|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 DEBT FOR DIPLOMA : AN EXAMINATION OF STUDENT LOAN BORROWI NG AND INDEBTEDNESS AMONG 2007 GRADUATES By LYLE MCKINNEY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTI AL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010
2 2010 Lyle McKinney
3 T o my family
4 ACKNOWLEDGMENTS I am indebted (pun intended) to a host of wonderf ul individuals whose guidance and unwavering support have contributed immeasurably to my growth as a person and educational scholar. Through every phase of my academic career, my family has been a source of great encouragement and inspiration. The joys of my personal accomplishments are a lways sweetened when shared with my family whom I love dearly Mom, Dad, Stuart, and Zack thank you for your unconditional love and support. My doctoral experience has been enhanced by the opportunity to work under the daily mentorship of Dr. Luis Ponjuan. I am grateful for the thoughtful attention he has given to my research agenda and to my preparation as a higher education scholar Undoubtedly I still have much to learn, but I feel confident in my abilities because o f the time I have spent collaborating While constantly challenging me to deliver my best work, he also provided the support I need ed to be successful. His sense of humor and dedication to our countless research projects made our office a grea t place to work. Thank you Dr. Ponjuan for the time and energy you have invested in me I wish you, Laurel, and Davis the very best I am especially thankful for the privilege to study and work closely with Laura Waltrip over the past four years. As my cl assmate, co teacher, and colleague on the Ponjuan research team Laura has played an integral role in my doctoral experience I could not have asked for a better research colleague or friend. keen insights have broadened my worldview and I will sor ely miss working with her on a daily basis More importantly, Laura compassion for other people serves as an example for how I want to live my own life Thank you for everything Laura You are the best! The four faculty members I selected to serve on my committee proved to be helpful and supportive beyond belief Dr. Ponjuan, my passion for conducting meaningful research is a direct
5 result of our time together Dr. Mendoza, I never walk away from our conversations without learning something. Dr. Honeyman helped me discover the value of multidisciplinary research Dr. Mike, your accessibility to students is overshadowed only by your commitment to helping them succeed. I am thankful for the inva luable feedback each of my committee members provided on earlier drafts of my dissertation Moreover, I appreciate the sincere interest they took in my life outside of the classroom. The department office staff has demonstrated unbelievable patience in helping answer my countless questions and special requests over the last four years. Angela Rowe, I will miss you from the bottom of my heart. You took such good care of me during my time in the department. I will miss our big hugs and chatting about Gato r athletics Go Gators! Patty, you are one of the most genuine about your latest fishing adventures. You are a I can state without hesitation that Angela and Patty played an equally important role in the completion of my degree as did the members of my dissertation committee. Thank you both! My journey to become Dr. McKinney has been marked by triumphant experiences and considerable challenges. Fortunately, I had the sup port of great family, friends, and colleagues who helped me persevere through it all. God has blessed my life in ways I never expected or thought possible. I am grateful for where I have come from, and hopeful for where I am going. As my time in Gainesvill e draws to a close my greatest hope is that I will capitalize on all that I have learned so that I may lead a life mark ed by compassion and meaningful service to others.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 9 LIST OF FIGURES ................................ ................................ ................................ ....................... 11 ABSTRACT ................................ ................................ ................................ ................................ ... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 14 Purpose of the Study ................................ ................................ ................................ ............... 17 Research Questions ................................ ................................ ................................ ................. 18 Rationale for the Study ................................ ................................ ................................ ........... 18 Scope of the Study ................................ ................................ ................................ .................. 23 Significance of the Study ................................ ................................ ................................ ........ 23 Chapter Summary ................................ ................................ ................................ ................... 24 2 LITERATURE REVIEW ................................ ................................ ................................ ....... 26 Definition of Key Term s ................................ ................................ ................................ ......... 26 Reasons for the Rise in College Student Indebtedness ................................ ........................... 29 Student Borrowing and Debt Burden: A Review of the Literature ................................ ........ 32 ................................ ......................... 33 Student Loan Default ................................ ................................ ................................ ....... 36 Borrowing Behavior and High Loan Debt Burden ................................ .......................... 37 Summary of Existing Research ................................ ................................ .............................. 40 Conceptual Framework ................................ ................................ ................................ ........... 41 Layer 1: The Student and Family Context ................................ ................................ ...... 46 Demographic characteristics ................................ ................................ .................... 47 Soci ological perspectives: Social and cultural capital ................................ ............. 47 Economic perspectives: Human capital theory ................................ ........................ 49 Layer 2: School and Communi ty Context ................................ ................................ ....... 51 Layer 3: Higher Education Context ................................ ................................ ................. 52 Layer 4: Social, Economic, and Policy Context ................................ .............................. 54 Application of the Conceptual Model to the Present Study ................................ ............ 56 Chapter Conclusion ................................ ................................ ................................ ................ 57 3 METHODOL OGY ................................ ................................ ................................ ................. 58 Proposed Hypotheses ................................ ................................ ................................ .............. 58 Data Source ................................ ................................ ................................ ............................. 59 Sampling Design ................................ ................................ ................................ ..................... 60
7 Data Sample ................................ ................................ ................................ ............................ 61 Dependent Variables ................................ ................................ ................................ ............... 64 Independent Variables ................................ ................................ ................................ ............ 65 Layer 1: The Student and Family Context ................................ ................................ ...... 68 Demographic Characteristics ................................ ................................ .......................... 68 Social and cultural capital ................................ ................................ ........................ 69 ................................ ................................ .... 70 Supply of resources ................................ ................................ ................................ .. 71 Expected benefits ................................ ................................ ................................ ..... 73 Expected costs ................................ ................................ ................................ .......... 73 Layer 2: School and Community Context ................................ ................................ ....... 74 Layer 3: Higher Education Context ................................ ................................ ................. 74 Layer 4: Social, Economic, and Policy Context ................................ .............................. 75 Methodological Considerations ................................ ................................ .............................. 75 Analyzing Secondary Datasets ................................ ................................ ........................ 75 Estimating Causal Effects with Secondary Data ................................ ............................. 77 Analytic Methods ................................ ................................ ................................ .................... 79 Preliminary Data Analysis ................................ ................................ ............................... 80 Advanced Data Analysis ................................ ................................ ................................ 80 Regression Diagnostics ................................ ................................ ................................ ... 82 Limitations of the Study ................................ ................................ ................................ ......... 84 Contributions of this Study ................................ ................................ ................................ ..... 85 4 DATA ANALYSIS AND RESULTS ................................ ................................ .................... 87 Preliminary Data Analysis ................................ ................................ ................................ ...... 87 Dependent Variable One: The Probability of Borrowing through Student Loans .......... 87 Dependent Variable Two: Levels of Student Loan Debt among Borrowers ................... 89 Advanced Data Analysis ................................ ................................ ................................ ......... 94 Logistic Multivariate Regression: The Probability of Borrowing through Student Loans ................................ ................................ ................................ ............................ 94 Linear Multivariate Regression: Levels of Student Loan Debt among Borrowers ......... 99 Summary of Results ................................ ................................ ................................ .............. 105 5 DISCUSSION ................................ ................................ ................................ ....................... 107 Purpose of the Study Revisited ................................ ................................ ............................. 107 Discussion of Results ................................ ................................ ................................ ............ 108 Demographic Characteristics ................................ ................................ ......................... 109 Social and Cultural Capital ................................ ................................ ............................ 112 Demand for Higher Education ................................ ................................ ....................... 115 Supply of Resources ................................ ................................ ................................ ...... 116 Expected Benefits ................................ ................................ ................................ .......... 118 Expected Costs ................................ ................................ ................................ .............. 120 Institutional Characteristics ................................ ................................ ........................... 121 Summary of Discussion ................................ ................................ ................................ ........ 123
8 6 RECOMMEND ATIONS AND FUTURE RESEARCH ................................ ...................... 127 Recommendations for Policy and Practice ................................ ................................ ........... 128 Financial Literacy Education ................................ ................................ ......................... 128 Financial Aid Counseling ................................ ................................ .............................. 130 Financial Aid Reform ................................ ................................ ................................ .... 132 Recommendations for Future Res earch ................................ ................................ ................ 134 New Federal Student Financial Aid Policy ................................ ................................ ... 134 ................................ ................................ ............................ 136 Credit Card Debt ................................ ................................ ................................ ............ 138 Methodological Approaches ................................ ................................ .......................... 139 Closing Words ................................ ................................ ................................ ...................... 140 LIST OF REFERENCES ................................ ................................ ................................ ............. 141 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 151
9 LIST OF TABLES Table page 3 1 Frequency of gender (n=25,671) ................................ ................................ ........................ 62 3 2 Frequency of race/ethnicity (n=24,779) ................................ ................................ ............. 63 3 3 Frequency of dependency st atus (n=25,671) ................................ ................................ ..... 63 3 4 Frequency of institution sector (n=23,457) ................................ ................................ ........ 64 3 5 Summary of dependent variables ................................ ................................ ....................... 64 3 5 Frequency of borrowing among 2007 08 (n=25,671) ................................ ........................ 65 3 6 Descriptives for amount still owed on all undergraduate loans (n=16,333) ...................... 65 3 7 Summary of independent variables and indices ................................ ................................ 66 3 8 Summary of multivariate regression models ................................ ................................ ..... 81 4 1 Chi square tests for the probability of borrowing through student loans .......................... 88 4 2 Frequencies and percentages for the probability of borrowing through student loans ...... 88 4 3 t test of mean cumulative level of student loan debt (n=16,333) ................................ ...... 90 4 4 e/ethnicity (n=15,730) ................ 90 4 5 ................................ ................................ ................... 91 4 6 ........ 91 4 7 s ................................ ................................ ............ 92 4 8 .... 92 4 9 ................................ ................................ ........ 93 4 10 ANOVA for level of student loan debt by type of postseco ndary institution (n=14,988) ................................ ................................ ................................ .......................... 93 4 11 postsecondary institution (n=14,988) ................................ ................................ ................ 94
10 4 12 Binary logistic multivariate regression model measures for the probability of borrowing through student loans (n=25,671) ................................ ................................ .... 95 4 13 Results for binary lo gistic multivariate regression model for the probability of borrowing through student loans ................................ ................................ ....................... 95 4 14 Unstandardized beta coefficients for blocked entry regression on the cumulative level of stu dent loan debt (n=16,333) ................................ ................................ .............. 101
11 LIST OF FIGURES Figure page 2 1 ................................ ................................ ........................ 44 2 2 .................. 45 3 1 NPSAS:08 variables depicted within the conceptual model ................................ .............. 67
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DEBT FOR DIPLOMA : AN EXAMINATION OF STUDENT LOAN BORROWING AND INDEBT EDNESS AMONG 2007 GRADUATES By Lyle McKinney August 2010 Chair: Luis Ponjuan Co chair: Pilar Mendoza Major: Higher Education Administration This study examined the probability of borrowing and cumulative levels of student loan de the 2008 National Postsecondary Student Aid Study conducted by the National Center for Education Statistics. college choice was utilized as a conceptual framework to guide this study. This model offered a new theoretical approach to the study of student loan borrowing and indebtedness. Descriptive statistics t tests, ANOVAs, and m ultivariate regression techniqu es were used to examine the individual, familial, and institutional factors associated with loan borrowing behavior. characteristics, social and cultural capital, supply of financial resources, and postsecondary institution type were strongly associated with the probability of borrowing and cumulat ive level were more likely to be fe male, Black, independent, and have higher levels of unmet financial need after receiving all grant aid In addition, students who demonstrated poorer academic
13 performance and graduated from private postsecondary institutions were more likely to borrow and have higher levels of loan debt Findings from this study highlight the importance of providing students, and their families, with pertinent information about the costs and benefits of borrowing through student loans to finance their college educati on. Financial literacy programs and better financial aid counseling could help students acquire the knowledge and financial management skills they need to make informed decisions about using student loans. Furthermore, results corroborate the need to incre ase the amount of grant aid available to college students and reduce the financial hardships of graduates who leave higher education with high levels of outstanding student loan debt. Promisingly, President Barack Obama signed a new student loan reform bi ll into law in March 2010 that has the potential help assuage the skyrocketing levels of indebtedness among recent college graduates. The implications of this study are discussed in light of this new federal legislation and recommendations are made to help strengthen future research related to student loan borrowing. Collectively findings from this study can better inform public policies and unmanageable debt burden.
14 CHAPTER 1 INTRODUCTION unprecedented. The average student loan debt for graduating seniors has nearly tripled since 1990, with the average debt at $2 3 2 00 for borrowers who graduate d during the 200 7 200 8 academic year ( National Postsecondary Student Aid Study [NPSAS] 200 8 ). In addition, more students are borrowing today than ever before to cover the escalating costs associated with two graduated with federal and/or private education loan debt in 2008 ( NPSAS 200 8 ) A growing percentage of these students are borrowing at levels placing them at greater risk for loan default or even bankruptcy aft er graduation. In 1993, only 1 % of graduating seniors had borrowed $40,000 or more in student loans to pay for college. The proportion of all graduates borrowing in excess of $40,000 had risen to 1 0 % by 20 08 ( NPSAS 200 8 ). ncreased reliance on loans to pay for college represents a shift from the public to private financing of higher education in the United States (Heller & Rogers, 2006). As tuition rates have increased rapidly within the last two decades and governmental fin ancial support for higher education has decreased, individuals have been forced to pay a greater share of the total costs of attending college. As a direct result, higher education has become less affordable for a growing number of students and their famil ies (Price, 2004a). Many students are left with no choice but to borrow through loans in order to bridge the gap between their available resources and the escalating costs associated with earning a college degree. Consequently c ollege student indebtedness has drawn widespread attention in the media and public policy debates (Gladieux & Perna, 2005) and has been identified as an impending national crisis (Carey & Dillon, 2009).
