i SOME APPROACHES TO MODELLING NEED-BASED FINANACIAL AID TO NEEDY STUDENTS IN THE UNIVERSITY OF GHANA By Patience Mamle Agbedor (10363038) This thesis is submitted to the University of Ghana, Legon in partial fulfillment of the requirement for the award of MPhil Statistics Degree June, 2013 University of Ghana http://ugspace.ug.edu.gh ii DECLARATION I hereby declare that this submission is my own work towards the University of Ghana Masters of Philosophy in Statistics degree and that, to the best of my knowledge, it contains no material previously published by another person nor material which has been accepted for the award of any other degree of the University, except where due acknowledgement has been made in the text. Patience Mamle Agbedor (10363038) -------------------------------- Signature ----------------------------- Date Certified By: Dr. E. N. N. Nortey (Supervisor) -------------------------------- Signature -------------------------------- Date University of Ghana http://ugspace.ug.edu.gh iii ABSTRACT It was asserted that University and university systems around the world are faced with rapid growing demand and decreasing or static government investment (Marcucci and Johnstone, 2010). In response to this assertion, many countries introduced cost-sharing in order to preserve the quality of higher education. In order not to deny academically talented young people from poor families from accessing higher education, governments and individual institutions started offering financial assistance to needy students. For this same cause the University of Ghana established a Student Financial Aid Office (SFAO) in 2005 which aims at awarding scholarships to needy but brilliant students. Little is known about how the SFAO awards its scholarship. Many countries have adopted the means testing method to enable them target the scarce funds to only needy students. Hence the aim of this study is to develop a statistical model (a means testing statistical model) for assessing the need of a student who applies for financial aid and awarding the scholarship accordingly. A random sample of 384 undergraduate regular University of Ghana students was selected to fill a questionnaire on a wide range of questions. Factor analysis was used to extract critical factors which were used to assess the need levels of the students and responses of each respondent were scored based on weights assigned to the variables. The scores were then used to compute the Relative Need Index for every respondent; furthermore, students were categorized into five need groups according to their need levels. It was found out that only 2.7% of students sampled fell in the most needy group and 15.6% were in the least needy group. On the other hand, majority of the students were in the middle level class which are the needy and the less needy groups, constituting 23.1% and 47.3% of the total sample respectively. It was concluded that information on income is difficult to come by in our part of the world, therefore the family income component was not included in the analysis. It was also established that, even though means testing has its challenges it is adopted by countries and institutions in order to allot financial assistance to students efficiently. The developed means testing formula was recommended to the University of Ghana for adoption. University of Ghana http://ugspace.ug.edu.gh iv ACKNOWLEDGEMENT I am happy to express my deepest appreciation to the Almighty God for His divinely guiding, guarding me, granting me the necessary wisdom and strength throughout this course, and most of all for bringing me to this successful end. My deepest and profound gratitude goes to my supervisor, Dr. E. N. N. Nortey for his, encouragement and immeasurable support, his significant suggestions and inputs without which I would not have had the urge to complete this work. I say God bless you for urging me on to the end. Also my gratitude goes to my second supervisor, Dr. F. O. Mettle for his advice and constructive criticisms, and also to Dr. I. K. Baidoo for making time to study my questionnaire and giving me the necessary advice. To my colleagues, Webster Ahli and Benedict Mbea-Basiden, I say God bless you for providing me with necessary research materials and scholarly arguments which went a long way to positively impact my work. And Phidelia Smith for her encouragement, I say thank you all for being a blessing to me. I say thank you very much and to anyone who assisted in diverse ways to make this work a success. May God bless you all. University of Ghana http://ugspace.ug.edu.gh v DEDICATION I dedicate this work to the Almighty God for his great love, protection, and guidance during the years of this study. Even though it was stressful and hectic, I am grateful to God for seeing me through. This work is also dedicated to my dear parents Jonathan Korli and Lucy Otoo for their support and encouragement; this programme will not have been successful without them. To my siblings Eunice Nyema, Esther Addo, Mary Doe and Gideon Korli for their prayer and moral supports. Finally, I owe my most cherished husband Michael Agbedor a million thanksgiving for his encouragement, support, enormous sacrifices and love throughout my study, not withstanding, my children Elikem, Emefa and Eyram Agbedor for their numerous sacrifices and understanding. I say thank you all for being a blessing to me and may the Almighty God show you kindness in your lives. University of Ghana http://ugspace.ug.edu.gh vi TABLE OF CONTENTS CONTENT PAGE Title Page ..…………………………………………………………………… i Declaration …………………………………………………………………… ii Abstract ……………………………………………………………………….. iii Acknowledgement ……………………………………………………………. iv Dedication ……………………………………………………………………….. v Table of Content ………………………………………………………………. vi List of Tables ………………………………………………………………….. xi List of Figures …………………………………………………………………. xiii List of Abbreviations/Acronyms ……………………………………………… xiii Definition of Terms …………………………………………………………… xiii Class Range of the University of Ghana……………………………………….. xiv CHAPTER ONE INTRODUCTION 1.1 Background of the Study …………………………………………………. 1 1.2 Problem Statement ………………………………………………………... 4 1.3 Objectives of the Study …………………………………………………… 4 1.4 Significance of the Study …………………………………………………. 5 1.5 Limitations of the Study …………………………………………………… 6 1.6 Organisation of the Study ………………………………………………….. 7 CHAPTER TWO REVIEW OF RELATED LITERATURE 2.0 Introduction ………………………………………………………………… 8 University of Ghana http://ugspace.ug.edu.gh vii 2.1 Higher Education Financing………………………………………………... 8 2.2 Cost Sharing in Higher Education ………………………………………….. 10 2.3 Financial Aid ………………………………………………………………... 11 2.4 Merit-Based versus Need-Based Financial Aid …………………………...... 13 2.5 Means Testing ……………………………………………………………….. 15 2.5.1 Definition of Means Testing ………………………………………………... 17 2.5.2 Means Testing Formulae used by Some Countries …………………………. 18 2.5.2.1 Means Testing Procedure of Australia ……………………..……………... 18 2.5.2.2 Means Testing Procedure of Chile ……………………………………… 19 2.5.2.3 Means Testing Procedure of Colombia …………………………………. 20 2.5.2.4 Means Testing Procedure of Costa Rica …………………………........... 21 2.5.2.5 Means Testing of Procedure of Kenya ………………………………….. 21 2.5.2.6 Means Testing Procedure of South Africa ……………………………… 23 2.5.2.7 Means Testing Procedure of the United States ………………………… 24 2.5.3 Factors Considered by some African Countries in Conducting Meansgf Testing............................................................................................................ 28 2.5.4 Justification for Means Testing …………………………………………. 29 2.5.5 Conducting of Means Testing …………………………………………… 30 2.5.6 Limitations of Means Testing …………………………………………… 33 CHAPTER THREE METHODOLOGY 3.1 Introduction ………………………………………………………………. 34 3.2 Research Design ………………………………………………………….. 34 3.3 Study Population……………………………………………………........... 35 University of Ghana http://ugspace.ug.edu.gh viii 3.4 Source of Data ……………………………………………………………. 37 3.5 Sampling Procedure ………………………………………………………. 37 3.5.1 Sample Size Estimation …………………………………………………. 37 3.5.2 Sampling Design ………………………………………………………… 40 3.6 Data Collection ……………………………………………………………. 40 3.7 Method of Data Analysis ………………………………………………….. 41 3.7.1 Preliminary Analysis …………………………………………………….. 41 3.7.1.1 Data Cleaning ………………………………………………………….. 42 3.7.1.2 Treatment of Missing Values and Outliers …………………………….. 42 3.7.1.3 Description of Data ……………………………………………………. 43 3.7.1.4 Testing of Assumptions ……………………………………………….. 44 3.7.2 Factor Analysis ………………………………………………………….. 45 3.7.2.1 The Orthogonal Factor Model ………………………………………… 48 3.7.2.2 Method of Estimating Factor Loadings ……………………………….. 49 3.7.2.3 The Principal Component Method …………………………………….. 51 3.7.2.4 Factor Rotation ………………………………………………………… 52 3.7.2.5 Computation of Factor Scores …………………………………………. 53 3.7.2.6 Computation of Relative Need Index (RNI) …………………………… 55 3.8 Model Validation …………………………………………………………... 56 3.9 Ethical Issues ………………………………………………………………. 57 3.10 Challenges ………………………………………………………………… 57 CHAPTER FOUR RESULT OF DATA ANALYSIS 4.1 Introduction …………………………………………………………………. 58 University of Ghana http://ugspace.ug.edu.gh ix 4.2 Presentation of Preliminary Analysis of the Data………………………......... 58 4.2.1 Demographic Description of the Sample…………………………………… 58 4.2.1.1 Gender of Respondent ……………………………………………………. 59 4.2.1.2 Age Distribution of Respondents …………………………………………. 59 4.2.1.3 Respondent’s Region Hailed from and Permanent Region of Residence … 61 4.2.1.4 Localities Respondents Hailed from and Permanently Resided ………….. 62 4.2.2 Respondent’s Student Status …………………..……………………………. 63 4.2.2.1 Respondent’s Level of Study …………………………………………….. 63 4.2.2.2 Course of Study and Performance of Respondent ……………………….. 64 4.2.2.3 Residential Status of Respondent ………………………………………… 66 4.2.2.4 Type of Accommodation of Non-Residence Respondent ………………... 66 4.2.3Information on Respondent’s Educational Background …………………… 67 4.2.3.1 Respondent’s Highest Education Qualification Attained ……………….. 67 4.2.3.2 Type of SHS and JHS …….…………………………………………….. 68 4.2.3.3 Locality of SHS and JHS Attended …………………………………….. 68 4.2.3.4 Sponsor of Respondents’ Primary/JHS Education ……………………… 69 4.2.3.5 Sponsor of Respondents’ Secondary Education ………………………… 70 4.2.3.6 Main Source of Funding for this Academic Year ……………………….. 