gender differences in marketing communication: An historical/contemporary analysis. ACR Gender and Consumer Behavior. Choi, S. M., & Rifon, N. J. (2012). It is a match: The impact of congruence between celebrity image and consumer ideal self on endorsement effectiveness. Psychology & Marketing, 29(9), 639-650. https://doi.org/10.1002/mar.20550 Courtright, J. A. (1996). Standpoint: Rationally thinking about nonprobability. D′Astous, A., & Bitz, P. (1995). Consumer evaluations of sponsorship programmes. European Journal of Marketing, 29(12), 6-22. https://doi.org/10.1108/03090569510102504 88 University of Ghana http://ugspace.ug.edu.gh Dang, Huy. (2020). Re: How to deal with cross loadings in Exploratory Factor Analysis?. Retrievefrom:https://www.researchgate.net/post/How_to_deal_with_cross_loadings_in_Ex ploratory_Factor_Analysis/5efd7afd40c451634674020e/citation/download. Davies, F., & Slater, S. (2015). Unpacking celebrity brands through unpaid market communications. Journal of Marketing Management, 31(5-6), 665-684. https://doi.org/10.1080/0267257x.2014.1000941 Diabah, G. (2018). The representation of women in Ghanaian radio commercials: Sustaining or challenging gender stereotypes? Language in Society, 48(2), 261-283. https://doi.org/10.1017/s0047404518001343 Dzisah W. E, & Ocloo, C. E. (2013). Celebrity endorsement and consumer buying behaviour; enhancing the promotion function of marketing in the central business area of Accra, Ghana. European Journal of Business and Management. Edem Adzovie, D., Holm Adzovie, R., & Boateng, E. (2019). Gender in audio-visual advertisements in Ghana: A semiotics analysis. Proceedings of The World Conference on Social Sciences. https://doi.org/10.33422/worldcss.2019.09.542 Edwards, S. M., & La Ferle, C. (2009). Does gender impact the perception of negative information related to celebrity endorsers? Journal of Promotion Management, 15(1-2), 22-35. https://doi.org/10.1080/10496490902837940 Eli-Zafoe, D. (2013). Analyzing the effectiveness of celebrity endorsement on consumer buying behavior in the Telecommunication industry in Ghana (Doctoral dissertation). Erba, J., Ternes, B., Bobkowski, P., Logan, T., & Liu, Y. (2018). Sampling methods and sample populations in quantitative mass communication research studies: A 15-Year census of six journals. Communication Research Reports, 35(1), 42-47. https://doi.org/10.1080/08824096.2017.1362632 Erdogan, B. Z. (1999). Celebrity endorsement: A literature review. Journal of Marketing Management, 15(4), 291-314. https://doi.org/10.1362/026725799784870379 89 University of Ghana http://ugspace.ug.edu.gh Erfgen, C., Zenker, S., & Sattler, H. (2015). The vampire effect: When do celebrity endorsers harm brand recall? International Journal of Research in Marketing, 32(2), 155-163. https://doi.org/10.1016/j.ijresmar.2014.12.002 Farhat, R., & Khan, B. M. (2011). Celebrity endorsement: A congruity measure of personalities. Research for Rural Development, 19(9), 1-7. Fathi, F., & Kheiri, B. (2015). Celebrity endorsement and its effect on consumer behavior: A literature review. International Journal of Review in Life Sciences, 5(10), 619-626. Fink, J. S., Cunningham, G. B., & Kensicki, L. J. (2004). Using athletes as endorsers to sell women’s sport: Attractiveness vs. expertise. Journal of Sport management, 18(4), 350-367. Fink, J. S., Parker, H. M., Cunningham, G. B., & Cuneen, J. (2012). Female athlete endorsers: Determinants of effectiveness. Sport Management Review, 15(1), 13-22. Freiden, J. B. (1984). Advertising spokesperson effects-An examination of endorser type and gender on 2 audiences. Journal of advertising research, 24(5), 33-41. Friedman, H. H., Termini, S., & Washington, R. (1976). The Effectiveness of Advertisements Utilizing Four Types of Endorsers. Journal of Advertising, 5(3). Ghana* Hofstede Insight. https://www.hofstede-insights.com/country/ghana/. Gravetter, F. J., & Forzano, L. A. B. (2018). Research methods for the behavioral sciences. Cengage Learning. Guido, G., Peluso, A. M., & Moffa, V. (2011). Beardedness in advertising: Effects on endorsers' credibility and purchase intention. Journal of Marketing Communications, 17(1), 37-49. https://doi.org/10.1080/13527260903157383 Gupta, V. (2017). Celebrity Endorsement-An Analysis of Brand Image and Celebrity Image. International Journal of Applied Marketing and Management, 2(2). Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2009). Multivariate Data Analysis 7th Edition Pearson Prentice Hall. Hansen, I., Erlandsson, R., & Mokhtari, H. (2013). Celebrity Endorsement: A gender perspective of consumer behavior in the fashion industry. Hinton, P. R. (2014). Statistics explained. London: Routledge. Hoyer, W. D., & MacInnis, D. J. (2004). Consumer behavior. Boston, MA: Houghton Mifflin. 90 University of Ghana http://ugspace.ug.edu.gh https://www.nexttv.com/news/nielsen-global-survey-finds-believability-key-advertising- effectiveness-127624 Hussain, D. (2019). Celebrity use in Indian advertising: Analysis and appraisal. International Journal of Indian Culture and Business Management, 1(1), 1. https://doi.org/10.1504/ijicbm.2019.10019277 Kamins, M. A., Brand, M. J., Hoeke, S. A., & Moe, J. C. (1989). Two-sided versus one-sided celebrity endorsements: The impact on advertising effectiveness and credibility. Journal of Advertising, 18(2), 4-10. https://doi.org/10.1080/00913367.1989.10673146 Kang, J., Hubbard, G. T., & Hong, S. (2019). Gender and credibility in branded storytelling. Gender in Management: An International Journal, 34(8), 702-714. https://doi.org/10.1108/gm-02-2019-0015 Kasana, J., & Chaudhary, N. (2014). Impact of celebrity endorsement on consumer buying behavior: A descriptive study. International Research Journal of Commerce Arts and Science, 5(2), 319-347. Kim, S. S., Lee, J., & Prideaux, B. (2014). Effect of celebrity endorsement on tourists’ perception of corporate image, corporate credibility and corporate loyalty. International Journal of Hospitality Management, 37, 131-145. Kim, S., Kim, S. & Han, H. (2018). Effects of TV drama celebrities on national image and behavioral intention. Asia Pacific Journal of Tourism Research, 24(3), 233-249. https://doi.org/10.1080/10941665.2018.1557718 Klaus, N., & Bailey, A. A. (2008). Celebrity endorsements: An examination of gender and consumers’ attitudes. American Journal of Business, 23(2), 53-62. https://doi.org/10.1108/19355181200800010 Korchia, M., & Fleck-Dousteyssier, N. (2006). Celebrities in advertising: The role of congruency (No. 123456789/4202). Paris Dauphine University. Kumar, R., & Singh, N. (2015). Impact of Celebrity Endorsement on Consumer Buying Behaviour—A Case Study of Durable Products in District Udhampur, J & K. Asian Journal of Multidisciplinary Studies, 3(7), 201-208. 91 University of Ghana http://ugspace.ug.edu.gh Lazăr, L. (2020). Gender influence on the perception of celebrity endorsement. Studia Universitatis, Vasile Goldis” Arad – Economics Series, 30(2), 13-23. https://doi.org/10.2478/sues-2020-0009 Lee, J., & Thorson, E. (2008). The impact of celebrity–product incongruence on the effectiveness of product endorsement. Journal of Advertising Research, 48(3), 433-449. https://doi.org/10.2501/s0021849908080446 Leedy, P. D., & Ormrod, J. E. (2001). Practical Research: Planning and Design, Merrill Prentice Hall. New Jersey. Levert, K. A. D. (2018). Congatulations! You have a match!: How (in) congruence between the type of endorser and type of product, along with innovativeness of the consumer, affects a consumer's purchase intention, the endorser personality evaluation and product personality evaluation (Master's thesis, University of Twente). Lien, N., Chou, H., & Chang, C. (2012). Advertising effectiveness and the match-up hypothesis: Examining spokesperson sex, attractiveness type, and product image. Journal of Current Issues & Research in Advertising, 33(2), 282-300. https://doi.org/10.1080/10641734.2012.700809 Lin, T. M., Lu, K., & Wu, J. (2012). The effects of visual information in eWOM communication. Journal of Research in Interactive Marketing, 6(1), 7-26. https://doi.org/10.1108/17505931211241341 Lutz, C. (1985). Ethnopsychology compared to what? Explaining behavior and consciousness among the Ifaluk. Lutz, R. J. (1985). Affective and cognitive antecedents of attitude toward the ad: A conceptual framework. Psychological process and advertising effects: Theory, research, and application, 45-63. Mac-Kingsley, I., & Kalu, S. E. (2016). Impact of celebrity endorsements on brand loyalty in sports products. International Journal of Engineering and Advanced Technology, 8(6S3), 473-478. https://doi.org/10.35940/ijeat.f1087.0986s319 Magaji, M., Coordinators, S., Galadima, D. J., & Dana’an, M. G. Seminar, (2012). Presentation On Celebrity Endorsement: Relevance And Risks In Nigerian Advertising. 