15 The past two decades have brought about dramatic shifts in how students and th eir families pay for college. Student loans have become the system and have replaced grants as the dominant source of student financial aid in the United States (Dowd, 2008; Heller & Rogers, 2006) Education l oa ns offered through the federal government are now the primary source of funding that undergraduates use to finance their college education During the 2007 08 academic year, federal loans accounted for 4 5 % ($56.5 billion) of all financial aid delivered to undergraduates (College Board Trends in Student Aid 200 9 ). Borrowing through the federal student loan program s has simply become part of the college experience for the majority of undergraduates A growing number of students are also relying on private or alternative, loans to pay for college. Private loans are typically provided through a bank or private lending agency and have less favorable terms than federal student loans. Unlike private loans, student loans delivered through the federal go vernment are guaranteed and carry a fixed interest rate. For the 200 8 0 9 academic year, the interest rate was 6% for newly originated subsidized federal loans to undergraduate students. In contrast, private loans are not guaranteed by the government and th ere are no limits on the interest rates and fees that private lenders can charge. Some private lenders have variable rates on student loans as high as 19% (Block, 2006). In 1997 98 private loans represented 7% of all education loans, but by 2007 08 privat e loans accounted for 23% of the total loan volume (College Board Trends in Student Aid 2008). The most recent data from the College Board (2009) suggest that private loans had declined to approximately 13% of total borrowing in 2008 09. Students who acqu ire private loans often do so in order to borrow more than the maximum amount allowed by the federal loan programs,
16 which could potentially lead to unmanageable levels of debt among these private borrowers ( Carey & Dillon, 2009; King, 1998). Debt burde n among college student borrowers remains an important public policy issue because of its adverse affects on individuals and society. At the individual level, unmanageable loan debt burden may lead a student to drop out before earning their degree (Cofer & Somers, 2000 ; Gladieux & Perna, 2005 ) or decide not to attend graduate school ( Heller, 2001; Millet, 2003). Students who graduate with high debt burden may even be forced to postpone life milestones such as purchasing a home, getting married, or having ch ildren (A merican Association of State Colleges and Universities [AASCU] 2006 ; Baum & Sanders, 1998 ). In addition, students with excessive levels of education loan debt have a greater likelihood of defaulting on their loans during repayment. Defaulting on a student loan can result serious in financial hardships for students such as credit report damage, denial of a professional license, ineligibility for federal financial aid in the future, legal action, and forfeiture of tax refund payments (Gladieux & Pe rna, 2005). The student debt problem also has significant consequences at the societal level. T he federal government incurs tremendous costs when large numbers of student borrowers default on their federal education loans because the government covers the losses to lenders (Christman, 2000; Flint, 1997). Consequently, a higher federal student loan default rate means greater costs to may suffer when large numbers of college graduates are forced to devote a signif icant share of their income to the repayment of debt (Harrast, 2004). Heavy debt burden can also discourage students from pursuing valuable public service careers, such as teaching or social work, because these careers have low earning potential (AASCU, 20 06). Furthermore, since minority and low income students are at the greatest risk of
17 acquiring excessive debt burden, the student indebtedness problem threatens to exacerbate longstanding social inequalities (Price, 2004a; 2004b). The many adverse affects of excessive debt burden corroborate why policymakers, financial aid experts, and the general public have all ased reliance on loans to finance their college education. Purpose of the Study While most students a re borrowing for higher education at reasonable levels, the fact that a growing number of students are acquiring high levels of loan debt and experiencing difficulty making their monthly repayments cannot be ignored (Baum & Schwartz, 2006). In 2008, 15% of who borrowed had accumulated $40,000 or more in loan debt during their time in college ( NPSAS, 2008 ). The severity of the student debt problem is further intensified by the fact that more students than ever before are using cre dit cards to pay for tuition and other college related expenses (Sallie Mae, 2009) S tudents with high debt burden are more likely to experience diminished economic returns to their investments in higher education and encounter financial hardships after gr aduation. Perpetual tuition increases, reductions in federal and state appropriations for higher education, and increases in private lending all threaten to exacerbate the student debt problem (Carey & Dillon, 2009). The purpose of the present study wa s to examine the decision to borrow and level of student loan indebtedness among 2007 performed descriptive and multivariate analysis of data available from the 2008 National Postsecondary Student A id Study (NPSAS:08) conducted by the U.S. Department of Education. At the time the present study was conducted, NPSAS:08 offered the latest and most comprehensive national data available pertaining to use of student loans to pay for college The comprehensive nature of the NPSAS:08 dataset allowed for statistical analysis
18 of copious variables that could potentially influence initial decision to borrow for college and their cumulative level of indebtedness upon graduating with their Research Questions T here was a primary research question and two sub questio ns guiding this empirical study: What individual, familial, and institutional factors influence d the probability of borrow ing and cumulative level of student l oan degree from United States postsecondary institutions during the 2007 08 academic year? 1. What individual, familial, and institutional factors are associated with 2007 degree gr 2. What individual, familial, and institutional factors are associated with the total level of student loan debt among 2007 The conceptual framework g uiding this st udy, which will be introduced and discussed in detail in subsequent chapters, provided the rationale for examining individual, familial, and institutional variables. In addition, it is important to note that this study examined the total amou private, state, and institutional Rationale for the Study Dramatic increases in tuition rates and burgeoning levels of student debt propose an important question: Is American higher education worth what it cost? Every year the federal and state governments, along with millions of citizens, collectively invest billions of dollars in higher education. Since the immediate economic sacrifices associated with expenditures on higher education are not trivial, clearly governments and individuals expect to realize substantial benefits in return for their investments. The notion that education has some benefits may be self eviden t, but determining whether these benefits actually exceed initial expenditures is a question of great importance (Psacharopoulos, 2006). Traditionally, researchers and policymakers have
19 relied heavily upon the rational behavior approach proposed by human c apital theory to examine why individuals and governments invest in education (Paulsen & Toutkoushian, 2008). Building upon concepts first introduced by Adam Smith in his classic book The Wealth of Nations economists in the 1960s first applied human capit al theory as a way of estimating the value of education for individuals and the larger society (Bloom, Hartley, & Rosovsky, 2007). The seminal work of Schultz (1961) and Becker (1964) used human capital to suggest that individuals and society derive benefi ts from investments in people (Sweetland, 2006). Since the 1960s, countless research studies have used human capital theory as a framework to identify and measure the specific benefits that result from investments in education. Researchers have typically c lassified these benefits as private or public. The private benefits of an investment in education refer to those benefits enjoyed directly by the individual (Cohn & Geske, 1990). The most obvious private benefit associated with participation in higher edu cation is an increase in personal financial earnings. A college education is correlated with higher wages for both men and women, and among all racial/ethnic groups ( Baum & Ma make $1.2 million more throughout his/her working career than someone with only a high school diploma (Kantrowitz, 2007). College graduates are also less likely to become unemployed and more likely to have higher saving levels than individuals who did not earn a college degree (Bloom et al., 2007). In addition to these economic benefits, college graduates profit in many other ways from their participation in postsecondary education. Research suggests that attending college can enhance the overall quality of a p expectancy, family life, consumptive behavior, and asset management (Baum & Ma, 2007; Bowen, 1997; Geske & Cohn, 1998).
20 The public benefits of education are those benefits that society derives beyond those en joyed by the individual (Cohn & Geske, 1990). The far reaching social benefits of investment s being and vitality. From an economic perspective, producing a more educated citizenry has numerous b enefits for the federal and state governments. The federal government collects $132,762 more in income tax revenue (Kantrowitz, 2007). Adults with higher levels of ed ucation are less likely to become reliant upon federal assistance programs or become incarcerated, which decreases demand on public budgets (Baum & Ma, 2007). Furthermore many of the public benefits of higher education help fuel and strengthen democracy. Individuals who attend college are more likely to vote and are 30% more likely to be interested in politics than people who did not attend (Bloom et al., 2007). Participation in higher education has also been shown to increase charitable giving, volunteering, and openness to the opinions of others (Baum & Ma, 2007). College graduates are also more likely than non graduates to lead healthy lifestyles (Baum & Ma, 2007), which can result in considerable tax savings for federal and state governments. The collective findings from numerous research studies suggest that both individuals and governments experience benefits from investments in higher education (Bowen, 1997; Cohn & Geske, 1990 ; Paulsen, 2001 ). The private benefits of higher education hel p explain why millions of individuals are willing to endure immediate financial and social sacrifices in order to pursue a college degree. These personal sacrifices may include forgoing potential earnings that would result from working a full time job or a cquiring student loan debt in order to pay the costs associated with attending college. In addition, the public benefits resulting from investments in
21 higher education illuminate why governments allocate funding for postsecondary institutions and provide a ssistance directly to college students through financial aid programs. Through these investments, governments expect to realize an overall increase in human capital as the result of a more educated, engaged, and productive citizenry (Psacharopoulos, 2006). Traditionally, higher education has been viewed as a social good that provides a multitude of benefits to the entire nation (Ehrenberg, 2006). During the 1960s and early 1970s, the federal government invested heavily in higher education primarily by prov iding appropriations directly to public colleges and universities. The underlying purpose of this strategy was to strongly support the public institutions so they could in turn offer low cost tuition to the general public. Proponents of this approach asser ted that while higher education undoubtedly benefited the individual, these private gains were far outweighed by the overall public benefits experienced by society (Alexander, 2002). From this perspective, policymakers view higher education as a public obl igation and not a private privilege. Within the past two decades, however, the burden of paying for higher education has dramatically shifted from federal and state governments to students and their families (Heller & Rogers, 2006). A decrease in federal and state government financial support for postsecondary institutions and escalating tuition rates have forced students to pay for a greater share of the costs of higher education (Alexander, 2002; Heller & Rogers, 2006). In addition, many research studies examining the returns on investments for higher education have estimated that the private returns outweigh the larger benefits to society (see Bloom et al., 2007). These estimates, in conjunction with the widening earning differential between college educ ated and less educated privilege and not a public obligation. The perspective that higher education is a private good
22 reflects the notion of academic capitalism and ha s resulted in the rationale that students, and their families, should be responsible for paying the bulk of their college costs (Slaughter & Rhoades, 2005). The privatization of higher education has resulted in a dramatic increase in the number of stu dents who borrow to cover their college related expenses. From the perspective of human capital theory, students borrow to pay for the immediate costs of higher education with the expectation that this borrowing will help them realize benefits over the lon g term (Becker, 1993) The accessibility of federal and private loans has made attending college a reality for thousands of students who would have been unable to afford higher education without this financial support. Student l oans have also given a growi ng number of students the ability to decide whether, and how much, they want to invest in their own human capital. Conversely the growing reliance on credit to pay for college related expenses has led many students to acquire excessive levels of debt. E because of the monetary returns associated this investment However, high levels of student loan debt burden threaten to significantly diminish the monetary return s o n this investment for a growing number of undergraduate borrowers. There are growing concerns that high debt burdens a significant portion of college graduate ment Accordingly, i t is critically important that the welfare of college students and society be improved, and not exacerbated, as a result of student borrowing (Harrast, 2004). Identifying the levels of debt burden he is essential in order to accurately estimate the private and public returns to investments in higher education.
23 Scope of the Study The scope of this study was to examine the borrowing behavior and cumulative level of indeb ted ness from student loans 2007 08 nationally representative sample. In particular, NPSAS:08 data set was used to identify the factors salient in predicting debt levels among undergraduates who relied upon st udent loans in route to earning their four year degree. The degree during the 2007 08 academic year. Undergraduates who borrowed but dropped out before ed in the sample. In addition, it was beyond the scope of this study to examine borrowing behavior and debt levels among students enrolled in graduate school or pursuing a professional degree. The NPSAS:08 dataset contains information on students attendi ng public and private non profit and for profit institutions, community colleges, and four year colleges and universities. Therefore, results of this study reflect student borrowing behavior and debt burden among undergraduates who attended all types of po stsecondary institutions in route to earning Findings from this study reflect the borrowing behavior and levels of student loan indebtedness among undergraduate borrowers in the United States and should not be generalized to other countries. Significance of the Study Empirical and current research on college student indebtedness is needed to inform targeted policy responses that can help assuage the burgeoning levels of debt burden among recent college graduates. The present study identifie d the factors that decision to borrow, or not borrow, through student loans in order to pay for their college related expenses In addition, this study examined the factors contributing to different levels of student loan deb t among 2007 08 Identification of these factors is the
24 foundation for the development and implementation of public policies and institutional practices aimed at protecting college students from acquiring unmanageable debt burde n. Reducing the percentage of college graduates with high debt burdens can help decrease student loan default attainment across the country. Chapter Summary Thi s chapter highlighted the problem of rising student loan debt burden undergraduate s More students are borrowing than ever before to finance their college education and a growing number of students are accumulating unmanageable levels of debt The purpose of the present study was to examine the decision to borrow and level of student loan indebtedness Particular attention was given to undergraduate borrowers who acquire th e greatest amount o f debt in route to earning their degree The NPSAS:08 dataset from the U.S. Department of Education was utilized because at the time of this study it represented the latest and most comprehensive source of national data pertaining to the topic of student i ndebtedness. Identifying the levels of debt burden among student borrowers is paramount in order to accurately estimate the private and public returns to financial investments in higher education. While the problem of student debt has received widespread attention in the media and public policy debates in recent years, there are relatively few empirical research studies that focus on the borrowing behavior and cumulative levels of student loan debt burden among different es Specifically, the present study examines leve ls of indebtedness from federal, private, state, and institutional loans aimed at prote cting undergraduate borrowers from acquiring unmanageable debt burden.