70 4.2.3.7 All Other Sources of Funding for This Academic Year ………………… 71 4.2.4 Respondent’s Parent’s Status ……………………………………………... 72 4.2.4.1 Respondent’s Parents Alive …………………………………………….. 72 4.2.4.2 Employment Status of Parent …………………………………………… 73 4.2.4.3 Occupation of Father ……………………………………………………. 74 4.2.4.4 Occupation of Mother …………………………………………………… 74 4.2.5 Respondent’s Socio-economic Status …………………………………….. 75 University of Ghana http://ugspace.ug.edu.gh x 4.2.5.1 Marital Status of Respondent …………………………………………… 75 4.2.5.2 Number of Children of Respondents ……………………………………. 76 4.2.5.3 Employment Status of Respondent ……………………………………… 76 4.2.5.4 Material of the Roof of Dwelling ………………………………………. 77 4.2.5.5 Material of Walls of Dwelling …………………………………………. 78 4.2.5.6 Source of Drinking Water ……………………………………………… 79 4.2.5.7 Toilet Facility used by household ……………………………………… 80 4.2.5.8 Main Fuel used for Cooking …………………………………………… 81 4.2.5.9 Main Fuel used for Lighting …………………………………………… 82 4.2.5.10 Appliances Owned by Respondents’ Household …………………….. 83 4.2.6 How Respondent’s Medical Expenses are Borne ………………………… 84 4.2.7 Total Cost of Attending the University per Semester ………………….... 85 4.3 Results of Main Analysis …………………………………………………… 86 4.3.1 Factor Analysis Results ……………………………………………………. 86 4.3.1.1 KMO and Bartlett’s Test…………………………………………………. 86 4.3.1.2 Total Variance Explained and Number of Factors Extracted ……………. 87 4.3.2 Relative Need Index Statistics .……………………………………………. 89 4.4 Discussion of the Results ……………………………………………………. 91 CHAPTER FIVE SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 5.1 Summary ……………………………………………………………………... 94 5.2 Conclusions …………………………………………………………………… 96 5.3 Recommendations …………………………………………………………….. 97 References …………………………………………………………………………. 99 University of Ghana http://ugspace.ug.edu.gh xi Appendix A: Results from Field Data ……………………………………………. 102 Appendix B: Questionnaire ………………………………………………………. 112 Appendix C: Factor Loading Matrix……………………………………………… 118 LIST OF TABLES TABLE TITLE PAGE 4.2.1.1 Gender of Respondents 59 4.2.1.2 Age Distribution of Respondents 60 4.2.1.3 Region Respondents Hailed from and Permanent Region of Residence 61 4.2.1.4 Respondents’ Permanent Locality of Residence and Locality Hailed from 63 4.2.2.1 Level of Study 64 4.2.2.2 Course of Study by Performance Class of Respondents 65 4.2.2.3 Residential Status 66 4.2.2.4 Type of Accommodation of Non-Resident Respondents 66 4.2.3.1 Respondent’s Highest Education Qualification Attained 67 4.2.3.2 Type of SHS and JHS 68 4.2.3.3 Locality of SHS and JHS Attended 69 4.2.3.4 Sponsor of Respondents’ Primary/JHS Education 69 4.2.3.5 Sponsor of Respondents’ Secondary Education 70 4.2.3.6 Main Source of Funding for this Academic Year 71 4.2.3.7 All other Sources of Funding this Academic Year 72 University of Ghana http://ugspace.ug.edu.gh xii 4.2.4.1 Respondents Parents Alive 73 4.2.4.2 Employment Status of Parent 73 4.2.4.3 Occupation of Father 74 4.2.4.4 Occupation of Mother 75 4.2.5.1 Marital Status of Respondent 76 4.2.5.2 Number of Children of Respondent 76 4.2.5.3 Employment Status of Respondent 77 4.2.5.4 Material of roof of Dwelling 78 4.2.5.5 Material of the Walls of Dwelling 79 4.2.5.6 Source of Drinking Water 80 4.2.5.7 Toilet Facility used by the Household 81 4.2.5.8 Main Fuel used for Cooking 82 4.2.5.9 Main Fuel used for Lighting 83 4.2.5.10 Appliances Owned by Respondent’s Household 84 4.2.6.1 How Respondents Own Medical Expenses are Borne 85 4.2.7 Total Cost of Attending the University per Semester 85 4.3.1.1 KMO and Bartlett’s Test 86 4.3.1.2 Total Variance Explained 87 4.3.2.1 Relative Need Index 90 University of Ghana http://ugspace.ug.edu.gh xiii LIST OF FIGURES FIGURE TITLE PAGE Figure 1 Scree Plot for Extracted Factors 88 Figure 2 Relative Need Index 90 Figure 3 Relative Need Groups 90 LIST OF ABBREVIATIONS AND ACRONYMS SFAO ----------------- Students Financial Aid Office SISBEN --------------- Sistema de Seleccion de Beneficiarios para Programas Sociales NSFAS --------------- National Student Financial Aid Scheme FAFSA --------------- Free Application for Federal Student Aid EFC ------------------ Expected Family Contribution fpc ---------------------- Finite Population Correction CGPA------------------ Cumulative Grade Point Average ICETEX---Instituto Colombiano de Crédito Educativo y Estudios Técnicos en el Exterior ACCESS ---------------- Acceso Con Calidad a la Educacion Superior DEFINITION OF TERMS Communality – Total amount of variance an original variable shares with all other variables included in the analysis. Correlation Matrix – Table showing the intercorrelations among all variables. Factor Loading – Correlation between the original variables and the factors. Factor Matrix – Table displaying the factor loading of all variables on each factor. Factor Rotation – Process of manipulation or adjusting the factor axes to achieve a simpler and pragmatically more meaningful factor solution. University of Ghana http://ugspace.ug.edu.gh xiv Factor Scores – They are Composite score estimated for each respondent on the derived factors by factor analysis (Naresh, 2004). Multicollinearity – Extent to which a variable can be explained by the other variables in the analysis. Orthogonal – Mathematical independence (no correlation) of factor axes to each other. Varimax – The most popular factor rotation methods focusing on simplifying the columns in a factor matrix. SISBEN is a composite welfare index used as a targeting system for social programs in Colombia. It is a function of a set of variables that can be grouped into four categories:(1) housing and home appliances; (2) public utility services; (3) human capital endowment and economic risk (schooling of oldest wage earner, mean schooling for family members 12 years and older); and (4) family demographics, unemployment, dependency ratio and per capita income (Marcucci and Johnstone, 2010). fpc – the finite population (fpc) factor is used to adjust a variance estimate for an estimated mean or total, that this variance only applies to the portion of the population that is not in the sample. Class Ranges of the University of Ghana CGPA Class Division 3.60 – 4.00 First Class Division 3.25 – 3.59 Second Class Upper Division 2.50 – 3.24 Second Class Lower Division 2.00 – 2.49 Third Class Division University of Ghana http://ugspace.ug.edu.gh xv 1.5 – 1.99 Pass Below 1.5 Fail University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE INTRODUCTION 1.1 Background Marcucci and Johnstone (2010) asserted that universities and university systems around the world are faced with rapidly growing demand and decreasing or static government investment. Marcucci and Johnstone again acknowledged that, many countries have introduced tuition fees and other elements of cost sharing in order to preserve or even expand capacity and protect quality in higher education as government investment curtails. At the same time, universities are also working to enhance higher education’s accessibility to academically talented young men and women from poor and rural families who are not in a position to cover significant tuition and other fees by offering them financial assistance (Marcucci and Johnstone, 2010). The University of Ghana is no exception to this assertion. In an attempt to alleviate or minimize the problem of denying talented but needy students the opportunity to attain higher academic credentials, the University plausibly established a Students Financial Aid Office (SFAO) in the year 2005. The mission of the SFAO is as follows: The SFAO supports the mission of the University to develop world-class human resources with capabilities to meet national development needs. University of Ghana http://ugspace.ug.edu.gh 2 It also aims at significantly reducing or eliminating financial barriers that might prohibit or inhibit students’ access to education at the University of Ghana. They provide financial assistance to students who, without such assistance, may probably not be able to readily access or meet educational and other expenditure at the University (SFAO flyer). Since the establishment of the SFAO in 2005 to 2012, a total of 1,210 students out of the 2,012 who applied were awarded the scholarship, that is, approximately 60% of the applicants were successful (SFAO flyer). From the number of beneficiaries 1,074 were male and 136 female out of the 1,800 male and 212 female applicants respectively (SFAO flyer). Looking at this all important initiative by the University of Ghana, the researcher deemed it appropriate to develop a statistical model. This model could serve as a tool for the SFAO to scientifically select needy applicants to be awarded the scholarship by using the means testing methodology. This is in order to enhance the accuracy and efficiency of the existing selection criteria, amount given to each student based on level of need and also to eliminate any possible human bias during selection. According to Tekleselassie and Johnstone (2004), the sharp increases in tuition fees and other parent- or student – borne costs must be met with some form of targeted subsidies in the form of means tested grants and/or loans if cost-sharing is not to preclude the University of Ghana http://ugspace.ug.edu.gh 3 possibility of higher education for the majority of families with low incomes. Tekleselassie and Johnstone again drew attention to the fact that one of the very great dilemmas for higher educational policy in Africa and virtually all developing countries is means testing – determining and verifying the amount that a family can reasonably be expected to contribute towards its children’s higher education. Marcucci and Johnstone (2010) hold the view that means testing is a form of subsidy targeting which attempts to distribute at least some of the higher educational subsidies, such as low or no tuition fees, grants or subsidized loans, and/or access to lodging, on the basis of the student‘s or his/her family’s need, or its estimated ability to pay for some of the underlying costs of education. Marcucci and Johnstone again argued that the success of student assistance policies in meeting their objectives in a financially sustainable manner ultimately rests on fair and accurate means testing that ensures financial assistance to eligible students and families and avoids or minimizes awards to non-poor students. Ngolovoi (2008) emphasized that means testing is specifically used to winnow out students from wealthy backgrounds who can access and participate in higher education, leaving only the needy to benefit from the Financial Aid. Malik and Chanthy’s study (as stated in Qingyue, Beibei Y. and Liying J, 2010) revealed that, means testing was thought of being the most effective and expensive targeting method compared with other targeting methods. University of Ghana http://ugspace.ug.edu.gh 4 1.2 Problem Statement Although the University of Ghana has an SFAO which seeks to award scholarships to needy students, there has been no study to ascertain the effectiveness of the financial aid they offer to the students or applicants. Since the globally accepted method (means testing) of targeting needy but academically able students is not employed, there is a high tendency of awarding scholarships to students from rather affluent backgrounds. If this happens then the aspirations of the University of Ghana to significantly reduce or eliminate financial barriers that might prohibit or inhibit students’ access to the University might not be realized. According to Ngolovoi (2008), in developing countries, where higher education is heavily subsidized, it is necessary to employ means testing. It is therefore prudent to institute a means testing methodology in allocating the scarce resource of the University to needy but brilliant applicants. Hence the study. 