92 University of Ghana http://ugspace.ug.edu.gh McCracken, G. (1989). Who is the celebrity endorser? Cultural foundations of the endorsement process. Journal of Consumer Research, 16(3), 310. https://doi.org/10.1086/209217 Menon, K., Boone, L. E., & Rogers, H. P. (2001). Celebrity Advertising: An assessment of its relative effectiveness. Unpublished manuscript. Meyers-Levy, J., & Tybout, A. M. (1989). Schema congruity as a basis for product evaluation. Journal of Consumer Research, 16(1), 39. https://doi.org/10.1086/209192 Min, J. H. J., Chang, H. J. J., Jai, T. M. C., & Ziegler, M. (2019). The effects of celebrity-brand congruence and publicity on consumer attitudes and buying behavior. Fashion and Textiles, 6(1), 10. Min, J., Ziegler, M., Chang, H. J., & Jai, T. (. (2017). The effects of celebrity-brand congruence and publicity on consumer attitudes, buying intention, and loyalty. https://doi.org/10.31274/itaa_proceedings-180814-1804 Mishra, A. S., Roy, S., & Bailey, A. A. (2015). Exploring brand personality-celebrity endorser personality congruence in celebrity endorsements in the Indian context. Psychology & Marketing, 32(12), 1158-1174. https://doi.org/10.1002/mar.20846 Mishra, A. S., Roy, S., & Bailey, A. A. (2015). Exploring brand personality-celebrity endorser personality congruence in celebrity endorsements in the Indian context. Psychology & Marketing, 32(12), 1158-1174. https://doi.org/10.1002/mar.20846 Misra, S., & Beatty, S. E. (1990). Celebrity spokesperson and brand congruence. Journal of Business Research, 21(2), 159-173. https://doi.org/10.1016/0148-2963(90)90050-n Moy, P., & Murphy, J. (2016). Problems and prospects in survey research. Journalism & Mass Communication Quarterly, 93(1), 16-37. Nyarko, I. K., Asimah, V., Agbemava, E., & Tsetse, E. K. (2015). The influence of celebrity endorsement on the buying behavior of the Ghanaian youth: A study of fan milk Ghana ads. International Journal of Business and Management Review, 3(11), 1-16. Osei-Frimpong, K., Donkor, G., & Owusu-Frimpong, N. (2019). The impact of celebrity endorsement on consumer purchase intention: An emerging market perspective. Journal of marketing theory and practice, 27(1), 103-121. Park, H. J., & Lin, L. M. (2020). The effects of match-UPS on the consumer attitudes toward internet celebrities and their live streaming contents in the context of product endorsement. 93 University of Ghana http://ugspace.ug.edu.gh Journal of Retailing and Consumer Services, 52, 101934. https://doi.org/10.1016/j.jretconser.2019.101934 Parker, H., Mudrick, M., & Fink, J. (2018). The Impact of Gender Expression on Female Athlete Endorser Effectiveness. Sport Marketing Quarterly, 27(4), 264–274. https://doi.org/10.32731/smq.274.122018.05 Peetz, T. B., Parks, J. B., & Spencer, N. E. (2004). Sport Heroes as Sport Product Endorsers: The Role of Gender in the Transfer of Meaning Process for Selected Undergraduate Students. Sport Marketing Quarterly, 13(3). Petrescu, M., Korgaonkar, P., & Gironda, J. (2015). Viral advertising: A Field experiment on viral intentions and purchase intentions. Journal of Internet Commerce, 14(3), 384-405. https://doi.org/10.1080/15332861.2015.1080057 Petrescu, M., Mingione, M., Gironda, J., & Brotspies, H. (2019). Ad scepticism and retouch-free disclaimers: Are they worth it? Journal of Marketing Communications, 25(7), 738-762. https://doi.org/10.1080/13527266.2018.1437552 Petty, R. E., & Cacioppo, J. T. (1984). The elaboration likelihood model of persuasion. Communication and Persuasion, 1-24. https://doi.org/10.1007/978-1-4612-4964-1_1 Phang, G., & de Run, E. C. (2007). Celebrity