25 T he subsequent chapter examine s how changes in the landscape of American higher education and dramatic shifts in the financial aid system have contributed to the rise in college stu dent indebtedness In addition, the chapter gives attention to the immediate and impending student loans to finance their college education. Relevant research literature pertaining to student borrowing and loan debt is presented. oice is adapted for the purposes of this study and used as a conceptual framework for understanding and their cumulative level of student loan debt upon gra
26 C HAPTER 2 LITERATURE REVIEW The purpose of this chapter is to provide a better understanding of the problem of student indebtedness in American higher education and develop the theoretical framework guiding this stu dy The chapter begins by defining the key terms and concepts that will be referenced throughout the remainder of this study. In the second section, attention is given to how increases in tuition rates and changes in federal financial aid policy have trigg ered increased reliance on student loans to pay for college. The next section provides a review of the extant research literature on student borrowing behavior and debt The conceptual framework used to guide this study, el of college choice is presented and discussed in the subsequent section. The literature review concludes with a summary of the highlights presented throughout the chapter. Definition of Key Terms Debt burden. The level of difficulty or hardship a stude nt borrower experiences in repaying his/ her education loans been measured by the ratio of monthly student loan repayments to gross monthly income (Baum Fed eral education loans. Funding made available by the federal government allowing qualified students to borrow a specified amount of money each year while they are enrolled in higher education. However, the money borrowed must be repaid with interest. The ma jority of federal education loans awarded directly to postsecondary students are delivered through the Stafford Loan program, which offers both subsidized and unsubsidized loans (defined elsewhere in this section). The loans disbursed through the Stafford Loan program are guaranteed by the federal government and are awarded through the government, private lenders, or banks. Loans
27 provided by the government are administered by participating postsecondary institutions through the Direct Loan program. Gran ts. Financial assistance awarded to students in the form of scholarships, fellowships, tuition waivers, and/or employer tuition reimbursements. Grants to students attending higher education are provided through numerous sources (e.g. federal government, st ates, postsecondary institutions, civic organizations, private foundations) and students do not have to repay these or a combination of both merit and need. Higher Education Act (HEA) Landmark federal legislation authorized in 1965 that introduced the first broad based, federally funded financial aid programs in America. Title IV of the HEA of 1965 established the Educational Opportunity Grant program (re named Pell Grants in 1980), the Guaranteed Student Loan Program, and the College Work Study program. Today, all three of these programs continue to play a significant role in the disbursement of student financial aid. The HEA goes through legislative reaut horization periodically as changes to federal financial aid policy are deemed necessary. Loan default Default on an education loan occurs when the student borrower can no longer make the minimum repayments on his/her student loan debt. Per the 1998 HEA amendments, a federal student loan is considered in default when the borrower has not made required payments for 270 days on loans with monthly payments or 330 days for loans with less frequent payments. Student loan default can result in credit report damage, loss or denial of a professional license, ineligibility for future federal financial aid, wage garnishments, forfeiture of tax refund payments, and legal action (Gladieux & Perna, 2005).
28 Merit based financial aid Merit based financial aid progra ms use some form of eligibility criteria to determine whether a student merits financial assistance (Creech & Davis, 1999). A student may demonstrate merit on the basis of his/her academic, artistic, or athletic achievement. In addition, a student may be c onsidered meritorious if he/she provides a valuable service to society, such as serving in the armed forces or intending to become a teacher in an underserved community. Need based financial aid Need based programs use some type of eligibility criteria t o much the student could reasonably pay for one year of higher ed ucation at a particular college or overall level of financial n eed. Need based financial aid programs are centered on the goal of access and seek to increase opportunities for low income students to attend college (Doyle, 2008). Pell Grants Pell Grants are federally funded and are awarded to financially needy underg ular college or university (McPherson & Schapiro, 1998). For the past four decades, Pell Grants have been the primary means through which federal assistance for college has been delivered to students with demonstrated financial need (Heller & Rogers, 2006)
29 Private education loans Private, or alternative, loans are provided through a bank or private lending agency and have less favorable terms than student loans provided by the federal government. Private loans are generally considered a riskier form of lo an borrowing since they are not guaranteed by the government and there are no limits on the interest rates and fees that private lenders can charge. Subsidized federal loans financial need. The federal government pays the interest on these loans while the student is enrolled in college and for a grace period of six months after the student leaves higher education. tus (dependent or independent) and class year. Unsubsidized federal loans Unsubsidized loans are available to students regardless of their financial need, but the student is responsible for repaying the interest accrued. Interest accumulates on the princ ipal loan amount the student borrows while in school and not yet in repayment. Like subsidized loans, the maximum amount a student may borrow annually through unsubsidized loans is determined by his/her dependency status and class year. Reasons for the Ris e in College Student Indebtedness The dramatic increase in borrowing and debt burden has been attributed to at least three major factors. First, as tuition rates at all types of postsecondary institutions continue to increa se a growing number of st udents have turned to loans (Heller & Rogers, 2006) to pay for their college related expenses Secondly, federal Pell Grants have been unable to keep pace with rising tuition rates, forcing a growing number of low income students to take out loans to pay for college (Curs, Singell, & Waddell, 2007; Redd, 2004). Third, several key legislative acts have broadened eligibility requirements for federal education loans and raised the maximum dollar amount students may borrow annually (Ki ng, 2005). The interplay of
30 these factors has contributed to the rise in student borrowing and triggered widespread concerns graduates Considerable attention has been given to the skyrocketing cos ts associated with attending higher education. Recent data from the College Board (2009) indicate that from 1999 2000 to 2009 10 tuition and fees at public four year universities rose at an average annual rate of 4.9%. This rate of increase was higher tha n in either of the previous two decades. Tuition and fees have also continued to steadily climb at private colleges and universities for profit postsecondary institutions, and community colleges. In addition to paying for tuition and fees, students must a lso purchase books and supplies and pay for living expenses such as housing and food (College Board, 2009). Consequently, a growing percentage of college students have turned to education loans in order to bridge the gap between their existing financial re sources and the rising costs associated with attending higher education. The declining purchasing power of federal Pell Grants has forced a growing number of low income s tudents to borrow through loans to pay for college (Curs, Singell, & Wad dell, 2007; Gladieux & King, 1995). For the past four decades, federal Pell Grants have been the primary means through which need based financial aid for college students has been delivered (Heller & Rogers, 2006). Approximately 14% ($1 8.2 billion) of all financial aid awarded to undergraduates during the 200 8 0 9 academic year was delivered through Pell Grants (College Board, 200 9 ). While Pell Grants have helped make college a reality for thousands of low income students the program has been criticized for failing to keep pace with the rapid increases in college tuition and fees (Redd, 2004). During the 1987 88 academic year, the maximum Pell Grant award covered 50% of tuition and fees at public four year institutions and 20% at private four year
31 institutio ns. However, Pell Grants covered only 32% of public costs and 13% of private costs during the 2007 08 academic year (College Board, 2008). The federal government provides education loans to college students primarily through the Stafford Loan program. Ove r time, Stafford Loans have transformed from a small program designed to supplement Pell Grants into the largest student financial aid program in the countr y. The reauthorization of the Higher Education Act in 1972 triggered a shift in federal financial ai d from grants to primarily loans. While this shift provided more financial aid dollars overall for students, it significantly increased the share of college costs paid by students and their families (Price, 2004b) By 1980 the Stafford Loan program had su rpassed Pell Grants as the dominant source of federal student aid (Mumper & Vander Ark, 1991). Even more students began turning to federal loans during 1980s and 1990s when college tuition rose faster than the rate of inflation (Alexander, 2002). By the 19 94 95 academic year, federal loans provided more than twice as much money as all other federal student financial aid prog rams combined (Hartle, 1996). The 1992 reautho rization of the HEA triggered yet another dramatic increase in student borrowing. This legislation broadened eligibility for federal subsidized loans, increased the annual borrowing limits, and established the unsubsidized loan program that was open to all students (King, 2005). The result was a rise in student borrowing that began to emerge almost immediately after this legislation was passed. Between 1992 93 and 1995 96, undergraduate borrowing through the federal loan programs had increased by more than 100% (King, 1999). By the 200 8 0 9 academic year, federal loan disbursements had increas ed to $ 56.5 billion and represented 4 5 % of all financial aid disbursed to undergraduates (College Board, 200 9 ). The dramatic rise in federal student loan borrowing has heightened concerns about student loan default rates Per the 1998 HEA amendments, a fe deral student loan is considered in default
32 when the borrower has not made required payments for 270 days on loans with monthly payments or 330 days for loans with less frequent payments. Student loan default has severe consequences for borrowers postseco ndary institutions, and the government For student borrowers, loan default can result in credit report damage, loss or denial of a professional license, ineligibility for future federal financial aid, wage garnishments, forfeiture of tax refund payments, and legal action (Gladieux & Perna, 2005). A colle ge or university with excessive default rates forfeits their eligibility to participate in the federal student financial aid programs. Since t he government covers the financial losses lenders experience whe n borrowers default on their federal student loans, increases the national default rate produce tremendous costs for the federal government (Christman, 2000; Flint, 1997). The default rate among student borrowers increased during the late 1980s until it p eaked in 1991 when 22% of borrowers defaulted within their first two years of repayment (Fossey, 1998). Default rates began to decline after 1991 as a result of significant changes in the federal lending policy. Throughout the early 2000s, the annual stude nt default rate remained between 4% and 6% (U.S. Department of Education, 2009). However, the national student loan cohort default rate for the 2007 fiscal year increased to 6.7%, which was a notable rise from the 2006 rate of 5.2%. The da ta indicate that more than 225, 300 student borrowers, whose first repayments where due between October 2006 and September 2007, had defaulted on their loans by September 2008. This recent increase in the national cohort default rate may legitimize concerns (Pilon, 2008) th at the latest economic downtown will cause a growing number of borrowers to default on their student loans. Student Borrowing and Debt Burden: A Review of the Literature A review of the extant research on college student indebtedness reveals an abundance of studies education loans to pay for college. Researchers from the
33 fields of education, economics, sociology, psychology, and finance have all contributed to the growing body of literature on student borrowing behavior and de bt burden. In addition, policy and non profit organizations have played an important role in generating policy reports that examine the student debt problem (Dowd, 2008). The majority of research on student indebtedness to date has focused on us e of federal education loans to finance their college education. However, in recent years a growing number of research studies and policy reports have examined the impact of private loans. This literature review addresses three topics germane to student borrowing and high debt burden 1 First, attention is given to the body of research that has examined perceptions about borrowing to pay for college. The second section examines research studies focusing on student loan default Thir d, a review is provided of the studies and policy reports that identify those students at risk of acquir ing the highest levels of loan debt These three areas of research are important for the present study because these studies illuminate many of the fact ors The literature review concludes with a summary of several key themes from the existing research literature on college student indebtedness Attitudes and Perce ptions of Borrowing A growing body of research literature has examined towards borrowing to pay for college and how borrowers perceive their debt burden Researchers have investigated the attitudes and perceptions held by high schools s tudents (Perna, 2008), borrowers enrolled in Nicholson, 2006), 1 The present study is concerned with the cumulative debt burden held by borrowers who have already earned their making and behavior while enrolled (e.g. persistence) are not included in this literature review. Readers are encouraged to see Dowd (2008) for a review of existing literature on this topic, including potential solutions to many of the methodological challenges that have typ ically plagued this line of research.
34 borrowers who dropped out of college (Gladieux & Perna, 2005), and borrowers who graduated from college (Millett, 2003). The bulk of existing research on this topic perceptions of credit card s with fewer studies giving attention to d perceptions of education loan debt Collectively, this line of research provide s a better understan ding of the diverse factors that sh views of borrowing and debt burden and help s explain differences in borrowing behavior across groups. In general, researchers have found that the majority of college students have favorable attitudes regard ing the use of credit (Lyons, 2008; Xiao, Noring, & Anderson, 1995). S tudents with more tolerant attitudes towards debt typically have higher levels of debt burden (Davies & Lea, 1995) Furthermore, college students tend to be even more tolerant of debt on ce they have become indebted (Davies & Lea, 1995). These findings suggest that students who have the most liberal attitudes towards borrowing via student loans or credit cards, to pay for college are at the greatest risk of acquiring unmanageable debt bur den. Another noteworthy finding within this line of research is that some student groups are more averse to borrowing than others Research suggests that A sian, Hispanic, and low income students are less willing to borrow to pay for college (Callender & J ackson, 2005) As a result many s tudents belonging to these particular groups make a conscious decision not to take out education loans (Burdman, 2005; Cunningham & Santiago, 2008). While an aversion to borrowing can prevent students from accumulating exc essive debt burden, it may also function as a barrier to college access or cause a student to drop out because of a lack of financial resources. Furthermore, s tudents who avoid borrowing may be forced to attend college part time or work more hours, both of ir degree (Dowd, 2008).