1.3 Objectives of the Study The main objective is to develop a statistical model that would be used for targeting needy students for financial assistance and also to determine their level of need. The specific objectives are to i. identify the factors that would be more relevant to the development of the model; University of Ghana http://ugspace.ug.edu.gh 5 ii. measure several socio-economic variables to be used to develop a model that discriminates needy students from non-needy students; iii. group needy students by the level of their needs. 1.4 Significance of the study This study stipulates suggestions which are useful in augmenting the scholarship programme and very important issues of equity, and it also advocates increased targeting of students from poor families for Student Financial Aid. This research is employing means testing methodology to make targeting of needy students more effective. The study is also to help ensure that the scarce funds available is reaching the target population (comprised of needy but academically able students) and to increase equity and access by providing funds to only needy students (Ngolovoi, 2008). If the scholarship reaches the target population, it would first of all go a long way to benefit the individual by way of getting better jobs as well as non-monetary benefits such as the prestige associated with credentials and enhance their lives and live in good health. Secondly, the University of Ghana would benefit by achieving the mission of providing world-class human resource to the workforce. The beneficiaries might also in future contribute to the development of the University which can elevate it especially in the world ranking of universities. Finally, if the financial aid is well targeted, it will contribute in breaking the cycle of poverty for those at the bottom of the socio-economic structure and also give the opportunity to such people to return to society in the form of a University of Ghana http://ugspace.ug.edu.gh 6 more talented and productive workforce (Marcucci and Johnstone, 2010). They are more likely to contribute to tax revenue and to the nation’s economic and cultural productivity. This study is sorting to develop an improved formula or method for determining the financial need of students. Heller (2004) made a very challenging statement that in the era in which tuition fees are rising much faster than the ability of families to pay for university education – particularly lower-income families; public and institutional financial aid need to be used in the most effective and efficient manner as possible. Heller again argued that in the economic world of highly constrained social welfare maximization, giving scarce financial aid resources to people who don’t need them is wasteful, unnecessary, unproductive, and comes as the price of adequate and appropriate student financial aid for others who could not afford to complete without assistance. 1.5 Limitations of the Study 1. The study involved only regular undergraduate students in levels 200 to 400 of the University of Ghana. 2. This model can be used only in the University of Ghana because of the method used. It cannot be applied to data from a different institution. University of Ghana http://ugspace.ug.edu.gh 7 1.6 Organisation of the Study The study comprises of five chapters. This introductory chapter is the first and it presents the background of the study, problem statement, objectives and relevance of the study. It is followed by chapter two in which literature related to the study is reviewed – issues reviewed include cost sharing in higher education, the concept and financial aid policies and formulae in colleges, juxtaposing merit-based with need-based financial aid, the means testing approach in awarding financial assistance to needy students and finally a review of the means testing formulae used by some countries. Chapter three discusses the methodology of the study. The results of the data analysis are presented and discussed in chapter four and finally, the summary, conclusions and recommendations of the study are reported in chapter five. University of Ghana http://ugspace.ug.edu.gh 8 CHAPTER TWO REVIEW OF RELATED LITERATURE 2.0 Introduction This chapter sets out a frame work that forms conceptual foundation for the research. It is in six divisions, comprising higher education funding, cost sharing in higher education, financial aid, means testing, discussion of the two types of financial aid (need-based and merit-based) and means testing formulae adopted by some countries. 2.1 Higher Education Financing According to Johnstone (2003), higher education at the beginning of the 21st century has never been in greater demand, both from individual students and their families. For the occupational and social status and the greater earnings it is presumed to convey, as well as from governments for the public benefits it is presumed to bring to the social, cultural, political and economic well-being of countries, there has been an increasing demand for the past few decades (Johnstone, 2003). The World Bank (2010) concludes that in most Sub-Saharan African countries, enrolment in higher education has grown faster than financing capabilities and that public funding in most countries are already overstretched. According to the report this problem of lack of resources has resulted in severe decline in the quality of instruction and that it will not be sufficient to respond to the growing demand for access to higher education while University of Ghana http://ugspace.ug.edu.gh 9 delivering a level of quality that provides students with the skill necessary to succeed in current and future markets. The report made mention of some easy ways out that some countries are implementing – private higher education which is experiencing a spectacular growth in Africa and Cost-Sharing programmes which are being implemented in many universities accompanied by students loans and financial aid for low – income students. Higher education plays a key role in training qualified individuals to establish more enterprises and institutions and thus allocate resources more efficiently and through research and increased knowledge in higher education can also help to address the challenges arising from population growth, endemic diseases, urbanization, energy costs and climate change (World Bank, 2010). The report further challenged Sub-Saharan African countries that in order for them to reap the benefit of this investment in human capital, higher education institutions must have financing to provide quality training and sound professional prospects to their students. The World Bank challenging African countries to provide quality training to students supported their observation of increased demand for higher education with some figures. The reports claims that over the past 15 years, the total number of students pursuing higher education tripled, climbing from 2.7 million in 1991 to 9.3 million in 2006 (an annual average rate of 16 percent), while public resources allocated to current expenditure only doubled (an annual average rate of 6 percent). University of Ghana http://ugspace.ug.edu.gh 10 The Ghana government’s funding of tertiary education started in 1948 when the University of Gold Coast, now University of Ghana, was established to produce the manpower requirements of the country (www.ghanaweb.com/GhanaHome page/feature/article.php). During that era, University students were treated as first-born babies and were provided with everything, including pocket money by government just to ensure that the needed psychological and physiological comfort was obtained for smooth scholarly work (www.ghanaweb.com/GhanaHome page/feature/article.php). Due to the higher demand for higher education, the government could not adequately fund these institutions leading to lack of basic services such as professors, laboratories, equipments, housing, and other facilities so needed (Amenya, 2009). Amenya alluded that as a result of the lack of basic facilities, many people were denied access and that the model of basically free for all which in principle did not discriminate against anyone with the basic standards for entry needed to be reviewed. Amenya (2009) argued that higher education financing has always been a thorny issue for parents, policy makers and other stakeholders in the arena of higher education. 