endorser and respondents gender: Its impact on company, behavioral and attitudinal variables. Jurnal Kemanusiaan, 5(2). Pornpitakpan, C. (2003). Validation of the celebrity endorsers' credibility scale: Evidence from asians. Journal of Marketing Management, 19(1), 179-195. https://doi.org/10.1362/026725703763772015 Putrevu, S. (2001). Exploring the origins and information processing differences between men and women: Implications for advertisers. Academy of marketing science review, 10(1), 1-14. Rajaobelina, L., Lacroix, C., & St-Onge, A. (2019). Does experiential advertising impact credibility? International Journal of Bank Marketing, 37(1), 171-191. https://doi.org/10.1108/ijbm-01-2017-0021 Rogers, J., & Révész, A. (2019). Experimental and quasi-experimental designs. The Routledge Handbook of Research Methods in Applied Linguistics, 133-143. https://doi.org/10.4324/9780367824471-12 94 University of Ghana http://ugspace.ug.edu.gh Rogers, J., & Révész, A. (2020). Experimental and quasi-experimental designs. The Routledge handbook of research methods in applied linguistics. New York: Routledge. Ross, S. M., & Morrison, J. R. (2014). Measuring meaningful outcomes in consequential contexts: Searching for a happy medium in educational technology research (Phase II). Journal of Computing in Higher Education, 26(1), 4-21. https://doi.org/10.1007/s12528-013-9074-6 Rowley, M., Gilman, H., & Sherman, S. M. (2019). Investigating the celebrity effect: The influence of well-liked celebrities on adults’ explicit and implicit attitudes to brands and brand choice. Psychology of Popular Media Culture, 8(4), 402-409. https://doi.org/10.1037/ppm0000199 Rozendaal, E., Lapierre, M. A., Van Reijmersdal, E. A., & Buijzen, M. (2011). Reconsidering advertising literacy as a defense against advertising effects. Media Psychology, 14(4), 333- 354. https://doi.org/10.1080/15213269.2011.620540 Samarasinghe, H.M.U.S.R. (2018). “Moderating Role of Consumer’s Gender on Effectiveness Celebrity Endorsement towards Consumer’s Purchasing Intention” Global Journal of Management and Business Research: EMarketing. Saunders, M. N., & Lewis, P. (2012). Doing research in business & management: An essential guide to planning your project. New York: Pearson. Shan, Y., Chen, K., & Lin, J. (. (2019). When social media influencers endorse brands: The effects of self-influencer congruence, parasocial identification, and perceived endorser motive. International Journal of Advertising, 39(5), 590-610. https://doi.org/10.1080/02650487.2019.1678322 Shapiro, M. A. (2002). Generalizability in communication research. Human Communication Research, 28(4), 491-500. https://doi.org/10.1111/j.1468-2958.2002.tb00819.x Silvera, D. H., & Austad, B. (2004). Factors predicting the effectiveness of celebrity endorsement advertisements. European Journal of Marketing, 38(11/12), 1509-1526. https://doi.org/10.1108/03090560410560218 Soh, H., Reid, L. N., & King, K. W. (2009). Measuring trust in advertising. Journal of Advertising, 38(2), 83-104. https://doi.org/10.2753/joa0091-3367380206 Soni, A. (2016). Impact of celebrity endorsed advertisements on gender in attitude Formation. BVIMSR’s Journal of Management Research, 8(2), 157. 95 University of Ghana http://ugspace.ug.edu.gh Sridevi, J. (2012). Effectiveness of celebrity endorsement in brand recall and brand recognition. Zenith International Journal of Business Economics & Management Research, 2(5), 203-209. Stanley, S. M., Clow, K., & James, K. E. (2011). The impact of visual strategy and race and gender congruency on source credibility of print advertisements. Marketing Management Journal., 21(2), 81-94. Thomas, R. J., Cook, J. A., McGrew, C., King, J. H., Pulsipher, D. T., Yeo, R. A. & Campbell, R. A. (2018). Convergent and discriminant validity of the ImPACT with traditional neuropsychological measures. Cogent Psychology. 