35 Students attitudes and perceptions towards borrowing appear to be influenced by a variety of cation loans to pay for college are shaped by messages received from their parents, school counselors, and teachers. Peer groups also play a role in helping students make decisions about borrowing for college (Tierney & Venegas, 2006 ) Trent et al. (2006) found that students expecting to earn a professional degree and students who believed that good luck is important for success were more likely to acquire education loans. Other studies suggest that a college (Joo, Grable, & Bagwe ll, 2003 ; Trent et al., 2006 ) and perceptions of money in general (Hayhoe, Leach, & Turner, 1999; McDonough & Calderone, 2006) shape their views about borrowing and the prospect of becoming indebted. Researchers have examined how borrowers perceive the co llege related debt they have associated with repayment. However, they also found that some students had negative attitudes toward their education debt, and that lingering loan debt was a major reason why many students did not pursue graduate school. Several studies indicate that negative attitudes and perceptions of being. For example, students with high debt burden report higher levels of overall st ress lower self esteem and a decreased sense of ability to manage their money (Lange & Byrd, 1998; Norvili tis et al., 2006). M any students appear to have considerable misperceptions regarding the use of credit to pay for college King and Frishberg (2001) found that students tend to underestimate the total costs of their loan borrowing and overestimate the am ount they will e arn upon entering the workforce In addition, students are likely to be confused by interest rates and underestimate the
36 total amount of interest when facing an extended repayment period (Lewis & van Venrooij, 1995). T hese findings suggest many student borrowers do not fully understand the consequences of acqui ring debt through student loans Student Loan Default Numerous studies of student indebtedness have focused on the factors that lead borrowers to default on their student loans. Howe ver, the most robust and methodologically sound research on student loan default was conducted during the late 1980s and in the mid to late 1990s, in a different historical context (Gross, Cekic, Hossler, & Hillman, 200 9 ). In reviewing the extant research literature on student loan default, Gross and colleagues assert that few default studies employ ing multivariate statistical methods and using national databases have been conducted in recent years. Considering the increases in student borrowing and the cha nges in federal loan disbursement and default policies that have occurred within the last decade new research studies on loan default would represent a valuable contribution to the literature on student indebtedness. Prior research studies have primarily examined the individual and institutional factors that are associated with student loan default. In general, findings from these studies suggest that institutional characteristics are not strong predictors of ( Knapp & Se aks, 199 2 ; Volkwein & Szelest, 1995; Wilms, Moore, & Bolus, 1987). A s Volkwein et al. (1998) propose, evidence from several studies indicates that a demographic characteristics and level of success in college are more influential in predicting d efault than is the type of postsecondary institution the borrower attends The most consistent finding among existing studies is that borrowers who do not graduate are more likely to default on their loans than borrowers who do graduate (Dynarski, 1994; Knapp & Seaks, 1992; Podgursky et al., 2000; Volkwein & Szelest, 1995; Woo, 2002). This finding is attributed to the fact that while burdened with loan debt that must be repaid, borrowers who drop
37 out before graduating do not experience the increase in fin ancial earnings that a college degree provides (Gladieux & Perna, 2005) Several other variables related to a academic performance while in college have been identified as predictors of loan default. Borrowers with lower GPAs (Christman, 2000; Fl int, 1997; Volkwein & Szelest, 1995 ; Volkwein et al., 1998 ) and who fail credit hours (Christman, 2000; Steiner & Teszler, 2003) are at greater risk of defaulting on their loans. In addition, studies have found Flint, 1997; Volkwein & Szelest, 1995 ), enrollment pattern ( Podgursky et al., 2002; Woo, 2002 ), and employment status (Volkwein et al., 1998) can play a role in predicting default. A demographic characteristics can exert a strong influence on their l oan repayment behavior R esearchers Dynarski, 1994; Flint, 1997; Volkwein et al., 1998 ; Woo, 2002 ), income status (Dynarski, 1994; Knapp & Seaks, 1992), and age (Flint, 1997; Podgursky et al., 2002 ; Woo, 2002) ar e significant predictors of loan default. Specifically, t he likelihood of default is highest for borrowers who are African American Hispanic, or who come from low income families. Older students are at greater risk of default ing than younger students per haps because older students are more likely to have accumulated more overall debt (e.g. credit card, mortgage) and have dependents ( Gross et al., 200 9 ). income ( Dynarski 1994; Volkwein & Szelest, 1995 ; Woo, 2002 ), marital status (Dynarski, 1994; Volkwein et al., 1998) level of income and education ( Knapp & Seaks, 1992; Volkwein et al., 1998 ; Woo, 2002 ). Borrowing Behavior and High Loan Debt Burden A growin g number of studies have examined borrowing behavior and average loan debt burden among different student groups To date, p olicy reports and white papers published by non profit higher education organizations represent t he bulk of existing research on tre nds in
38 student loan borrowing These reports have typically conducted descriptive and/or multivariate statistical analysis of national datasets available through the National Center for Education Statistics. In addition, many existing policy reports and wh ite papers have placed particular emphasis on describing the characteristics of borrowers who acquire the highest levels of student loan debt. A smaller number of peer reviewed journal artic les have focused on high loan debt burden among college students A consistent theme among existing reports is that more students are borrowing more money via education loans than ever before to finance their college education (AASCU, 2006; Boushey, 2005; Choy & Li, 2006; College Board, 2009; King, 2005; King & Bann on, 2002; King & Frishberg, 2001; Project on Student Debt, 2009; Steele & Baum, 2009). In particular, these reports tend to highlight annual increases in loan borrowing and/or describe average debt levels across different groups of student borrowers Most extant reports have focused specifically on the debt levels of degree recipients. The latest data indicate that a pproximately two thirds of all college students who graduated during the 2007 08 academic year had acqu ired student loan debt and the average debt level for these students was $23,200 ( NPSAS, 2008 ). In conjunction with research on student loan default, policy reports and journal articles identifying the characteristics of borrowers with high debt burden he lp provide a more complete picture of the student debt problem 2 Several p olicy reports and white papers have examined student loan debt burden based upon demographic characteristics. In general, 2 Default studies undoubtedly represent an important contribution to the extant research literature on student indebtedness. However, default research alone does not provide a complete picture of the student debt pro blem. Not all borrowers who acquire high debt burden default on their loans during repayment. While these borrowers may not become an official default statistic, this does not negate the fact that many of these students will experience financial hardships and diminished life choices as a result of their heavy debt burden. The negative consequences resulting from high levels of indebtedness extend far beyond simply whether or not a borrower defaults.
39 findings from these reports suggest low income, A frican American, and Hispanic student borrow ers are more likely to graduate with the highest levels of loan debt burden ( Kantrowitz, 2009; King, 2005; King & Banno n, 2002 ). In addition, f indings indicate that students who borrow through private loans (Care y & Dillon, 2009 ; College Board, 2009; Project on Student Debt, 2009 ) or attend higher priced private or for profit institutions ( Boushey, 2005; College Board, 2009; Kantrowitz, 2009 ) are at greater risk of accumulating high debt burden A smal ler number of peer reviewed journal articles have examined the factors leading to high levels of loan debt burden among undergraduate borrowers A notable except ion is the work of Price (2004a ), who examined debt burden levels among 1992 93 ee graduates. Using a nationally representative sample, Price utilize d the 8% rule as a threshold to identify those students at the greatest risk of having excessive levels of loan debt four years after graduation. The 8% rule was a common benchmark used by the U.S. Department of Education and many lending agencies in late 1980s and 1990s to determine what constitutes an acceptable level of student loan debt for a college graduate. The rule asserts that no more than 8% of a graduation incom e should be devoted to student loan repayment and students who exceed this threshold are more likely to default (Baum & Schwartz, 2006). Price found that low income, Black, and Hispanic students were disproportionately represented among students who devote d more than 8% of their monthly income to the repayment of their education loans. In addition, Price discovered that students employed in legal occupations and professional careers were more likely to have excessive loan debt burden. Harrast (2004) examin ed borrowing behavior and loan debt levels among students attending a large, research extensive university. He found that race/ ethnicity, GPA, age, academic major, and number of semesters required for degree completion were associated with
40 levels of loan d ebt for the borrowers in his sample. Hispanic students had the highest debt burden loan debt. Furthermore, each one point increase in college GPA reduced a stud $4,402. Harrast also found that students majoring in special education, computer engineering, sociology, art history, and risk management and insurance had higher levels of debt, though he acknowledged it is difficult to determine the reason for the higher debt contribution among these particular majors. Summary of Existing Research College student indebtedness has received increased attention in the research literature within the last decade and r esearchers from numerous academic disciplin es have strengthened o ur understanding of this topic Several key themes emerged from this review of the literature. First, there is a dearth of recent, empirical research that uses multivariate statistic methods and a national representative sample to exa subsequent level of student loan indebtedness. Second, consistent f indings across multiple studies suggest that low income, Latino and African American students are at the greatest risk of defa ulting and acquiring the highest levels of student loan debt. Third Latino and Asian students appear to be the most adverse to borrowing and the least tolerant of debt T he hesitancy among Latino students to borrow or acquire debt seems to be justifiable considering they are more likely than their peers to graduate with high levels of student loan debt. One particular finding that emerged from this review of the research literature deserves further attention. There are several possible explanations for the seemingly contradictory finding s that Latino students are loan averse, but yet are disproportionally repre sented among borrowers who experience financial hardships as a result of their loan borrowing Findings are lik ely to have lower starting salaries upon
41 entering the workforce than White students ( Saenz & Ponjuan, 2009 ). Therefore, Latino students who borrow may experience increased hardships during loan repayment because their loan debt levels are high relative to their income A second explanation for the contradictory findings regarding the borrowing behavior of Latino students is the research methods that have been used to derive these findings (Dowd, 2008). In recent years, several researchers have highlighted the methodological problems that are evident in many existing causal studies of student financial aid (Cellini, 2008; Chen, 2008; Dowd, 2008). These problems could have potentially resulted in statistical bias and i naccurate results. In addition, few exist ing studies that address student loan borrowing have incorporated interaction terms to Researchers have advocated for the use of interaction terms in future studies of student fin ancial a function of their level of income. Conceptual Framework Borrowing to pay for higher education is a choice each college student must make. The majorit y of undergraduate students in American can gain access to federal and/or private loans with relative ease. T he initial decision a student must make is simply whether or not he/she will rely upon credit to cover the immediate expenses associated with atten ding higher education For students who do make the ini tial decision to borrow, these individuals must then decide how much debt they are willing to accrue in order to earn their degree or achieve their academic goals. Therefore, a framework for understand ing how college students make choices has inherent value in examining student borrowing behavior and debt burden. The present study adapts through student loans and the cumulative level of loan debt among borrowers.
42 Traditionally, many research studies examining student choice have either explicitly or implicitly relied upon economic models of human capital investment Studies conducted using this framework place the individual at the center of the decision making process (Long, 2007) and view students as rational actors whose college related choices are based upon their comparison of the perceived costs and benefits of these choices (Perna, 2006a). Utilizing this economic framework many existing studies have treated student decision making as a linear and sequential process of cost continues with reenrollment and persistence, and the n culminates with graduation and career choice (Dowd, 2008). A major contribution of t he rational human capital investment model has been its attention to finance variables such as financial aid, tuition rates, and family income that frequently exert a st related decisio ns (Paulsen, 1998 ; Perna, 2006a). While the human capital approach has proven valuable for studying how finances affect college going choices and behavior, the explicit focus on individua l level decision making has significant limitations in explaining differences in college choices across student groups (Perna, 2006a) Consequently, a growing number of researchers have drawn attention to the inherent limitations of relying exclusively on the rational human capital approach to understand issues related to college access, choice, and success ( Dow d, 2008; Perna, 2006a; St. John & Paulsen, 2001 ). These r esearchers have assert ed that exogenous factors such as institutional characteristics and public policies, often influence education and have advocate d for the use of a multi disciplinary approach in studying college choice.
43 Accordingly, Perna (2006a) developed a conceptual model of college choice that in corporates key constructs from both economic and sociological perspectives (see Figure 2. 1) Her model also takes into consideration relevant social structures and resources that can influence student decision making about higher education Specifically P of college choice college erna, 2006b, p. 1621). In addition to offering a multi disciplinary approach to the study of college choice a ma attention given to exogenous variables like institutional characteristics and public policy that can s hape student choice and behavior. have applied the model to guide their own research However, the model may prove to be especially useful in helping researchers dev elop a better understanding of how different sources of financial aid affect student decision making and behavior. Financial aid research in particular has be inclined to rely heavily on the human capital framework and researchers have been slow to conside college aspirations, and socioeconomic characteristics influence college choices (Dowd, 2008). By accounting for many relevant sociocultural factors that tend to be omi tted in financial aid researchers with a more sophisticated understanding of how financial aid impacts student choice and college going behavior. Additionally, P erna (2008) has demonstrated the relevance of her model in studying financial aid by using it as a framework to examine the factors that shape high
44 Figure 2 1. del of college choice
45 Figure 2 2. understand undergraduate borrowing behavi or (see Figure 2.2). Specifically, the model is applied
46 approach v iews student borrowing primarily as an individual offers a multilayer context that accounts for the multitude of external factors that can influence a ssumes that students will make decisions about borrowing through student loans within a situated context. That is, by their family context, the characteristic s of the high school and postsecondary institutions they attend, and by the larger social, economic, and policy context (Perna, 2008). While most existing research studies of student borrowing and indebtedness have either explicitly or implicitly relied on a rational human capital investment framework the present decision to borrow and the level of student loan i The followi ng call attention to the specific constructs within each layer that are salient in explaining student borrowing behavior. In addition, r elevant research on student borrowin g and indebtedness is highlighted within each contextual layer of the model to corroborate the applicability of the model for this study Layer 1: The Student and Family Context Layer 1 cs as well as his/her access to social and cultural capital, exert an influence on college choice (Perna, 2006a). class g 1704).