2.2 Cost Sharing in Higher Education According to Johnstone (2003), “cost – sharing is generally thought of as the introduction of or especially sharp increase in, tuition fees to cover part of the costs of instruction, or of user charges to cover more of the costs of lodging, food and other expenses of student living that may have hitherto been borne substantially by governments (taxpayers) or institutions”. University of Ghana http://ugspace.ug.edu.gh 11 Cost sharing was introduced in Ghana in the 1997 through the adoption of the ‘Akosombo Accord’ that divided responsibility for university funding between the government (responsible for 70 percent of total funding) and three other sources (30 percent) including university internal revenue-generation, private donations and student tuition fees. Student academic and residential facility user fees were introduced in 1998 (www.ghanaweb.com/GhanaHome page/feature/article.php). At this juncture students from very low income families would have difficulties in pursuing higher education; this gave rise to institutions introducing Financial Aid to assist needy students pursue and complete their higher education. 2.3 Financial Aid Carey (2007) drew attention to the fact that every year, university education gets more expensive. On the question of student financial aid, this study supports the position of Carey who maintains that while policy makers focus on student loans, another important form of student financial assistance has received less scrutiny and that is aid provided directly by individual universities. Fenske, Porter and DuBrok (2000) made an assertion about financial aid policy and programmes that they are the primary vehicles to ensure economic status is not a barrier, hence, the importance of understanding the nature of financial aid received by students from low-income families as it relates to their persistence and degree completion. It was argued in the World Bank (2010) report that scholarships and other forms of student financial aid need to be better targeted and rationalized to better meet the goals of University of Ghana http://ugspace.ug.edu.gh 12 equity and efficiency. The report also claims that in many countries, grant and scholarship allocation criteria are linked to academic performance rather than to socioeconomic disadvantages, or priority disciplines for the country’s development, which this study is not in full support of. Financial aid is very crucial in higher education since it will go a long way to benefit both the individual and the entire nation. Fenske et al (2000) argued that if we educate tomorrow’s workforce, then we are also meeting technical labour force of the nation. Several authors have investigated the impact of financial aid on university persistence and graduation but their results were inconclusive and range from positive to negative and to no effect altogether (Alon, 2005). The results of Alon’s study actually argues that interrelationship between aid eligibility and graduation mask the positive impact of financial aid on graduation. Going on further with the findings made, Alon contended that financial aid eligibility (except for merit-based aid) exerts a negative effect on persistence while increase in the dollar amount is positively related to successful completion of university education. Even though this subject is not most germane to this study, the researcher thinks its findings are worth noting. Since in the University of Ghana it is only a very small proportion of the student population that apply for financial aid, many may think it is not worth researching into, but this study is in agreement with the findings of Doyle (2010), and wrote “some authors describe the history and background of each type of aid, and concluded: students University of Ghana http://ugspace.ug.edu.gh 13 financial aid programmes play a major role in who participates in post secondary education”. 2.4 Merit-Based versus Need-Based Financial Aid Several authors have written about the advantages and disadvantages of need-based and academic merit-based financial aid. This section juxtaposed what researchers have said about the two types of financial aid and to find out which of these two methods is more beneficial than the other. Toutkoushin and Shafiq (2009) put across that access to higher education for low-income populations may have been hindered by the trend of shifting financial support away from need-based aid towards academic merit-based financial aid because low-income populations are less likely to qualify for merit-based aid. Their findings show that states are better off if they awarded financial aid on the basis of need rather than merit. They also claim that if it is true that students who would be eligible for merit-based aid are more likely than students who are normally eligible for need-based aid to go through college, then need-based aid leads to larger gains. In lieu of maintaining their findings, Toutkoushin and Shafiq claimed that it is not to say, however, that universities should not award merit-based aid to students, and made mention of an equally compelling argument which could be made that universities award merit-based aid and not need-based aid to students based on the assumption that the goal of universities is to maximize their prestige or reputation. University of Ghana http://ugspace.ug.edu.gh 14 Heller (2004), interestingly put across that both states and institutions have used measures of academic merit in place of financial need as the basis for awarding grants and scholarship and also observed that both are abandoning the concept of “exceptional financial need” as the factor determining who should receive aid. Heller holds the view that the difference between the cost of a student attending the university (including tuition, hall/hostel, books transportation and other expenses) and the contribution of the family must be the criteria to determine the amount of financial aid for which the student would qualify, and also thinks that this can be achieved by employing a complex formula as “needs analysis” which takes into account family income, assets and other characteristics to determine the amount that a student and family could afford to contribute to university education. Heller reiterated that merit-based scholarships go disproportionately to students who would have gone through the university education even without the universities assistance, while need-based aid helps those, according to his research findings, require assistance to complete whatever programme they pursued. Heller argued that students least likely to be awarded merit scholarship come from families with poor economic background and populations that have traditionally been underrepresented in the university, and said that this hinders the potential to increase university graduation among low-income students. Heller found it interesting that even within the category of need-based aid, students with higher income background saw the largest growth in grant dollars, indicating that institutions probably used increasingly liberal definitions of financial need in the United States of America. It is therefore important for institutions to develop a formula or model University of Ghana http://ugspace.ug.edu.gh 15 for determining the financial need of a student. This concern expressed by Heller is the key aim of this research work, which is, developing a model to determine the financial need of a student who applies for the financial aid. The increasing use of academic merit rather than financial need, as the fundamental criterion for the awarding of financial aid has important implications on college access in the United States (Heller, 2004). According to Heller, “research on tuition prices and financial aid over the past three decades have consistently found that, short of keeping tuition prices as low as possible, financial aid targeted at needy students is the best policy for increasing college access among underrepresented students”. Even though this research focuses mainly on need-based financial aid, it considers giving academic merit some room in granting financial assistance. Financial Aid is a very crucial element in higher education and once somebody becomes a student of the University of Ghana, that person could access this financial aid. 2.5 Means Testing Tekleselassie and Johnstone (2004) in their report put across the idea that means testing is a form of subsidy targeting which attempts to distribute at least some higher education subsidies on the basis of need or estimated ability to pay. Ngolovoi (2008) thinks that means testing is specifically used to screen out students from wealthy backgrounds who can afford to access and participate in higher education, leaving only the needy to benefit from the financial aid. University of Ghana http://ugspace.ug.edu.gh 16 Ngolovoi (2008) reiterated that averagely, families in most part of Africa have extremely low income and resources available to many or most of them are insufficient to meet new expectations of paying tuition fees as well as costs of students living. Ngolovoi emphasized that sharp increases in tuition fees and other parent- or student-borne costs must be met with some form of targeted subsidies in the form of means tested grants and/ or loans if cost-sharing is not to preclude the possibility of higher education for the majority of families with low incomes. Means testing methodology is simply to test the ability of a student’s family to contribute to the financing of his/her university education and the degree to which they can contribute. There are several factors in determining the means of a family. Factor Analysis is employed to summarize the factors by grouping correlated ones. On the premise of means tested scholarships, the university aims to achieve efficiency by targeting only the very needy who really require fund to access and participate in higher education by screening out students from affluent backgrounds (Ngolovoi, 2008). Teklesalasie and Johnstone (2004) confirmed by saying that it is apparent that people from the middle and upper social strata benefit from public services more than those from the low socio-economic strata. Ngolovoi (2008) made an assertion that in developing countries, where higher education is heavily subsidized, it is necessary to employ means testing. The design of the means test is critical. Means testing systems, like all household targeting systems, need to be designed with care. Castañeda and Lindert (2005) identified University of Ghana http://ugspace.ug.edu.gh 17 numerous factors that need to be considered in the design including: a) appropriate data collection strategy; b) adequate management; c) feasibility and potential accuracy of verification mechanisms; d) institutional arrangements, and e) monitoring and oversight mechanisms to ensure transparency, credibility and control of fraud. Systems take time to design; therefore, piloting and implementation are inherently context specific. A means testing system that works well in a particular country cannot simply be replicated in another and each of these factors must be considered in light of particular country characteristics and existing infrastructure (Marcucci and Johnstone, 2010). 