5(1), 1430199. Thomas, T., & Johnson, J. (2019). The effect of celebrity trustworthiness on endorsement effectiveness: A comparison of congruence and hybrid model. Vision: The Journal of Business Perspective, 23(3), 275-286. https://doi.org/10.1177/0972262919850922 Yen, C. H., & Teng, H. Y. (2015). Celebrity involvement, perceived value, and behavioral intentions in popular media-induced tourism. Journal of Hospitality & Tourism Research, 39(2), 225-244. Zakari, M., Dogbe, C. S., & Asante, C. (2019). Effect of celebrity endorsement on telecommunication companies’ reputation. Management Research Review, 42(12), 1297- 1314. https://doi.org/10.1108/mrr-12-2018-0470 96 University of Ghana http://ugspace.ug.edu.gh APPENDIX I: EXPERIMENTAL TREATMENTS AND QUESTIONNAIRE Dear respondent, Welcome to this study! I am an MPhil student in the Department of Communication Studies, University of Ghana. My experiment seeks to investigate effects of gender of a celebrity endorser and the advertised product on audiences’ perceptions of celebrity endorsed advertisements. The experiment which is expected to take 30 minute to an hour of your time is in two parts. In the first part, there are few sets of questions meant to understand your demographic background, your celebrity involvement level and your advertising literacy among others. In the second part, you would be presented with pictures of celebrities and products to view after which you will be asked to answer some questions. Kindly read the instructions on each section carefully and answer the underlying questions as best as you can. All information provided will be kept confidential. No participant will be directly identified in any way during analysis and discussion of the data. Please know that you are free to opt out of the study at any point in time when you feel you no longer want to participate. For further information or questions regarding this research, you can kindly contact my supervisors, Dr. Sarah Akrofi-Quarcoe and Dr. Abena Animwaa Yeboah-Banin or the researcher on 0245626763/amensah161@gmail.com Sincerely, Alice Mensah Department of Communication Studies, UG Do you agree to participate in this study? 97 University of Ghana http://ugspace.ug.edu.gh SECTION A: DEMOGRAPHIC INFORMATION GENDER LEVEL OF EDUCATION a. Male a. 100 b. Female b. 200 c. 300 d. 400 WHAT IS YOUR AGE? a) 18-23 b) 24-29 c) 30-36 SECTION B Perceptions about Advertising and Celebrity Endorsement Please indicate the extent of your agreement or disagreement with each statement. As a person, 1. S.D – Strongly Disagree S.D D N A SA 2. D- Disagree 3. N -Neutral 4. A- Agree 5. SA- Strongly agree 98 University of Ghana http://ugspace.ug.edu.gh I think adverts are meant to make me think 1 2 3 4 5 positively about the advertised product. 2. I think adverts always tell the truth 1 2 3 4 5 3. I think adverts are targeted at a specific 1 2 3 4 5 group of people at all times 4. Celebrities are often used in adverts to 1 2 3 4 5 encourage purchase 5. I think adverts tell half truth 1 2 3 4 5 6. I think adverts are paid for by someone 1 2 3 4 5 to influence people As a person, I find a lot of my life is organized around celebrities 1 2 3 4 5 Celebrity activities have a central role in my life 1 2 3 4 5 I enjoy discussing celebrity activities with my friends 1 2 3 4 5 I find a lot of my life is organized around celebrity 1 2 3 4 5 activities As an individual, I learn about fashions and about what to wear to 1 2 3 4 5 impress others from celebrity endorsed ads Celebrity endorsed ads are a valuable source of 1 2 3 4 5 information about local sales 99 University of Ghana http://ugspace.ug.edu.gh celebrity endorsed ads tell me what people with 1 2 3 4 5 lifestyle similar to mine are buying and using Celebrity endorsed ads tell me which brands have the 1 2 3 4 5 features I am looking for Celebrity endorsed ads present a true picture of the 1 2 3 4 5 product advertised Celebrity endorsed ads help me keep up to date about 1 2 3 4 5 product/services available in the marketplace. For me, When a male celebrity endorses computers, I consider 1 2 3 4 5 him as more dependable than a female celebrity When a male celebrity endorses computers, I see him 1 2 3 4 5 to be more honest than a female celebrity When a female celebrity endorses food, I perceive her 1 2 3 4 5 to be more sincere than a male celebrity endorser, When a male celebrity endorses computers, I consider 1 2 3 4 5 him as more reliable than a female celebrity I see male celebrities endorsing computers to be more 1 2 3 4 5 trustworthy than female celebrities. 