47 habitus can e xert a strong influence on the decisions they make about borrowing through loan s to finance the ir college related expenses Demographic c haracteristics The conceptual model suggests that individual characteristics like gender, race/ethnicity, and socioeconomic status influence student decision making about higher education. These three demographic characteristics have been analyzed in numerous existing research studies pertaining to college student borrowing and indebtedness (see Flint, 1997; Kantrowitz, 2009; Price, 2004a; Volkwein et al., 1998) In general, most of these studies have found little difference between male and female college students with regards to their use of education loans. and socioeconomic status impact their borrowing behavior and debt le ve ls (see Flint, 1997; Kantrowitz, 2009; King & Bannon, 2002; Price, 2004a; Volkwein et al., 1998) Findings suggest that Asian Hispanic, and low income students are more averse to borrowing than other student groups and may completely avoid using loans t o finance their college education (Burdman, 2005; Cunningham & Santiago, 2008 ; Mendez & Mendoza, 2008 ). For students who are debt averse, the perceived costs of borrowing appear greater than the perceived benefits of using credit as a means of financing th eir degree. Furthermore, findings suggest that Latino and low income students who do borrow are at the greatest risk of defaulting on their loans (Volkwein et al., 1998) and accumulating high debt burden (Price, 2004 a ). Sociological p erspectives: Social a nd c ultural c apital Social capital and cultural capital are useful concepts in studying college choice because they provide a framework for examining how socioeconomic characteristics influence student decision making about higher education. These sociolo gical perspectives reflect differences in
48 models (Perna, 2006a). Theories of social and cultural capital are particularly useful in helping researchers develo p a better understanding of the non college related choices. Applied to the present study, sociological approaches can help explain differences in the ways students obtain information about education loans, and she d light on the borrowing behaviors of different student populations. The social capital framework suggests that individuals experience positive benefits from their involvement and affiliation with groups (Tierney & Venegas, 2006). Students leverage the ir social capital in order to accumulate other forms of capital (e.g. human, cultural) and gain access to institutional resources and support (see Perna, 2006a). Notable forms of social capital that students utilize when making decisions about college incl ude their parents, schools, teachers, counselors, and peer groups. The amount of social capital a particular student has access to is determined by the size of their social networks and the amount of economic, cultural, and social capital that is possessed by the individuals within these networks (Bourdieu, 1986). Students rely heavily on their social capital to make numerous decisions about higher education, including choices about borrowing to pay for college. As the model indicates, social capital can h elp students navigate the often confusing process of applying for and obtaining financial aid. Parents, school counselors, and teachers all represent forms of social capital that high school students can use when making decisions about whether or not to bo rrow for college (Perna, 2008). In addition, Tierney and Venegas (2006) found that peer groups can help high school students make decisions about applying to college and understanding how to pay for ormation, and understand the true costs and benefits of borrowing, is strongly influenced by his/her social capital
49 Cultural capital refers to the language skills, cultural knowledge, and mannerisms that are concept of cultural capital has been used to explain how membership in particular groups is advantageous for some individuals. For example, students from high and middle class families typically possess th e types of cultural capital that are the most valuable (McDonough, 1997). As it pertains to educational attainment and success, cultural capital is delivered from parents to their children in the form of knowledge and attitudes needed to successfully navig ate the education system. tudents with college educated parents often benefit from a home environment that offers a wealth of information about college to include advice about borrowing that students can access (McDonough, 1997). Conversely, first generation college students may be discouraged from using student loans because their parents are debt averse or do not understand the financial aid system. While most parents at middle and high resource high schools expect their students will need to rely upon loans to pay for college, parents at low income high schools generally do not want their students to borrow (Perna, 2008). borrowing behavior. Economic p erspective s : Human c apital t heo ry The rational human capital investment model is the foundation From an economic perspective, an individual is motivated to pursue higher education when they believe the long term benefits outweigh the immediate costs ( Becker, 199 7 ; Man ski & Wise, 1983). The immediate direct costs associated with attending college include tuition, fees, books, and living
50 expenses. In addition to these direct costs, the student also experiences the indirect costs associated with relinquishing the income t hat would be derived from working a job instead of being enrolled in college. The economic benefits accrued from participation in higher education include enhanced knowledge and training, as well as greater earning potential throughout the remainder of t he 1998). Economists assume that rational individuals will weigh these expected costs and benefits when deciding whether to invest in their own human capital by pursuing higher education (Paulsen, 2001). The cost benefit comparison is a fundamental tenet of human capital investment theory and is useful in understanding student s decision to borrow for college. In essence, students borrow in the present under the assumption that the long term monetary and nonmonetary benefi ts of earning a college degree will outweigh the immediate costs of becoming indebted. The theory claims that individuals perform the cost benefit comparison based upon the information that is available to them and does not presume that individuals have a ll of the pertinent information when making decisions (DesJardins & Toutkoushian, 2005). This suggests that knowledge regarding the actual costs and benefits of borrowing for college can vary significantly across student groups Furthermore, relying on inc omplete information or inaccurate assumptions can help explain why many students underestimate their loan borrowing and overestimate the amount they will earn upon entering the workforce (King & Frishberg, 2001). H uman capital theory (Perna, 2008 pg. 591 ). As depicted in the conceptual education is strongly i nfluenced by his/her academic preparation and performance. Students with greater academic preparation and who excel academically before college are more likely to
51 pursue postsecondary education In addition students who possess greater academic preparatio n and achievement may demonstrate a willingness to borrow for college in order to achieve their academic goals. financial resources, including family income and availability of financial aid, can influence his/her borrowing behavio r Students from high income families or students who have received generous financial aid in the form of grants may be able to afford college without the use of education loans. Conversely, many students from low income families ( Christou & Haliassos, 200 6 ) and students who receive a smaller share of grant aid are more likely to need to borrow in order to make attending college a reality. The supply of financial resources available to a student may also be dependent upon the maximum amount they are eligibl e to borrow through student loans Layer 2: School and Community Context Layer 2 of the conceptual model context and reflects what McDonough (1997) refers to as organizational habitus In partic ular, student college c While numerous studies have examined how schools impact the decisions high school students make about colle ge, little is known about how (Perna, 2008). Applied to the present study, t h is context ual layer addresses the specific types of resources and support systems (or lack thereof) a high school makes available to students that can influence their decisions about borrowing to pay for college. The high school context often reflects the amount of social capital that is available to a particular student. High income or p rivate high schools in comparison to low income or urban public schools, typically provide students with greater resources and support that can help them
52 make informed decisions about borrowing for college. Students attending higher income schools are mor e likely to have immediate access to knowledgeable financial aid counselors teachers, and peers who can provide valuable information about college costs and financing strategies ( Perna, 2008; Stanton Salazar, 1997 ; Tierney & Venegas, 2006 ). Conversely, lo wer income high schools may exhibit structural constraints that make it difficult for students to make informed decisions about the costs and benefits of borrowing For example, t he least informed counselors about college related issues are typically found in lower income schools (McDonough, 2005) and disbursing information about college costs and borrowing may not be a top priority for counselors employed at these types of high schools. High school counselors represent an important and often primary, sou rce of financial aid information for all types of students. Findings suggest that high school students are more likely to have a better understanding of college costs and financing when they consult with trained counselors (McDonough & Calderone, 2006). Ho wever m any high school counselors appear to be relatively uninformed about college financial aid and are unsure of how to advise students regarding the amount to borrow ( NACAC, 2007 ). Counselors also tailor the advice they provide based on their knowledge and assumptions about the students they work with (Dowd, 2008). For example McDonough and Calderone (2006) found that many counselors assume that African American and Latino students are averse to using loans and therefore direct them to low er cost commu nity colleges Collectively, these factors highlight the significant role of the school attitudes and perceptions of borrowing and their actual borrowing behavior. Layer 3: Higher Education Context Layer 3 of the model addres ses the higher education context and recognizes the major role related decisions (Perna, 2006a).
53 Applied to the present study, Layer 3 gives particular attention to the institutional characte ristics (e.g. institutional type, tuition rates, geographic location ) that may shape decision to borrow and cumulative level of student loan. In addition, this layer examines how institutional resources and support systems can influence student initial decision to borrow and their cumulative level of indebtedness. T he type of postsecondary institution a student attends can directly impact the extent of their debt burden For the 200 7 0 8 academic year, approximately 62 % of graduates at public four year institutions graduated with loan debt compared to approximately 72 % of graduates from private non profits and 96% of private for profit institutions ( NPSAS 2008 ). The average debt among graduates was $ 20,200 at public four year colleges, $ 27,65 0 at private non profits, and $33,050 at private for profits. In 200 7 08 48 % of community college graduates (i.e. students earn ing had student loan debt and the average debt among these borrowers was approximately half of the amount owed by borrowers graduating from public four year institutions ( Steele & Baum, 2009 ). T uition rates vary greatly across different types of postsecondary institutions and students who attend higher priced colleges and universities may be forced to borrowi ng more in order to complete their degree. The geographic location of the college or university a student decides to attend can also shape his/her borrowing behavior. Most postsecondary institutions have published tuition rates for both in state and out of state students, with the tuition charged to out of state students typically being significantly more expensive. Therefore, students who choose to enroll in an out of state college and do not receive generous grant aid that can help alleviate the total cost of attendance may find it necessary to use student loans to finance their education. In addition,
54 attending college in a city that has a relatively high cost of living may lead some student s to acquire higher debt burden in route to earning their bach Similar to the high school context addressed in Layer 2, the types of resources and support systems offered at higher education institutions can impact student borrowing behavior. Colleges and universities play a major role in prov iding students and their families with i nformation about financial aid (Perna, 2006b), which can help guide their dec isions about the use of student loans. College financial aid advisors are more likely than high school counselors to be knowledgeable about the costs and benefits of borrowing to pay for college. However, there may be vast differences across postsecondary institutions in the accessibility of these counselors and the types of information relayed to students about borrowing. While some colleges and universities offer their own institutional loan programs, others have taken steps to curtail their credit. For example, some community colleges have elect ed not to participate in the federal loan programs (Project on Student Debt, 200 9). This institutional practice has the potential to significantly influence student borrowing. Layer 4: Social, Economic, and Policy Context Layer 4 of the contextual model assumes the broader social, economic, and policy context impacts student decis ion making about higher education Applied to the present study, this layer provides a framework for understanding how economic conditions and changes in public policy can impact student borrowing and debt burden. This context ual layer also gives attention to how co llege costs and the media college students. As noted in a previous section of th is chapter, students and their families have been forced to pay a greater share of the total costs of higher education within the last two decades (Heller &
55 Rogers, 2006). This shift in the financial burden from federal and state governments to students re flects the idea that higher education is primarily a private privilege and not a public obligation. Th e current societal expectation that students and not governments, should pay the majority of college costs has contributed to the rise in student borrowi ng and debt burden (Heller & Rogers, 2006). Furthermore numerous media sources in recent years have portrayed the indebtedness of Many of these media reports have profiled student borrowers who are facing ser ious financial hardships as a result of the excessive loan debt burden they accumulated predominately negative portrayal of student borrowing may deter some students from using loans to finance their college education. F inancial aid policies especially those related to student loans, can significantly impact the borrowing behavior and debt burden of college students For example, history has shown that c hanges in the eligibility requirements and maximum award amount s f or the federal loan programs can trigger dramatic shifts in federal loan borrowing ( King, 1999 ). The borrowing behavior of students pursuing some academic majors (e.g. education, social work) or career paths (e.g. military service volunteer work ) may be i nfluenced by their intentions to take advantage of existing student loan forgiveness programs In addition, the newly established Income Based Repayment Plan may impact some students attitudes towards federal loan borrowing since the t required federal education debt payments will never exceed 15% of the (Steele & Baum, 2009, pg. 3) The economic context provides a lens for understanding how current economic conditions and characteristics can influence student borrowing and debt burden. The most recent economic
56 downturn and credit crunch has led to growing concerns that more students will face difficulties repaying their debt, and that the student default rate will increase ( Pilon, 2008 ). For students who lost their jobs due to the economic recession, many of these individuals may have no choice but to rely on student loans in order to continue attending college. Furthermore, t he current interest rates charged on different types of student loans (i.e. federal, private, state, institutional) can A pplication of the Conceptual Model to the Present Study Rather than relying solely on the economic models that have traditionally been used to examine student borrowing, the present study applies this promising new conceptual model that incorporates multip le theoretical perspectives and recognizes that decisions about borrowing occur within a situated context. The conceptual model offers a more comprehensive framework for examining the profusion of factors that can potentially g behavior and debt burden. While students do perform the cost benefit comparison predicted by human capital theory when making decisions about borrowing for college, the model suggests that assessments of the benefits and costs are shaped not only by the demand for higher directly and indirectly, by the family, school, and community context, higher education context, (Perna 2006a, pg. 119 ). Perna (200 6b ) suggests that because of the no one study can examine all of the potential relationships depicted in the conceptual model The present study will focus on the individual (i.e. Layer 1) and institution al (i.e. Layers 2 and 3) variables available in the 2008 National Postsecondary Student Aid Study that can impact the cumulative level of student loan indebtedness among 2007 variables that addr ess the current social, economic, and policy context (i.e. Layer 4) and
57 therefore the dataset has limited usefulness in estimating how this layer of the conceptual model influences student borrowing and debt burden. The model was however, particular usefu l for iden tifying relevant variables to be examined during statistical analysis. The se variables and the statistical techniques used to conduct this study are addressed in detail in the subsequent chapter. Chapter Conclusion R ising tuition rates the de clining purchasing power of Pell Grants, and legislation that has increased the accessibility of federal loans has led to skyrocketing levels of indebtedness among college students Loans are now the dominate source of student financial aid in the United States The extant research literature related to student borrowing and debt burden has and trends in student loan borrowing. While numerous research studies and pol icy reports have strengthened our understanding of student indebtedness, several notable gaps in the existing research literature were addressed by the present study. choice provided a new conceptual framework for exami ning student borrowing behavior and debt burden. The four layers of the conceptual model were described and relevant research was used to corroborate the applicability of the model in addressing the research questions guiding this study. The conceptual mod el was used as a lens to identify the demographic, familial, and institutional factors that influence student borrowing and that are associated with different levels of student loan 08 A better understanding of the se factors can lead to the development of policies and practices designed to protect college students from acquiring excessive loan debt and experiencing undue financial hardships after graduation.