2.5.1 Definition of Means Testing Coady, Grosh and Hoddinott study (as cited in Ngolovoi, 2008) defined means testing as a form of individual assessment that compares resources such as income belonging to an individual or a household with some cut off. Whiles Meritosis and Wolain study (as cited in Ngolovoi) simply put it as the method of determining who is able to pay and the proportion they can pay in relation to costs of higher education. Ngolovoi (2008) holds a similar view that means testing can be defined as the process of determining whether families of loan applicants are able to contribute and the degree to which they are able to finance the education of their children. Tekleselassie and Johnstone (2004) defined means testing as a form of subsidy targeting, which attempts to distribute at least some higher education subsidies on the basis of need or estimated ability to pay. Tekleselassie and Johnstone (2004) put across that means tested subsidy is a benefit (e.g., a grant, tuition fee discount, or access to a subsidized loan) that is targeted to families or University of Ghana http://ugspace.ug.edu.gh 18 directly to students with minimal means. The system may call for benefits that rise with the diminishing calculated family means. Or from the opposite perspective but with the same meaning, the system may call for a grant that diminishes with increasing incomes or measured means. 2.5.2 Means Testing Formulae used by some Countries There are several countries in the world which have employed means testing in awarding scholarships, allowances and loans and it is evident in the discussions below that there are different ways of arriving at the means tested awards in the various countries. The means testing formulae employed by some few countries in awarding various types of financial assistance are discussed below and the source of these formulae is (Marcucci and Johnstone, 2010). 2.5.2.1 Means Testing Procedure of Australia Youth Allowance Means testing formula: The means test is composed of a parental income test (for dependent students), a personal income test (for both dependent and independent students) and an assets test. For the parental income test, the basic youth allowance is reduced by A$1.00(US$0.72) for every A$4.00 (US$2.88) that the income is over the threshold of A$32,800 (US$23,597). The personal income test allowance, an income-free area of A$236 (US$170) per fortnight for dependent students. University of Ghana http://ugspace.ug.edu.gh 19 Income that exceeds the income free area will reduce allowance payment. Under the assets test, if a dependent student’s family assets exceed A$571,500 (US$ 411,150) no allowance may be paid. The assets test does not include the family home and the value of farm or business assets are discounted by 75 percent. (Source: Marcucci and Johnstone, 2010). It could be realized from the Austrian means testing formula for the youth allowance that income is the principal source of information used in the formula. This implies that income data is reliable in Australia. On the other hand, income data in Ghana is not easily accessible, even when provided is not reliable because they are over quoted or under quoted. 2.5.2.2 Means Testing Procedure of Chile Fondos Solidarios de Credito Universitario & Credito de la Ley 20.027 para Financiamento de Estudios de Education Superior Means testing formula: For the Fondos Solidarios de Credito Universitario, students must be from among the four poorest income quintiles. The loan finances all or part of the reference tuition fee for the course of study. For the Credito de la Ley 20.027 para Financiamento de Estudios de Education Superior, the applicant indicates the size of the loan that he/she is applying for (within minimum and maximum parameters set by the scheme). The Comision Ingresa (not the individual financial institutions) ranks the qualified applicants from poorest to richest and awards loans starting with the poorest. Academic performance is used to filter applicants (weed out those who are not eligible). Then socioeconomic level is the only indicator that is used to allocate credit. (Source: Marcucci and Johnstone, 2010) University of Ghana http://ugspace.ug.edu.gh 20 2.5.2.3 Means Testing Procedure of Colombia ICETEX ACCESS grant/loan Means testing formula: For targeting purposes several criteria are considered, including: the applicant‘s academic merit (given weight of 73.2%), his/her estrato socioeconomic (given weight of 11.5%), accreditation of applicant‘s higher education institution (11.5%), and the affirmative action and retention activities of the institution (3.8%). For allocation purposes, students in SISBEN levels 1 and 2 (poorest) are eligible for ICETEX loan covering 50 percent of their tuition fees and a grant covering an additional 25 percent. They also receive a subsidized interest rate. If a student does not have a SISBEN, his/her social strata are used for allocation. If she is in social strata 1 or 2, he/she will get credit covering 75 percent of tuition fees only, while if he/she is in strata 3 to 6, he/she will access credit for 50 percent. The estrato socioeconomic in Colombia is based on the outside characteristics of a neighborhood and its dwellings. Neighborhoods and rural areas are grouped into 6 strata (poor to rich) and used to target public services and subsidies in Colombia. (Source: Marcucci and Johnstone, 2010). University of Ghana http://ugspace.ug.edu.gh 21 2.5.2.4 Means Testing Procedure of Costa Beca de Asistencia Socioeconomica (and Beneficios Complementarios) Means testing formula: A regression formula is used to estimate the socioeconomic level of the applicant’s household and assign them to one of eleven categories that determines eligibility for tuition fee waivers and other financial assistance. (Source: Marcucci and Johnstone, 2010) 2.5.2.5 Means Testing Procedure of Kenya Higher Education Loans Board loan and bursary Financial information collected for use in means test: Family income Information collected as proxy indicators of likely family financial strength or for use in the means test: Information collected for corroboration purposes: Parental education children in school Information collected for use in the means test: ol attended University of Ghana http://ugspace.ug.edu.gh 22 Means testing formula (2009/2010): Family Income Loan Award Bursary Award ≤Kshs 250,000 (US$8,470) single parent household Ksh 55,000 (US$1,860) Female: Ksh 7,000 (US$237) Male: Ksh 6,000 (US$203) Ksh 250,000-400,000 (US$8,470-13,550) single parent household Kshs. 50,000 (US$1,695) Female: Kshs. 6,000 (US$203) Male: Kshs. 5,000 (US$169) Kshs 400,000 – 850,000 (US$8,470 – 28,795) single parent household Kshs. 45,000 (US$1,524) Female: Ksh 5,000 (US$169) Male: Kshs 4,000 (US$136) ≤Kshs 250,000 (US$8,470) two parent household Ksh 45,000 (US$1,524) Female: Ksh 5,000 (US$169) Male: Kshs 4,000 (US$136) Kshs 250,000 – 600,000 (US$8,470 – 20,325) Kshs. 40,000 (US$1,355) No bursary Kshs. 600,000 – Kshs. 850,000 (US$20,325 – 28,795) Kshs. 35,000 (US$1,186) No bursary >Kshs. 850,000 (US$28,795) No loan No bursary Source: Marcucci and Johnstone, 2010 The maximum loan amount is based on secondary school attended so that even if a student comes from a single parent household with an income of less than Kshs 250,000, if he/she attended a national or high cost private school the maximum loans to which he would be privy is Kshs. 35,000. The other loans maximums are as follow: Provincial school: maximum of Kshs. 45,000 District school: maximum of Kshs. 50,000 Day school: Kshs. 55,000. Source: (Marcucci and Johnstone, 2010) University of Ghana http://ugspace.ug.edu.gh 23 2.5.2.6 Means Testing Procedure of South Africa National Student Financial Aid Scheme [NSFAS] Definition of independent students: Students that are married, widowed, divorced or orphaned or who have supported themselves for more than three years are considered independent. Treatment of independent students: The treatment of independent students in the means test does not differ from that of dependent students in that means test is done on household members. Financial information collected for use in means test: year gross income (before tax – from salary, wages, grants paid out, business profit, profit from investments and any form of informal sector income) is used for all household members. Used by higher education institutions’ financial aid offices to calculate disposable income (money that a family has left over after taxes have been paid and household earnings allowances have been set aside for the family’s general subsistence needs as well as the individual needs of each family member). iption of dependents in the household Information collected as proxy indicators of likely family financial strength or for use in the means test: None of these is used in means test, but some higher education institutions collect some of the following information: University of Ghana http://ugspace.ug.edu.gh 24 ons and employers Means testing formula: The NSFAS means test is based on an applicant’s family’s disposable income defined as total gross income from which taxes and annual subsistence allowances (combination of General Household Subsistence Allowance and Personal Allowance), which are set annually based region and on family size, are deducted. (Marcucci and Johnstone, 2010) 2.3.2.7 Means Testing Procedure of United States Subsidized Stafford Loan, Federal Perkins Loan, Pell Grant, and Federal Supplemental Education Opportunity Grant Purpose of Means Testing Instrument: Targeting and allocation Definition of independent students: To be considered independent of his/her parents for the purposes of the FAFSA, a student must meet at least one of the following seven criteria: University of Ghana http://ugspace.ug.edu.gh 25 of the court, or was a ward of the court until the age of 18; whom a financial aid administrator makes a documented