100 University of Ghana http://ugspace.ug.edu.gh When it comes to endorsement food, I perceive female 1 2 3 4 5 celebrity endorsers as more experts than male celebrity endorsers. I consider female celebrity endorsers as more 1 2 3 4 5 experienced than male celebrity endorsers especially when it comes to endorsement of food With regard to endorsement of computers, I see male 1 2 3 4 5 celebrity endorsers as more knowledgeable than female celebrity endorsers When it comes to endorsement of food, I perceive 1 2 3 4 5 female celebrity endorsers to be more qualified than male celebrity endorsers When male celebrities endorse computers, I see them 1 2 3 4 5 to be more skilled than female celebrities. 101 University of Ghana http://ugspace.ug.edu.gh PART 2 Endorser and Product Fit Kindly indicate your opinion of the match between the endorser and the product. There is fit between the endorser and product 1. Strongly disagree 2. Disagree 3. Neutral 4. Agree 5. Strongly Agree 102 University of Ghana http://ugspace.ug.edu.gh SECTION B: Perceptions about current advert Please indicate the extent to which you believe the SD D N A SA claims in this ad. Your decision must be based exclusively on the gender of the endorser. I believe this advert is informative 1 2 3 4 5 I believe this advert is truthful 1 2 3 4 5 This advert is a reliable source of information about the 1 2 3 4 5 quality and performance of the product I believe this advert presents a true picture of the 1 2 3 4 5 product I believe I have been adequately informed after seeing 1 2 3 4 5 this advert I believe this advert is truth well told 1 2 3 4 5 103 University of Ghana http://ugspace.ug.edu.gh SECTION A: Endorser and Product Fit Kindly indicate your opinion of the match between the endorser and the product. There is fit between the endorser and product 1. Strongly disagree 2. Disagree 3. Neutral 4. Agree 5. Strongly Agree 104 University of Ghana http://ugspace.ug.edu.gh SECTION B: Perceptions about current advert Please indicate the extent to which you believe the SD D N A SA claims in this ad. Your decision must be based exclusively on the gender of the endorser. I believe this advert is informative 1 2 3 4 5 I believe this advert is truthful 1 2 3 4 5 This advert is a reliable source of information about the 1 2 3 4 5 quality and performance of the product I believe this advert presents a true picture of the 1 2 3 4 5 product I believe I have been adequately informed after seeing 1 2 3 4 5 this advert I believe this advert is truth well told 1 2 3 4 5 105 University of Ghana http://ugspace.ug.edu.gh SECTION A: Endorser and Product Fit Kindly indicate your opinion of the match between the endorser and the product. There is fit between the endorser and product 1. Strongly disagree 2. Disagree 3. Neutral 4. Agree 5. Strongly Agree 106 University of Ghana http://ugspace.ug.edu.gh SECTION B: Perceptions about current advert Please indicate the extent to which you believe the SD D N A SA claims in this ad. Your decision must be based exclusively on the gender of the endorser. I believe this advert is informative 1 2 3 4 5 I believe this advert is truthful 1 2 3 4 5 This advert is a reliable source of information about the 1 2 3 4 5 quality and performance of the product I believe this advert presents a true picture of the 1 2 3 4 5 product I believe I have been adequately informed after seeing 1 2 3 4 5 this advert I believe this advert is truth well told 1 2 3 4 5 107 University of Ghana http://ugspace.ug.edu.gh SECTION A: Endorser and Product Fit Kindly indicate your opinion of the match between the endorser and the product. There is fit between the endorser and product 1. Strongly disagree 2. Disagree 3. Neutral 4. Agree 5. Strongly Agree 108 University of Ghana http://ugspace.ug.edu.gh SECTION B: Perceptions about current advert Please indicate the extent to which you believe the SD D N A SA claims in this ad. Your decision must be based exclusively on the gender of the endorser. I believe