58 CHAPTER 3 METHODOLOGY The obj ective of this chapter is to describe the research methodology used to conduct this study. The chapter begins by revisiting the purpose and research questions guiding this study. The next section describes the data source and sample. Next, t he dependent an d independent variables used in this study are presented and operationally defined. The following section describes the statistical methods and techniques used to analyze these variables. Attention is then given to several methodological considerations tha t had important implications for the research design of this study and data analysis Finally, this chapter concludes by addressing the limitations inherent to this study. The purpose of the present study was to examine the decision to borrow and level of student loan indebtedness among 2007 Particular attention was given to the borrowers who acquired the highest levels of debt in route to earning their degree. To that end, there was a primary research question and two sub q uestions guiding this study: What individual, familial, and institutional factors influence d the decision to borrow and overall level of student loan from United States postsecondary i nstitutions during the 2007 08 academic year? 1. What individual, familial, and institutional factors are associated with 2007 2. What individual, familial, and institutional fa ctors are associated with the total level of student loan debt among 2007 Proposed Hypotheses Based upon the research questions guiding this study and the extant literature pertaining to college student indebte dness, the following hypotheses are proposed: 1. Low income students and some racial/ethnic minority groups are less likely to borrow through student loans than both White and high income students In particular, e xisting
59 research suggest s that low income, Hispanic, and Asian students are more averse to using loans to pay for college than other student groups (Burdman, 2005; Callender & Jackson, 2005; Cunningham & Santiago, 2008). 2. When they do borrow through student loans r acial/ethnic minority and low i ncome students are more likely to acquire higher levels of student loan debt than both White and high income students. Specifically, existing research literature suggests African American and Hispanic students are likely to have the highest loan debt level s (King, 2005; King & Bannon, 2002; Price, 2004a, 2004b). Data Source The data analyzed in this study were derived from the National Center for Education Statistics (NCES) 2008 National Postsecondary Student Aid Study (NPSAS:08) At the time the present study was conducted, NPSAS:08 offered the latest and most comprehensive national related expenses. Since 1987, NPSAS has been administered every three to four years by the NCES division of the U.S. Department of Education. The study represents the most inclusive nationally representative survey of student financing of higher education in the United States (Wei et al., 2009). Data collected from the NPSAS survey are designed to help research ers and policy analysts answer questions about the affordability of American higher education and understand how various sou rces of financial aid impact students NPSAS:08 was selected as the data source for this study for several reasons. First, I was in recent graduates No other existing data source provided the number of detailed variables related to student loan borrowing for this particular student population than NP SAS:08. Secondly, it would have been extremely time intensive and expensive for me to develop and administer a survey instrument that could result in findings that were nationally representative. For these reasons, NPSAS:08 was the logical data source to u se in order to address the research questions guiding this study.
60 College s tudents attending all types and levels of institutions are represented in the NPSAS surveys including public and private not for profit and for profit institutions, community co lleges, and four year colleges and universities. For NPSAS:08, approximately 114,000 undergraduates and 14,000 graduate students were statistically selected from more than 1,600 postsecondary institutions. These students represent about 21 million undergra duates and 3 million graduate students enrolled in American higher education between July 1, 2007 and June 30, 2008 (Wei et al., 2009). Therefore, findings from the present study can be generalized to reflect the probability of borrowing and levels of stud ent loan indebtedness among 2007 08 Like most NCES postsecondary datasets, the NPSAS studies are available through public access and restricted use (Hahs Vaughn, 2007). Public access data is available online through a software application known as the Data Analysis System (DAS). This application allows users to generate descriptive tables and estimate covariance analyses from the NCES datasets. However, DAS does not provide the user with a complete list of variables from the NCES dataset of interest, nor does it allow for sophisticated manipulation of these variables. The restricted use data files for the NCES datasets allow users to perform more detailed analysis of the data. These files are available only through a restricted data license because of NCES confidentiality legislation (Hahs Vaughn, 2007). Access to the complete restricted data file for NPSAS:08 was required to answer the research questions proposed by the present study. Therefore, I submitted the applic ation for the NPSAS:08 restricted data license to the NCES in July 2009 and received the restricted data file in December 2009. Sampling Design Most national datasets utilize a complex survey design because they employ multistage, cluster, and/or stra tified sampling strategies (Hahs Vaughn, 2007). Simple random sampling
61 methods used by many small scale empirical research studies ensure each subject in the population has an equal probability of being included in the sample. In contrast, complex sampling designs used by many national datasets typically oversample subjects and institutions with particular characteristics of interest so they will have a higher probability of selection. Oversampling ensures these subjects and institutions are included in the sample in sufficient numbers for the purposes of statistical analysis (Thomas & Heck, 2001). The sample for NPSAS:08 was established using a two stage sampling design (Wei et al., 2009). The first stage identified the institutions that would be included in the sample from approximately 3,200 eligible United States postsecondary institutions The institutional sampling frame was constructed from the IPEDS:2004 05 and 2005 06 Institutional Characteristics, Fall Enrollment, and Completion files. A total of 1 ,730 colleges and universities participated in NPSAS:08. The second stage of the sampling design involved selecting students from these 1,730 institutions. Sample institutions provided student enrollment lists that were used to construct the student sampli ng frame. In total, approximately 127,700 students participated in NPSAS:08. Detailed information about the sampling design is available in the NPSAS:08 methodology report. Data Sample The sample used for the present study consisted of NPSAS:08 undergrad uates who earned 2008 academic year ( n = 2 5 671 ). This sample coincides with the NPSAS:08 variable named COLLGRAD, which used institutional records and student interviews to determine if the student completed his/her academic year. This sample was selected because I was specifically interested in the decision to borrow and level of student loan and not necessarily the debt held by un dergraduates who had not yet earned their four year degree
62 After applying the NPSAS:08 survey weight variable named WTA000 descriptive results revealed the sample used in this study represents 2,194,518 students who earned degrees fr om U.S. postsecondary institutions during the 2007 08 academic year. However, conducting statistical analysis using such a large number of cases poses greater risk for committing Type I errors ( Thomas & Heck, 2001 ) Therefore, a normalized weight was calcu lated by dividing the raw weight variable (i.e. WTA000) by the mean weight for the selected sample Using a normalized or relative weight when analyzing large, secondary datasets is a common practice among researchers and h elps correct for oversampling in the survey design ( Hahs Vaughn, 2005; Thomas & Heck, 2001). The normalized weight reduces the overall sample size for statistical purposes but still preserves the appropriate proportions of the complex survey design. In th is study, t he normalized weight was used to conduct all statistical analysis. The following tables provide descriptive information about the sample using several key demographic variables As depicted in Table 3 1, g reater than half (56.8%) of Amer 08 bachelor With regards to race/ethnicity (see Table 3 2), t he majority of all graduates were White (68.5%) and the next largest groups were Hispanic/Latino (10.7%), Black (10.1%) and Asian (7.3%) O nly these fou r racial/ethnic groups were included in the analysis because of the difficulty in conducting statistical analysis with relatively small numbers of students belonging to the other racial/ethnic groups (i.e. American Indian, Native Hawaiian, more than one ra ce). Table 3 1 Frequency of gender (n=25,671) Gender Frequency Percent Male 11099 43.2 Female 14572 56.8
63 Table 3 2 Frequency of race/ethnicity (n=24,779) Race/ethnicity Frequency Percent White 17577 68.5 Black or African American 2590 10 .1 Hispanic or Latino 2737 10.7 Asian 1874 7.3 Table 3 3 provides information about the dependency status for the students in the sample. The greatest percentage of graduates were considered dependents (58.3%) for fe deral financial aid need analysis purposes. Students who were considered independent (i.e. not reliant upon their parents for financial support) and did not have dependents of their own represented 24.9% of all graduates. The smallest percent of graduates (16.8%) were independent and did have dependents. For the purposes of the NPSAS:08, spouses of independent students were not considered as dependents. Table 3 3 Frequency of dependency status (n=25,671) Dependency status Frequency Percent Dependen t 14997 58.3 Independent without dependents 6388 24.9 Independent with dependents 4306 16.8 I also examined the types of postsecondary institutions students in the sample graduated from during the 200 7 08 academic year (see Table 3 4). The greatest num ber of students earned year college or university (60.6%), or a private not for profit college or university (27.5%). A smaller percentage of students (3.3%) earned their degree from a private for profit instituti one institution during the 2007 08 academic year were omitted from data analysis. Furthermore, graduates from postsecondary institutions classified as public two year were excluded from this study beca use of the small number of students belonging to this group. While not included in this study, these particular graduates likely reflect the growing trend by community colleges across
64 the country to offer baccalaureate degree programs in selected academic disciplines (Floyd & Walker, 2009; McKinney & Morris, 2010). Table 3 4 Frequency of institution sector (n=23, 457 ) Institution Sector Frequency Percent Public 4 year 15552 60.6 Private not for profit 4 year 7070 27.5 Private for profit 835 3.3 Dependent Variables T wo dependent variables were used in this study to examine ( a) borrow, or not borrow, through any type or combination of student loans (i.e. federal, private, state, and/or institutional) and ( b) the total amount of student loan debt borrowers accrued in Table 3 5 provides a summary of the se two dependent variables. Table 3 5 Summary of dependent variables NPSAS Source Variable Variable Type Scale Range Borrowed throu gh any type of student loan CUMLNTP1 Dichotomous Recoded: 0 =No, 1 =Yes Amount still owed on all undergraduate student loans OWEAMT1 Continuous $1 47 $150,000 The NPSAS:08 variable named CUMLNTP1 and labeled loan type for was used to address the first research question guiding this study. Originally, this variable indicated whether the student had borrowed (a) federal loans only (b) non federal loans, (c) federal and non federal loans or (d) no loans ever This variable was recoded into a dichotomous (a) borrowed through any type of student loan (i.e. federal, private /non federal or both) or (b) no student loans ever I recoded this variable to determine whether or not the graduates in the sample had ever borrowed any type o r combination of student loans This dependent variable was used to examin e the factors associated with probability of borrowing through federal, private state, and institutional loans. Table 3 5 provides frequencies
65 for this variable and indica tes that approximately two thirds (65.2%) of 08 ed through student loans in route to earning their degree. Table 3 5 Frequency of borrowing among 2007 08 (n= 25,671) Student Loan Borrowing F requency Percent Borrowed student loans 16740 65.2 Never borrowed student loans 8931 34.8 The second s in the sample. This continuous measure is denoted by the v ariable name OWEA MT 1 in the NPSAS:08 dataset. T he variable is based on the higher of an estimate derived from student interviews, the National Student Loan Data System, cumulative federal loan amount outstanding plus private loans borrowed in 2007 08, o r the amount borrowed in 2007 08 from any source. This dependent variable allowed for identification of the cumulative level of federal and non federal loan debt still owed by 2007 and was used to address the second research question guiding this study The sample used to examine the second research question was delimited to graduates who had borrowed through student loans and still owed money on their loans at the time of their graduation (n=16,333) Table 3 6 provides desc riptive information for this dependent variable and indicates the average outstanding amount of student loan debt among 2007 degree graduates who borrowed was $2 3 243 Table 3 6 Descriptives for amount still owed on all undergraduate lo ans (n=16,333) Minimum Maximum Median Mean SD Amount still owed on all undergraduate loans 147 150000 20000.00 23242.81 16727.16 Independent Variables T he conceptual framework was used to select the independent variables to be included during statis tical analysis. The independent variables were organized into categories based upon
66 Table 3 7 summarizes each of the independent variables used in this study. Table 3 7 Summary of indep endent variables and indices Items Variable Type Scale Gender Dichotomous dummy Male (reference), Female Race/ethnicity Categorical dummy White (reference), Black or African American, Hispanic or Latino, Asian Dependency status Categorical dummy Depend ent (reference) Independent without dependents, Independent with dependents English as primary language Dichotomous dummy Recoded: Yes (reference) No Categorical dummy Recoded: High school diploma or less, Bach Advanced degree (reference) Discussed financial aid options with parents or family Dichotomous dummy Yes (reference), No Discussed financial aid options with financial aid counselor or staff Dichotomous dummy Yes (reference), No ACT derived com posite score Categorical dummy Recoded: 18 or lower, 19 25 26 or higher (reference) College grade point average Categorical dummy Recoded: Less than 3.0, 3.0 or higher (reference) Family income Categorical dummy Recoded: $18,787 or less (Low) $18,788 to $51,043 (Mid Low) $51,044 to $93,643 (Mid High) $93,644 or more (High, reference) Unmet financial need after all grants Categorical dummy Recoded: No unmet need (reference) $1 to $3,084 (Low) $3,085 to $6,983 (Mid Low) $6,984 to $12,861 (Mid High) $12,862 or higher (High) Field of study/ academic major Categorical dummy Recoded: Humanities, Social/behavioral sciences, Life/physical sciences, Math/computer science/engineering, Education, Business/management (reference) Health, Vocational Tech Hig hest level of education ever expected Categorical dummy Doctorate or first professional (reference) Institution type Categorical dummy Recoded: Public 4 year (reference), Private not for profit 4 year, Private for pr ofit 4 year High school type Categorical dummy Public (reference), Private, Attended a foreign high school Institution degree of urbanization Categorical dummy Recoded: City (reference), Suburb, Town, Rural Institution region Categorical dummy Recoded: New England/Mid East (reference), Great Lakes/Plains, Southeast, Southwest, Rocky Mountains/Far West
67 Figure 3 1. NPSAS:08 variables depicted within the conceptual model Figure 3 1 depicts how the independent variables used in this study fit within th e layers of the conceptual model. Existing research literature on student loan borrowing and indebtedness provided the rationale for selecting the independent variables used in this study. In the following subsections, t he independent variables are present ed and operationally defined based upon the
68 contextual layers of the conceptual model guiding this study. Most of the independent variables included in this study were transformed into dummy variables for the purposes of statistical analysis. For the dummy coded variables, the reference group is indicated in parentheses. Layer 1: The Student and Family Context towards borrowing and their cumulative level of loan ind ebtedness are major components of the benefits of borrowing through student loans to pay for college. The following sections addressing Layer 1 of the conceptual demographic characteristics, social and cultural capital, demand for higher education and supply of financial resources, and expected benefits and costs of pursuing higher education. Demogr aphic Characteristics Numerous research studies and policy reports have examined the impact of gender and race/ethnicity on student borrowing behavior and loan debt burden. Several studies have found that females are more likely than males to acquire highe r levels of student loan debt (Kantrowitz, 2009; Price, 2004 a G differences in borrow and levels of student loan debt between male (reference group) and female 2007 08 uates. ace/ethnicity has been identified as a salient factor in predicting loan borrowing behavior among undergraduate s Research suggest s that Asian and Latino students are more averse to borrow ing through student loans than other racial/eth nic groups (Callender & Jackson, 2005; Cunningham & Santiago, 2008). Furthermore, researchers have found that African American and Hispanic borrowers are most likely to acquire higher levels of loan debt burden (Kantrowitz, 2009; King, 2005; King & Bannon, 2002; Price, 2004 a ). T he NPSAS:08