determination of independence by reason of other unusual circumstances. Financial information collected for use in means test: Applicant (and spouse) information collected includes adjusted gross income (previous year); income tax paid; tax exemptions; balance of cash, savings and checking accounts; net worth of investments (including real estate, but not home); net worth of business or farms; education credits; child support; taxable earnings from need based employment programs; student grants and scholarships; combat pay; and untaxed income. If applicant is dependent, the following information is collected: adjusted gross income of parents; parents’ income tax; parents’ exemptions; parent’s earnings; parent’s current balance of cash, savings and checking accounts; net worth of their investments including real estate (excluding family home); net worth of their businesses and or investment farms; education credits; child support paid; taxable earnings from need based employment programs; student grant and scholarship aid; combat pay; and untaxed income. University of Ghana http://ugspace.ug.edu.gh 26 Information collected as proxy indicators of likely family financial strength or for use in means test If a student is dependent, the following information is collected: number of people in parents’ household; number of college students in household (Marcucci and Johnstone, 2010). Means testing formula: Information from the FASFA form is used by the Central Processing System to calculate an official Expected Family Contribution (EFC) that is provided to student in Student Aid Report (SAR). The schools listed in the student’s FAFSA receive this information in an electronic file called an Institutional Student Information Record. For dependent students, total allowances against parent’s income (income tax paid plus tax and income protection allowances) are deducted from total income to get available income. Assets (not including family home) are multiplied by 0.12 to get contribution from assets. Parent’s contribution is calculated by adding available income and contribution from assets to get adjusted available income and then using a table to get total parents’ contribution from adjusted available income. Total parents’ contribution from adjusted available income is divided by number in college to arrive at parents’ contribution. The Student’s contribution is calculated by deducting taxes and tax allowances from student’s total income and multiplying it by 0.50 to get student’s contribution from University of Ghana http://ugspace.ug.edu.gh 27 available income and adding this student’s contribution from assets (total assets times 0.20). The EFC is parents’ contribution plus student’s contribution from available income plus student’s contribution from assets. The EFC calculation is established by law. To determine need, EFC is subtracted from cost of attendance. Marcucci and Johnstone noted that both institutional need-based grants and merit-based grants provide larger awards to families with greater total income especially at high tuition private institutions. Scheme Award = costs (registration and tuition fees, essential books, accommodation and food) – bursaries – scholarships – academic rebates – EFC (Source: Marcucci and Johnstone, 2010) Some institutions choose not to fund first year students, and some take performance into account. After studying the means testing formulae of the various countries it was realized that socio-economic status is a major determining factor followed by levels of family income which almost all countries mentioned or considered in the formulae. In the United States and South Africa, for instance, independent students are treated differently from dependent students. Marcucci and Johnstone made an observation that higher awards are given to students from high income families. This implies that means testing formulae discussed are not adequately addressing the targeting issues and therefore needs reviewing and improvement. This problem could be attributed to students not providing University of Ghana http://ugspace.ug.edu.gh 28 correct information or errors in the computations and setting of thresholds for individual factors. This model or means testing formula when developed will be subject to amendments periodically. 2.5.3 Factors Considered by some African Countries in Conducting Means Testing Tekleselassie and Johnstone (2004) asserted that, it is required of parents to submit information about household income and assets in Mozambique. Merisotis and Wolanin (2002), also added that this income and asset information is normally supplemented with categorical information on parents’ occupation, whether the home has running water and/or electricity, and the key mode of family transportation (e.g., car, public transportation, car and driver provided by agency, etc.) “In Uganda, several proxy variables are used to signify income and determine ability to pay for higher education” (Tekleselassie and Johnstone, 2004). The father’s level of education and the mode of transportation used are the major barometers to classify students among three income groups (Mayanja, (1998) as cited in Tekleselassie and Johnstone (2004)). Classified as high income are families with professional fathers who have more than 15 years schooling (i.e., first degree or above); businessmen fathers with private or official vehicles; and professional fathers with 15 years or less of schooling but with a personal or official car (Tekleselassie and Johnstone, 2004). University of Ghana http://ugspace.ug.edu.gh 29 Classified as middle-income families are those whose fathers are professionals with 15 years or less of schooling but with cars and businessmen and farmers with no personal or official vehicles (Tekleselassie and Johnstone, 2004). Classified as low-income families are peasants and those who are not employed (Tekleselassie and Johnstone, 2004). 2.5.4 Justifications for Means Testing Several authors have put across their justifications for means testing; These are critically examined. Ngolovoi’s (2008) main justification for means testing is that it addresses efficiency in the use of scholarship funds, and wrote that this can be achieved if subsidized financing strategies reach the target group, which constitutes needy students. The second justification by Ngolovoi is the case for equity, argued in the line that because middle and upper income students are easily able to afford higher education, means tested financial aid should increase access and participation of needy students. Ziderman and Hoddinott (as cited in Ngolovoi) also put across a very interesting justification that since the financial aid is purely scholarship based, means testing becomes imperative to ensure that only the needy receive funds. Marcucci and Johnstone (2010) share their view on the importance of good means testing by emphasizing that the success of student assistance policies in meeting their objectives in a financially sustainable manner ultimately rests on fair and accurate means testing that ensures financial assistance to eligible students and families and avoids or minimizes awards to non-poor students. University of Ghana http://ugspace.ug.edu.gh 30 2.5.5 Constructing of Means Testing There have been some methods or suggestions described by different writers as to how means testing is conducted and we present some of the shortcomings of such methods. First of all the simplest way of testing the means of a student and his or her family is getting the income of the student and the family and then subtract all expenses made by the student (including tuition fees, lodging, food and other expenses). The difference is what tells whether the family has the means to actually contribute to the student’s education and the proportion it can contribute and then it would determine whether the student is eligible to receive financial assistance or not. Tekleselassie and Johnstone (2004) claimed that assets are used in addition to income to determine eligibility of targeted subsidies. They went ahead to draw attention to some shortcomings for using assets and said that the use of assets measurements may be unreliable, especially where information may be withheld from relevant officials and added that assets may not serve as a corroboration of reported current income but may be assumed to be a part of parental contribution in a means testing approach. Tekleselassie and Johnstone went on to establish that income is not the only indicator for assessing means or need and the fact that there are other indicators other than income and assets which are referred to as categorical indicators and also known as proxy means testing; a categorical approach generally employs multiple indicators to supplement whatever is available on income and assets. They attributed some strong advantages associated with categorical indicators; the first advantage was that categorical indicators maximize the social objective for which the scholarships are designed. Secondly, they are University of Ghana http://ugspace.ug.edu.gh 31 difficult to manipulate and relatively easy to observe (hence less costly to measure), they can be used either as an alternative or as a supplement to income testing. Lastly, in practice almost all means tested schemes are conditional not only on income but also on satisfying certain categorical criteria. Tekeleselassie and Johnstone gave examples of what the categorical indicators could be – occupation, type of housing, place of residence, automobile ownership, family size and age of children, gender, ethnicity and other characteristics. Coady, Grosh and Hoddinott (as cited in Tekleselassie and Johnstone, 2004) in agreement to what had been written previously, maintains that in proxy means testing, it is first important to identify variables that exist in the surveys that are highly correlated with household income that are observable and easily manipulated by households in an attempt to benefit from scholarships or other social programmes. They made an assertion which is paramount to this study that; countries, institutions or bodies can determine their choices of variables which they can then associate with different weights statistically. Coady et al., made contributions by claiming that a key feature for proxy means testing is that it has the merit of