this advert is informative 1 2 3 4 5 I believe this advert is truthful 1 2 3 4 5 This advert is a reliable source of information about the 1 2 3 4 5 quality and performance of the product I believe this advert presents a true picture of the 1 2 3 4 5 product I believe I have been adequately informed after seeing 1 2 3 4 5 this advert I believe this advert is truth well told 1 2 3 4 5 109 University of Ghana http://ugspace.ug.edu.gh SECTION A: Endorser and Product Fit Kindly indicate your opinion of the match between the endorser and the product. There is fit between the endorser and product 1. Strongly disagree 2. Disagree 3. Neutral 4. Agree 5. Strongly Agree 110 University of Ghana http://ugspace.ug.edu.gh SECTION B: Perceptions about current advert Please indicate the extent to which you believe the SD D N A SA claims in this ad. Your decision must be based exclusively on the gender of the endorser. I believe this advert is informative 1 2 3 4 5 I believe this advert is truthful 1 2 3 4 5 This advert is a reliable source of information about the 1 2 3 4 5 quality and performance of the product I believe this advert presents a true picture of the 1 2 3 4 5 product I believe I have been adequately informed after seeing 1 2 3 4 5 this advert I believe this advert is truth well told 1 2 3 4 5 111 University of Ghana http://ugspace.ug.edu.gh Endorser and Product Fit Kindly indicate your opinion of the match between the endorser and the product. There is fit between the endorser and product 1. Strongly disagree 2. Disagree 3. Neutral 4. Agree 5. Strongly Agree 112 University of Ghana http://ugspace.ug.edu.gh SECTION B: Perceptions about current advert Please indicate the extent to which you believe the SD D N A SA claims in this ad. Your decision must be based exclusively on the gender of the endorser. I believe this advert is informative 1 2 3 4 5 I believe this advert is truthful 1 2 3 4 5 This advert is a reliable source of information about the 1 2 3 4 5 quality and performance of the product I believe this advert presents a true picture of the 1 2 3 4 5 product I believe I have been adequately informed after seeing 1 2 3 4 5 this advert I believe this advert is truth well told 1 2 3 4 5 113 University of Ghana http://ugspace.ug.edu.gh SECTION C: Attitude towards celebrity endorser Kindly describe your overall attitude towards Nana SD D N A SA Ama Mcbrown I love Nana Ama Mcbrown 1 2 3 4 5 Kindly describe your overall attitude towards John SD D N A SA Dumelo I love Master Richard 1 2 3 4 5 SECTION E: Perceptions of product gender Kindly indicate the gender of the following products Sanitary pad a. Feminine b. Masculine c. Neutral Shaving cream a. Feminine b. Masculine c. Neutral Bottled water a. Feminine b. Masculine c. Neutral 114 University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA ETHICS COMMITTEE FOR THE HUMANITIES (ECH) P. O. Box LG 74, Legon, Accra, Ghana My Ref. No…ECH 103 /19-20 … May 28th, 2020 Ms. Alice Mensah Department of Communication Studies University of Ghana Legon ETHICAL CLEARANCE (ECH 103/19-20) The protocol tittle below has been reviewed and approved by the ECH Committee. TITLE OF PROTOCOL: DOES GENDER IN CELEBRITY ENDORSED ADVERTISEMENT INFLUENCE AUDIENCE PERCEPTION? A STUDY OF UNIVERSITY OF GHANA STUDENTS PRINCIPAL INVESTIGATOR: MS. ALICE MENSAH Please note that the final review report must be submitted to the Committee at the completion of the study. Your research records may be audited at any time during or after the implementation. Any modification of this research project must be submitted to ECH for review and approval prior to implementation. Please report all serious adverse events related to this study to ECH within seven (7) days verbally and in writing within fourteen (14) days. This certificate is valid till May 27th, 2021. You are to submit annual reports for continuing review. Please accept my congratulations. Yours Sincerely, Professor C. Charles Mate-Kole ECH Chair Cc: Dr. Sarah Akrofi-Quarcoe Department of Communication Studies, UG Dr. Abena Animwaa Yeboah-Banin, Department of Communication Studies, UG Tel: +233-303933866 Email: ech@ug.edu.gh