69 was used to examine the decision to borrow and loan debt levels among four racial/ethnic groups: White (reference group) Black, Latino, and Asian. Research also suggests that independent studen ts have higher levels of student loan debt than dependent students (Kantrowitz, 2009). Independent students are typically non traditional aged students and are more likely to have additional expenses (e.g. f amilies, mortgages, credit card debt ) than depend was included to examine differences in student loan borrowing and debt burden among three groups: dependents (reference group) independents without dependents, and independents with depe ndents. Social and c ultural c apital The amount of s ocial capital a student possess es can influence their borrowing behavior by providing him/her with information about the costs and benefits of borrowing through student loans, and by providing him/her wi th assistance in applying for financial aid. The NPSAS:08 dataset was limited with regards to measures that have been used in prior financial aid studies to represent social capital. However, the dataset did contain several a cquisition of financial aid information. Two of these variables were used as proxies for social capital in this study The two NPSAS:08 survey items related to social capital were responses stem ming from the following survey questio n: When you were making decisions about financial aid, which of the following did you do? : ( 1 ) discussed options with family and/or friends, ( 2 ) talked with a counselor or financial aid office staff in high school or college. T hese variables give particula r attention to the assistance students may have received from parents friends, or counselors during the financial aid process. Including these measures allowed me to examine they collected
70 through their social networks, influenced their probability of borrowing and cumulative level of student loan debt. ultural capital, as depicted in the conceptual model, is represented by cultural knowledge and th e value placed on higher education by the student and his/her family T wo NPSAS:08 variables were examined ng of financial aid processes (Singer & Paulson, 2004), the dichotomous NPSAS:08 variable that ome was also included. The reference group for this variable was students who indicated th at English was the primary language spoken in their homes. knowledge and values about college attainment (Perna & Titus, 2004). Research has found that education can influence the amount of loan money a student borrows in route to tal. This variable indicates the highest level of education achieved by either parent of the student and was recoded into three categories: al ; reference group ). d emand for h igher e ducation related decisions are often the A stu represents his/her initial stock of human capital (Perna, 2006a) and these factors can impact his/her demand for higher education. Applied to the present study, the conceptual model suggests emic preparation for college
71 and academic performance during college can influence their choice to borrow through student loans and a high er demand for postsecondary ed ucation may be more willing to borrowing in order to make attending and graduating from college a reality. may impact their decision to borrow through student loans and their cumulative le vel of indebtedness. Numerous higher education studies have used standardized test scores ( e.g. Perna & Titus, 2004 ) and grade point averages ( e.g. Flint, 1997 ; Harrast, 2004 ) as a measure of academic achiev ement Therefore, the NPSAS:08 variable ACT used in this study as one measure of academic achievement. In addition to using ACT scores for students who ACT co mposite score. The advantage of using this variable is that it allowed me to examine test scores for all of the students in the sample using a consistent measurement scale. This variable was recoded into three categories for the purposes of statistical ana lysis: 18 or less, 19 25, and 26 or above ( reference group). In addition college grade point average study as a measure of academic achievement Existing research suggests that lower grade point averag es in college are associated with higher levels of loan debt (Harrast, 2004) and greater risk of student loan default (Christman, 2000; Volkwein & Szelest, 1995) among undergraduate borrowers. This continuous NPSAS:08 variable was recoded into the followin g two categories based upon a 4.0 grading scale: less than 3.0, and 3.0 or higher ( reference group) Supply of r esources amount of money he/she borrow s in order to pay for college related expenses. Students with
72 limited financial resources are expected to rely more heavily on student loans than their more affluent peers. The conceptual model suggests that both family income and the amount of financial aid received ar financial resources. Family income can exert a strong influence on the total amount of debt an undergraduate corroborated that students from lower income families are more likely to acquire unmanageable levels of loan debt than students from middle or high income families (Dynarski, 1994; Price 2004 a as measure of the in this study This variable provides the AGI for the parents of dependent students, and the AGI of independent students (and their spouses if applicable). T his continuous variable was recoded into quartiles for t he p urposes of statistical analysis: $18,787.99 or less (Low), $18,787 to $51,043.99 (Mid low), $51,044 to $93,644.36 (Mid high), and $93,644.37 and above (High, reference group). The amount of financial aid a student receives can significantly influence t heir borrowing behavior through student loans In particular, researchers have found that students with higher level s of unmet financial need after all grant awards are taken into account ar e more likely to borrow through loans (Cunningham & Santiago, 2008 ). Therefore, t he NPSAS:08 continuous included in this study This variable measures the remaining cost of attendance (i.e. tuition and non tuition related expenses) during the 2007 08 academic year student s i (i.e. need based grants, merit based scholarships, employer tuition reimbursements, and private scholarships). T his variable was recoded into the following predetermined quartiles based upon an earlier descriptive analysis of this NPSAS measure : $0 (No unmet need reference group ), $1
73 to $3 ,084.99 (Low), $3,085 to $6,983.75 (Mid low), $6,983.76 to $12,861.50 (Mid high), and $12,861.51 and above (High ). Expected b enefits As d epicted in the conceptual assessment of the benefits and costs of borrowing to pay for college. T wo NPSAS:08 variables were examined efits of using student loans. Students pursuing higher earning majors may expect greater benefits from their investment in higher education ( Flint, 1997; Harrast, 2004; Price, 2004 a ) and thus be more willing to borrow through student loans in order to fin Therefore was recoded into the following eight categories for the pu rposes of statistical analysis: humanities, social/behavioral sciences, life/physical sciences, math/computer sc ience/engineering, education, business/management (reference group) health, and vocational/technical/professional. Research also suggests that perceived benefits of investments in higher educati on ( Perna, 2006a ) Undergraduates who expect to earn a graduate or professional degree in the future may be more willing to acquire student loan debt in order to achieve their long term academic goals. Accordingly, the three categories college. prof essional degree (reference group) Expected c osts Existing research has shown that the total cost of attendance can influence borrowing behavior (Cunningham & Santiago, 2008; Kantrowitz, 2009). Numerous studies and
74 policy reports have exam ined student borrowing as a function of the type of postsecondary institution attended. Findings suggest average loan debt levels are typically higher for students who graduate from private and for profit institutions (College Board, 2008; Kantrowitz, 2009 ; Steele & Baum, 2009). Therefore, a three was included to examine student loan borrowing among students graduating from the following institution types during the 2007 08 academic year: public four year (referenc e group), private not for profit four year, and private for profit four year. Layer 2: School and Community Context Layer 2 of the conceptual model gives particular attention to Research has shown that high school characteri loans to pay for col lege (Perna, 2008). The NPSAS:08 dataset is limited with regards to variables that specifically examined the availability of resources and the types of structural supports and barriers t hat exist within the hig h schools However, the dataset does contain a key variable that indicates variable was included during data analysis and addressed whether the studen t graduated from a public (reference group) private, or foreign high school. Layer 3: Higher Education Context The higher education context of the model examines how the characteristics of the colleges and universities students attend can impact their b orrowing behavior The Project on Student Debt ( December 2009 ) found differences in loan debt levels for students attending higher education in different geographic locations across the country. To examin e the potential influence of geogra phic location on student borrowing behavior (Project on Student Debt, December 2009) the NPSAS:08 variable s were included in this study.
75 to examin e student loan borrowing as a function four different categories: city (reference group), suburb, town, and rural. In addition, t categories for the purposes of statistical analysis: New England/Mideast (reference group) Great Lakes/Plains, Southeast, Southwest, and Rocky Mountains/Far West. Layer 4: Social, Economic, and Policy Context The fourth layer of the conceptual model gives attention to the economic and public policy characteristics that may i nfluence borrowing behavior among college students. As described in Chapter 2, the m s negative portrayal of student borrowing, changes in student loan policies, and economic downturn s can all potentially d their cumulative level of debt burden However, a limitation of NPSAS:08 with regards to the present study is that the dataset does not contain variables that directly address this contextual layer of the model. The present study, therefore, focuse d more specifically on concepts within Layers 1, 2, and 3 of the model may be associated with student loan borrowing and debt levels Methodological Considerations Before describing the statistical techniques that were used during data analysis, there are two methodological considerations pertaining to the research design and focus of the present study that deserve attention. Existing research corroborates the importance of addressing each of these methodological challenges (Cellini, 2008; Dowd, 2008; Hahs Vaughn, 2007; Thomas & Heck, 2001) The following sub sections address these methodological considerations in greater detail. Analyzing S econdary D ata sets R esearchers have documented the methodological challenges inherent in using large scale, sec ondary datasets like the one being utilized for this study ( Hahs Vaughn, 2007; Thomas & Heck, 2001 ; Thomas, Heck, & Bauer, 2005 ). When using NCES restricted data files such as
76 NPSAS:08, it is imperative for researchers to understand design effects and wei ghts in order to ensure the data is analyzed correctly (Hahs Vaughn, 2007). Since the complex survey design of NPSAS:08 oversampled some cases (i.e. institutions, students) with particular characteristics, researchers need to account for the homogeneous cl usters that may result from this type of sampling. Ignoring this complex sampling design lead s to (Thomas & Hec k, 2001 p. 53 4 ). Researchers have advocated for the use of complex sample survey software to overcome this particular methodological challenge ( Hahs Vaughn, 2005; Thomas & Heck, 2001) Accordingly, SPSS 18.0 Complex Samples software was utilized for the purposes of th is study. This relatively new add on to basic SPSS software is specifically designed to produce correct estimates for data collected t hrough complex sampling methods, such as data collected through NPSAS and many other NCES and National Science Foundation (NSF) surveys. This software was chosen over other existing programs (e.g. AM, a free software from the American Institutes for Research) because of my familiarity with SPSS. The Complex Samples add on was used to conduct all of the data analyses for this study. Since the NCES datasets do not use simple random sampling techniques weights and design effects are used to ensure the data is representative of the national population (Hahs Vaughn, 2007). Weights and design effects are applied during data analy sis to deemphasize the disproportionate contribution of the cases or subjects that were oversampled in the complex sampling design (Thomas & Heck, 2001). Failure to apply the weights and design effects ignores the complex survey design and will result in a nalyses that reflect the sample, but not the larger population of interest. Analyzing NCES data without using the weights can also violate the
77 independence assumption and generate parameter estimates that are biased (Thomas & Heck, 2001). The weights prov ided in NPSAS:08 restricted data file are intended to compensate for the unequal probability of selection of postsecondary institutions and students that comprise the final sample used for the survey. In addition, these weights and design effects adjust fo r multiplicity at the institutional and student levels, unknown student eligibility, nonresponse, and poststratification (Wei et al., 2009). The NPSAS:08 methodology report provide d detailed information about the specific purpose and uses of each weight in cluded within the dataset. T he methodology report and existing research studies (see Hahs Vaughn, 2005 & 2007; Thomas & Heck, 2001) were utilized to ensure the normalized weight for the sample was calculated correctly and applied accurately during data ana lysis A s mentioned previously in this chapter, this normalized weight was used to conduct all of the statistical analysis for this study. Estimating Causal Effects with Secondary Data In recent years, several researchers have documented the statistical challenges associated with accurately estimating causal effects in higher education and financial aid research using secondary data ( Chen, 2008; Cellini, 2008; Dowd, 2008; Titus, 2007). Like most social science and educational research, higher education s tudies cannot typically employ experimental designs or randomized trials because of logistical, ethical, political, and economic reasons (Titus, 2007). Consequently, most higher education researchers rely heavily on secondary data and researchers must be a ware of the statistical challenges and limitations inherent in using these types of data to claim causality (Titus, 2007). The primary statistical problem plaguing higher education research including financial aid studies, is self selection bias ( Cellini 2008; Dowd, 2008; T itus, 2007). Self selection means that
78 variable that is correlated with the outcome of interest (Millimet, 2001). Failure to control for this correlation during statistical analysis is problematic because it can result in model misspecification and inaccurate coefficient estimates (Titus, 2007). For higher education researchers using large, secondary datasets (e.g. NCES and NSF datasets), omit ted variable bias is a statistical concern that can result in self selection bias. Omitted variable bias occurs when an independent variable that might influence the outcome variable is not included in the regression model (Cellini, 2008). While NCES and o ther secondary datasets may offer a wealth of variables, one criticism of these datasets is they may not provide sufficient information about complex constructs to allow researchers to disentangle the effect of student characteristics that led them to make certain choices about college (Dowd, 2008; Perna & Titus, 2005). Without accounting for these variables in the regression model, omitted variable bias may result in unreliable estimates because one or more independent variables in the regression model are correlated with the error term. Proxy variables were utilized t o reduce omitted var iable bias in the present study The use of proxy variables represents one of the simplest and most common methods for dampening omitted variable bias in higher education research (Cellini, 2008). A proxy variable is a measurable or observable variable that is used in place of a variable that is unobservable. In order to serve as a useful proxy, the variable must be irrelevant in explaining the outcome variable once the oth er independent variables in the model have been controlled for and it must be sufficiently closely related to the omitted variable (Cellini, 2008). Examples of proxy variables that have been used in higher education research include test scores as proxies for ability and postsecondary plans as proxies for aspirations (St. John, 1990).