making replicable judgments using consistent and visible criteria which means it should guarantee “horizontal equity” – the same or similar households receive the same treatment even if evaluated by different officials or by the same official on different days. Tekeleslassie and Johnstone brought to the fore some problems that can be associated with proxy means testing in spite of their usefulness in supplementing the information University of Ghana http://ugspace.ug.edu.gh 32 obtained through determining or estimating income. The main problem that was pointed out was imperfect targeting which may arise either from a loose connection between the categorical indicator and the benefit (example, family size or place of residence and eligibility for scholarships), or from error of ambiguities in identifying the categorical indicator itself (example, place of residence or ethnicity). Atkinson’s, Sen’s and Walle’s studies (as cited in Tekeleselassie and Johnstone, 2004) revealed that these imperfections may lead to Type I errors, resulting in the exclusion of eligible families and they can also lead to Type II errors, which would result in scholarships awarded to students who are not in need and ought not have been eligible. From the discussions above on how to conduct means testing, its advantages and short comings, the researcher finds out that there has not been an almost perfect method for determining the means or the need of a student or a family. This study is conducted to make a contribution to the improvement of the method. This acknowledgement was confirmed by Tekleselassie and Johnstone (2004) by their observation that “even supplementing income/assets measurements with categorical indicators does not solve all the limitations of subsidy targeting and that the search for workable approaches is a continuous exercise – one which is just beginning in only few developing countries”. The method developed in this study is exhaustive enough to address some of the limitations faced by other means testing formulae. University of Ghana http://ugspace.ug.edu.gh 33 2.5.6 Limitations of the Means Testing Approach These following limitations were put across by Tekleselassie and Johnstone (2004) 1. There may be no effective way of getting information on the income except, perhaps, of those in the formal sector. 2. The market value of real property may not be clearly known. 3. Finally, to the extent that real property might be included in assessing financial means, there may be few ways to convert this asset to cash short of selling it. University of Ghana http://ugspace.ug.edu.gh 34 CHAPTER THREE METHODOLOGY 3.1 Introduction This chapter outlines the research methodology employed to achieve the objectives of the study which were listed in the first chapter. It first of all discuses the study design, followed by the study population, sample size estimation and sampling technique, method of data collection, data explorations (preliminary analysis) and the tool used for the data analysis. 3.2 Research Design The study design adopted in this work was quantitative. In this approach, a reasonable portion of the population was selected and a large amount of data was gathered on them. Hence, the information gathered from the participants was used to make generalizations which cover the entire population from which the participants were taken. The information on the sample was collected through questionnaire administration. The quantitative design was chosen for this study because the outcome of the study would be generalised to cover the entire regular student population. University of Ghana http://ugspace.ug.edu.gh 35 3.3 Study Population The University of Ghana was founded in 1948 as the University College of Gold Coast on the recommendation of the Asquith Commission on Higher Education in the British colonies (Handbook for Graduate Studies, 2010). It was established for the purpose of providing for and promoting university education, learning and research (Regulations for Junior Members and Students Facilities, 2012). The student population during the 2012/2013 academic year is 35,638 (with a male/female ratio of about 3:2) the University of Ghana is the oldest and largest of the six public Universities in Ghana. The total number of students included 4,437 at the Accra City Campus and 4,532 undertaking their studies by the Distance Mode. Also included in this number are 3,196 post-graduate students and 3,596 students on modular or sandwich programmes. (Regulations for Junior Members and Students Facilities, 2012). The campus of the University lies between about 13 kilometres north-east of Accra, the capital of Ghana, at an altitude of between 90 and 100 metres. Within this dimension are facilities such as Halls of Residence, Departments, Lecture Halls, Laboratories, the Balme Library and Auditoriums (Regulations for Junior Members and Students Facilities, 2012). There are markets and supermarkets on the campus and across the road, the Accra- Dodowa road from the main University gate is a Police Station, a University Hospital and housing for Junior Staff of the University. The College of Health Sciences has its administration as well as the Medical/Dental/Allied Health Sciences located at the Korle- University of Ghana http://ugspace.ug.edu.gh 36 Bu Teaching Hospital, which is about three kilometers west of the centre of Accra and about 18 kilometres from the main University campus (Regulations for Junior Members and Students Facilities, 2012). Academic life of the University of Ghana is centered around Colleges, Faculties, Institutes/Schools and Centres of Research/Learning (Regulations for Junior Members and Students Facilities, 2012). Colleges Under colleges we have Medical School, Dental School, School of Allied Health Sciences, School of Public Health, Noguchi Memorial Institute for Medical Research, School of Nursing and School of Pharmacy (Handbook on Regulations for Junior Members and Students Facilities, 2012). There is also the College of Agriculture and Consumer Sciences which is constituted by two Schools and a Research Institute, and they are School of Agriculture, School of Veterinary Medicine; and the Institute of Agricultural Research under which are Livestock and Poultry Research Centre – Legon, Soil and Irrigation Research Centre – Kpong and Forest and Horticultural Crops Research Centre – Kade (Regulations for Junior Members and Students Facilities, 2012). Faculties – Arts, Science, Law, Social Studies, Business School and Engineering Sciences. University of Ghana http://ugspace.ug.edu.gh 37 3.4 Source of Data The University of Ghana students comprise of graduate and undergraduate students and within these two groups are Accra City Campus, Distant Learning, Part-time, Fee-paying and Regular Students. The study sampled regular undergraduate students who are in levels 200, 300 and 400. Level 100 students were not included because there was an element of CGPA (Cumulative Grade Point Average) in the Questionnaire which they did not have at the time the study was conducted. 3.5 Sampling Procedure The items discussed here are estimation of sample size and the procedure by which the members of the sample were selected. 3.5.1 Sample Size Estimation Ascertaining the sample size is a very crucial part of the study and there is no mysterious formula that will tell us the perfect sample size for this study, we therefore need to choose a formula that would be more appropriate so far as this study is concerned. In the process of deciding on what proportion of the population to be directly studied, two factors were considered – margin of error and precision (Lohr, 1999). According to Lohr (1999), in designing a Simple Random Sample one must decide what amount of sampling error in the estimates is tolerable and must balance the precision of the estimates with the cost of the survey. There are several formulae for calculating the size of the sample to be University of Ghana http://ugspace.ug.edu.gh 38 studied depending on the nature of the research being conducted; what is expected of the sample and information on the target population available (Lohr, 1999). Even though many variables are measured in this study, attention was centered on only one response which is whether the student needs financial assistance, and this is for the purpose of estimating the sample size. The response on the need of financial aid is considered for this estimation because it is the main focus of the study. It is worth noting in conducting a survey that precision is a very crucial factor which needs to be considered right at the stage of determining the size of the sample. There is therefore the need to find an equation that relates sample size and expectation of the sample (Lohr, 1999). The desired precision which is often expressed in absolute terms as, is the simplest equation that relates the sample size and the researcher’s expectation of the sample Lohr (1999). This equation comes from the Confidence Interval for mean from a Simple Random Sample, that is but SE ( ) = (Lohr, 1999) To obtain absolute precision e we find a value of n that satisfies University of Ghana http://ugspace.ug.edu.gh 39 The vital outcome of this is how precise the estimates are and not the proportion of the population studied. As confirmed by Lohr (2010) that precision is obtained through the absolute value size of the sample and not the proportion of the population covered except for small populations. The only place that the population size occurs is in the finite population correction (fpc) factor in the variance formula of the Confidence Interval formula. The fpc has no effect on the variance of the estimator in larger populations. Since the population size for this study is 19,877 students, which is large, the fpc is ignored in the computation of the sample size using the precision formula. The formula now becomes and Since the response on the need of financial aid which is in proportions is the focal point of this study and therefore for the purpose of estimating the sample size, the standard error s2 would be replaced by pq where p and q are the proportions of students who need financial aid and students who don’t need financial aid respectively. Therefore, As the proportion of needy students in the population is not known we choose p = 0.5 because according to Lohr (2010) for large populations S2 ≈ p(1 – p) and it attains its maximal value when p = 0.5. The margin of error chosen is e = 0.05 and = 0.05 Hence, University of Ghana http://ugspace.ug.edu.gh 40 n ≈ 384 This sample size was settled on because as earlier said that for large populations the precision is attained not through the proportion of the population studied but the absolute size of the sample. This assertion was confirmed by Cochran (1977) that if the population exceeds 8000, the sampling fraction is less than 5% and no adjustment for fpc is called for. 3.5.2 Sampling Design The sampling procedure that was used is Simple Random Sampling; the design was achieved by acquiring the identification numbers of all the 19,877 students which served as the sampling frame. The 19,877 Identification numbers were entered into an excel worksheet and used the software was used to generate 384 ID numbers randomly. The 384 randomly selected ID numbers were traced to the corresponding names and departments of the students. 3.6 Data Collection Since the source of data for the study is primary, it was collected through questionnaire administration. Secondary data was not used because most of the information was not available at the University. The questionnaires were responded to by students of the various levels of study indicated earlier of which the majority was the level three hundred students and across the course of study more of the Arts and Science students responded very well. University of Ghana http://ugspace.ug.edu.gh 41 The questionnaire was designed in nine sections (from Section A to Section I) the least number of questions in a section was four, the longest section had twelve questions and on the average there were ten questions in a section. See Appendix B for questionnaire. The questionnaire had both types of items which are open and closed ended questions and the form of administration was self – administration. This form was used because it was a university community and it was presumed that students would understand the questions and answer accordingly without the researcher’s assistance and the second reason was budget and time constraint. In all three hundred and eighty four questionnaires were distributed and three hundred and seventy eighty were returned. 3.7 Method of Data Analysis This section entails data cleaning and preparation process, description of the data, testing of necessary assumptions, procedures used for the main analysis and discussion of the underlying theories. 3.7.1 Preliminary Analysis This section of the analysis involves examination of the data; identification of missing data and appropriate remedies to replace them, detection and handling of outliers and testing of the assumptions underlying factor analysis. The descriptive statistics of the data were computed and discussed to enable the researcher have a general overview of the University of Ghana http://ugspace.ug.edu.gh 42 study sample. These were critical because they helped to gain a basic understanding of the data, to acquire some primary information with regards to the demographics of the sampled population, the relationships between the variables, and also to check if the data to be used for the analysis meet the requirements for the specific analytical tool. 3.7.1.1 Data Cleaning During this exercise some issues were encountered such as errors in inputting the data; these errors were detected by juxtaposing the data with the questionnaires (each completed questionnaire was cross checked with the data using the questionnaire identification number as the reference point). The errors were duly corrected. Moreover, there were a few problems with the coding of some of the responses which were also changed accordingly. 3.7.1.2 Treatment of Missing Values and Outliers Examination and replacement of missing data are crucial issues because they could create problems in later data analysis, especially for complex once such as multivariate analysis. There was the need to replace missing data because the analytical tool does not accommodate that. Missing responses represent values of variables that are unknown, either because respondents did not provide answers to those questions or their answers were not properly recorded (Naresh, 2004). University of Ghana http://ugspace.ug.edu.gh 43 About six questionnaires were discarded due to the fact that the respective respondents did not answer up to fifty percent of the questions. The mean response which is considered as a neutral value was substituted for the missing responses of the questionnaires that were retained. As argued by Naresh (2004) that the mean remains unchanged and other statistics such as correlations are not affected much. There may be a few questions in connection with the logic of substituting a mean for respondents who, if they had answered, might have used either high or low ratings even though this approach has some merits (Naresh, 2004). The modal response was used to replace the missing values of the categorical variables. There were no clearly observed outliers in the data perhaps due to the nature of the grouping of certain variables. The eighth item of Section F of the questionnaire was not used in the analysis due to the reason that over fifty percent of the respondents did not answer the questions in that section. 3.7.1.3 Description of Data Tables were basically used to present the descriptive statistics of the data, frequencies were mainly used and a few cross tabulations in describing the nature of the data. This was because the researcher thought the tables depicted more characteristics (total number of respondents, and cumulative percentages) of the data as compared to charts. University of Ghana http://ugspace.ug.edu.gh 44 3.7.1.4 Testing of Assumptions There are a few conceptual assumptions which are associated with factor analysis and these issues were discussed before the statistical assumptions were considered. Conceptual Issues A basic assumption of factor analysis is that some underlying structure exits in the set of selected variables (Hair, Black, Babin, Anderson and Tathan, 2006). The researcher ensured that the observed patterns were conceptually valid and appropriate to employ factor analysis since the technique has no means of determining appropriateness other than correlations. If there are no patterns, the existence of correlated variables and the subsequent definition of factors do not achieve relevance even if the statistical requirements are met (Hair et al., 2006). This makes the conceptual issues very paramount in this study. Secondly, the sample must be homogeneous with respect to the underlying factor structure (Hair et al., 2006). With the issue of sample size, Leech, Barret and Morgan (2005) think that sample size is less crucial for factor analysis to the extent that the communalities of items with the other items are high, or at least relatively high and variable. The minimum sample size is 50. Hair et al. (2006) also suggest that the sample size should be preferably 100 or larger and the total response gathered for this study was 372. University of Ghana http://ugspace.ug.edu.gh 45 Statistical Issues The normality of the distributions of the variables was checked by computing the skewness and kurtosis statistics. Some degree of intercorrelations among variables or multicollinearity is desirable and this was simply checked by constructing a correlation matrix for the data. Due to the huge nature of the correlation matrix because of a large number of variables, visual inspection (to check if the matrix has considerable number of correlations that are greater than 0.30) was tedious. Therefore, the statistically significant Bartlett’s test of sphericity which was to test if sufficient correlation exits among the variables was computed. If the significance value is less than 0.05 it indicates that there is sufficient correlation and we can therefore proceed. The Kaiser-Meyer-Olkin measure of sample adequacy which is an index used to examine appropriateness of factor analysis was used and a value greater than 0.5 indicates that factor analysis is appropriate for the data. 3.7.2 Factor Analysis Multivariate statistical technique of factor analysis has experienced increased usage during the past decade in all fields of business-related research and as the number of variables to be considered increases, so does the need for increased knowledge of the structure and interrelationship of the variables (Hair et al., 2006). University of Ghana http://ugspace.ug.edu.gh 46 Factor analysis is employed in this study because of its ability to examine the underlying patterns or relationships for a large number of variables and to determine whether the information can be condensed or summarized in a smaller set of factors or components. The study has large number of variables used and hence the need to be reduced and summarized for subsequent analysis. The most important reason of factor analysis is to describe, if possible, the covariance relationship among many variables in terms of underlying, but unobservable, random quantities called factors (Johnson and Wichen, 2007). Empirically, there is an argument that motivates the factor model, and it is as follows: suppose all the variables within a particular group are highly correlated among themselves but relatively small correlations with variables in a different group, then it can be conceived that each group of variables represent a single underlying construct, or factor which is responsible for the observed correlations (Johnson and Wichern, 2007). For example, correlations from a need analysis data, the group of; family owning its dwelling, number of rooms, hectares of land owned by the family, household owning a large scale business, locality of dwelling, building material of dwelling, collected by spearman’s correlation suggests an underlying “socio-economic status” factor. A second factor representing type of Junior High and Senior High Schools attended, course of study and guardian’s highest qualification might correspond to “educational background” factor. University of Ghana http://ugspace.ug.edu.gh 47 The unique factors must be uncorrelated with each other and with common factors. The common factors themselves can be expressed as linear combinations of the variables of a few random variables f1, f2, …,fm (m