79 In the present study, the NPSAS:08 dataset provided several variables to sufficiently measure most of the concepts embedded within the four contextual layers of the conceptua l model. Many of the demographic and institutional variables used in this study have been examined in existing research studies of student financial aid, including some studies of student loan borrowing However, the NPSAS:08 dataset does not contain varia bles that have been used in prior studies to specifically measure the complex construct of social capital described within Layer 1 of the model Applied to the present study, social capital refers to the information a student possesses about financial aid and the level of assistance they receive from their social networks (e.g. parents, friends, high school and college counselors ) regarding financial aid. After reviewing all of the available variables in the N PSAS:08 dataset, two dichotomous measures were selected as reasonable proxies for the purposes of this study These variables represent acquisition of financial aid information and the su pport they received from family, friends, and /or counselors when m aking decisions about financial aid T his proxy variable captures the major elements of social capital as des Wooldridge (2002) suggests, even a reasonable imperfect proxy can help reduce omitted variable bias. By inc luding th ese proxy variable s during data analysis, it was possible level of social capital plays in their probability of borrowing a nd cumulative level of student loan debt. Analytic Methods This section provides an overvie w of the analytic methods that were performed to address the researc h questions guiding this quantitative study. Two stages of data analysis, preliminary and advanced, were conducted to generate the results presented in the following chapter.
80 Preliminary Data Analysis The preliminary analysis included descriptive statistics, t tests, and ANOVAs for student level and institutional level data. Descriptive statistics were utilized to examine the frequencies of students demographic characteristics and the typ es of postsecondary institutions graduated from during the 2007 08 academic year. In addition, descriptive analysis was used to examine the frequencies of the two dependent measures in this study. These descriptive s helped identify the percentage of all graduate s who borrowed through student loans and the average level of student loan debt held by 2007 Chi square s t test s and analysis of variance s (ANOVA) were performed to examine mean differences in the two dependent variables with regards to key demographic and institutional characteristics Specifically chi square tests were used to examine the decision to borrow through student loans (i.e. yes or no) by the following categorical independent va riables: gender, race/ethnicity, dependency status, family income, and institution type. T tests and ANOVAs were used to examine mean differences between these same independent variables the continuous dependent measure indicating the amount borrowers stil l owed on their student loans. Statistically significant findings from the chi square s t tests, and ANOVAs provided justification for conducting more advanced statistical procedures. Advanced Data Analysis A dvanced data analysis included one logis tic multivariate regression and one linear multivariate regression Specifically, logistic multiple regression was used to address the first resear ch question guiding this study. This research question sought to identify the factors that influenced student Since the dependent variable of interest in this research question was dichotomous and there were multiple independent variables, logistic regression was the most
81 appropriate statistical technique (Shavelson, 1996). Logistic regression allowed for an examination of multiple individual and institutional level factors that can influence probability of borrowing through student loans L inear m ultiple regression was used to address the second research question guiding this study. The continuous nature of the dependent variable of interest in this question justified the use of linear regression techniques (Shavelson, 1996). The linear regression model examined the individual and institutional level factors related to the cumulative amount of student loan debt among 2007 Table 3 8 Summary of multivariate regression models Independent Variables Bloc k 1 Block 2 Block 3 Block 4 Block 5 Block 6 Block 7 Demographic Characteristics Gender X X X X X X X Race/ethnicity X X X X X X X Dependency status X X X X X X X Social and Cultural Capital X X X X X X English as primar y language X X X X X X Discussed F.A. with parents X X X X X X Discussed F.A. with counselor X X X X X X Demand for Higher Education ACT derived composite score X X X X X College GPA X X X X X Supply of Resources Family income X X X X Unmet need after all grant aid X X X X Expected Benefits Field of study/major X X X Highest level of ed ucation exp X X X Expected Costs Postsecondary institution type X X Institutional Characteristics High school type X Degree of urbanization X Institution region X The independent variables used in this study were organized into seven blocks and added in succession to a baseline model to be regressed against each of the dependent variables (see Tabl e 3 8 ) The advantage of utilizing a hierarchical blocked entry approach is that it allows
82 researchers to group related categories of independent variables based upon a theoretical rationale. T he seven odel of college choice (2006a) used as a conceptual framework in this study to examine the choice to borrow and level of student loan indebtedness among 2007 The baseline model (Block 1) for the multivariate regressions cond ucted in this study was: 0 + 1 (gender) + 2 (race/ethnicity) + 3 n + e 1 In this model, Y represents a single dependent variable (i.e. borrowed or did not borrow; cumulative level of student loan debt) the coefficient, and e the constant or error term. Regression Diagnostics Regression diagnostics were performed i n order to determine the adequacy of the regression models used in this study First, I examined the distribution of the continuous dependent variable (i.e. cumulative amount of loan d ebt) used to address the second rese arch question. As anticipated, the distribution indicated a positively skewed curve since a small, but $ 75,000 or more) leve ls of loan debt Common statistical methods used to address issues of non homogeneity of variance involve transforming the data by applying a mathematic modification such as the square root, log, or inverse (Osborne, 2002). After applying each of these tra nsformations using SPSS, the output indicated that the square root function resulted in a normal distribution of this dependent variable. However, the transformation procedure fundamentally changed the nature of the variable and made interpretation of the results more complex (Osborne, 2002). In particular, after performing the square root modification the estimated coefficients for this dependent variable were no longer substantively interpretable in terms of dollar amounts. That is, the results of this r egression using the transformed data could only be discussed in terms of debt levels that were
83 either higher or lower than the designated reference group for each independent variable. I believed presenting the results in this manner would reduce the overa ll contribution of this study to the existing research literature on student loan borrowing. To address this issue, I ran the linear multivariate regression model both with and without applying the transformation to the dependent variable. Examination of the results indicated that the level of statistical significance for the independent variables was consistent across both models. That is, the same variables that were statistically significant in the model that applied the data transformation remained sig nificant in the model that did not apply the square root transformation. Therefore, I made the decision to present and discuss the coefficients from the regression model in ter ms of dollars is more meaningful for policymakers and researchers. Next, c ollinearity statistics were examined to determine if any of the independent variables analyzed in this study were highly correlated Multicollinearity occurs when two or more indep endent variables are highly correlated with one another and can lead to unstable estim ated coefficients (Chen, 2008). In regression the variance inflation factor (VIF) can be used to determine if the degree of correlation among the independent variables i s problematic (Myers & Well, 2003). Larger VIF levels often signify problems with multicollinearity in the regression model and a commonly recommended threshold is a VIF of 10. In this study, the VIF level s of the independent variables in both regression m odels (i.e. the logistic and linear) were well below this threshold. Once these diagnostic procedures were performed, I examined the overall efficacy of the final logistic and linear regression models used in this study. In particular, t he F tests wer e
84 reviewed to determine the goodness of it for each regression model. Statistically significant F values corroborated the efficacy of both the logistic and linear regression models In recent years, researchers have encouraged the inclusion of interactio n terms in studies of student financial aid (Chen, 2008; Dowd, 2008). An interaction effect occurs when levels of a particular variable change in relation to levels of a second variable (Huck, 2004). Failure to account for this interaction can result in re For example, Dowd (2008) suggests that differences in student loan borrowing behavior may exist by race/ethnicit y and family income was explored for the purposes of this study. These terms were borrowing and level of indebtedness. When analyzed, analysis of the change in R re vealed the interaction terms did not improve the efficacy of either model For that reason, interaction terms were not included in the final regression models that generated the results presented in the following chapter. Limitations of the Study There are several limitations to the present study that deserve attention Like many of the NCES datasets, the data represented in NPSAS:08 are cross sectional and not longitudinal The cross sectional nature of the dataset did not allow me to examine trends in student borrowing across different time points T he findings presented in this study therefore represent borrowing behavior and level s of student loan debt among 2007 08 degree graduates Some r esearchers have argued that many large, secondar y datasets often do not contain (Dowd, 2008) For example, personal characteristics such as self efficacy and locus of control may influence a
85 s amount of loan debt the student is willing to accrue in route to e the NPSAS:08 dataset and thus were not analyzed in this study. Furthermore, as highlighted earlier in this chapter, the NPSAS:08 dataset does not contain variables that specifically address ed Layer 4 (i.e. social, economic, and policy context) of the conceptual model. Another limitation of this study is related t o the self reported nature of several variables used during data analysis. While most of the variables used in this study were derived from institutional records, several measures relied solely on interviews from students who participated in the NPSAS:08 s urvey. Self reported data can lead to statistical bias because students may not be able to accurately recall their past behaviors or experiences, or they may respond to survey items in ways they believe are socially desirable. In this study, the variables used to represent were self reported. Therefore, these NPSAS:08 measures cannot be interpreted with the same degree of confidence as the other variables in t his study that were derived from institutional data and records. Contributions of this Study The present study represents a valuable contribution to the extant research literature on college student indebtedness in at least two ways. First, Pern allowed for the examination of many key v ariables (e.g. high school type social/cultural capital ) that have typically been absent from existing studies of college student indebtedness. The identification of these variables can result in a better understanding o f the factors associated with
86 Second, this study utilized the most recent nationally representative data to identify the characteristics of students who acquired the highest levels of loan d ebt in route to earning their of acquiring excessive loan debt, policymakers and higher education leaders can begin to develop strategies aimed at reducing deb t burden among this student population. Effective public policies and institutional practices that limit the number of college graduates with excessive loan grad uation, and increase degree attainment across the country.
87 CHAPTER 4 DATA A NALYSIS AND R ESULTS The objective of this chapter is to present the results from the data analyses that were conducted to address the two research questions guiding this study Results from chi squares, t tests, and analysis of variance ( ANOVA s ) are presented in the preliminary data analysis section. T he advanced data analysis section presents results from the logistic and linear multivariate regression models This chapter con cludes with a section that highlights findings from the data analyses. Preliminary Data Analysis C hi squares, t tests, and ANOVAs were performed to determine if the differences between five key independent variables were statistically significant for e ach of the dependent variables. The five status, family income, and postsecondary institution type. The results from these statistical tests are presented in the follow two su bsections. Each subsection specifically address one of the two dependent variables used in this study. Dependent Variable One: The Probability of Borrowing through Student Loans Chi square is a nonparametric statistical procedure that is used to test hypo theses about whether two or more categorical variables are independent of one another (Myers & Well, 2003). Since the five independent variables and the dependent variable used in these preliminary analyses were categorical, chi square tests were used to e xamine expected outcomes (i.e. the probability of borrowing) across each level of the independent measures. R esults for each of the five chi square tests are presented in Table 4 1. In addition, Table 4 2 provides frequencies and percentages for this depen dent variable across each level of the five independent variables.
88 Table 4 1 Chi square tests for the probability of borrowing through student loans Independent measure Pearson chi square df p value Gender 80.84 1 .000 *** Race/ethnicity 363.69 3 .000 *** Dependency status 328.02 2 .000*** Family income 1092.24 3 .000*** Institution sector 532.36 2 .000*** Table 4 2 Frequencies and percentages for the probability of borrowing through student loans Independent m easure Frequency % Did not borrow % Borrowed Gender Male 11,099 37.9 62.1 Female 14,571 32.5 67.5 Race/ethnicity White 17,578 36.1 63.9 Black 2,590 20.0 80.0 Latino 2,737 33.8 66.2 Asian 1,873 45.9 54.1 Dependency status Dependent 14, 978 39.3 60.7 Independent with dependents 6,388 28.3 71.7 Independent without dependents 4,306 28.6 71.4 Family income Low income 6,416 25.8 74.2 Mid low income 6,417 27.7 72.3 Mid h igh income 6,419 34.8 65.2 High income 6,417 50.8 49.2 Institut ion sector Public four year 15,553 38.5 61.5 Private not for profit four year 7,070 25.7 32.5 Private for profit four year 834 4.1 95.9 The first chi square analysis examined expected outcomes between male and female 2007 ates regarding the probability of borrow ing through student l oans. The results indicate d that female students were significantly more likely than males to borrow through student loans Furthermore 68% of females borrowed through student loans The second chi square analysis examin ed expected outcomes race/ethnicity and the probabil ity of borrowing Among 2007 Black students (80%) were most likely to borrow through students, followed by Latinos (66%),
89 Whites (64%), and Asians (54%) The chi square results for this test were statistically significant (p differences in borrowing between two or more of the racial/ethnicity groups were not due to chance. Next, a chi square analysis was performed to examine the probability of borrow ing by s. The res ults of this test were In addition, i ndependent students without dependents (72%) and independent students with dependents (71%) were considerably more likely to borrow through student loans than dependent st udents (61%). A chi square analysis was per formed to examine 2007 probability of borrow ing through student loans as a function of family income. The results were comparison of percent ages for the four income classifications indicate d that the probability of borrowing through student loans decreases as family income increases. Approximately 74% of graduates from low income families borrowed as compared to 49% of graduates from high inco me families. The final chi square analysis examined expected outcomes by postsecondary institution type The results of this test were statistically significant (p through student loans was highest for students attending private for profit institutions (96%), followed by private not for profits (71%) and public four years (62%). Dependent Variable Two : Levels of Student Loan Debt among Borrowers An independent samples t test was used to examine cumulative levels of student loan debt by gender among 2007 statistical t est are presented in Table 4 3 and indicate that female students are more likely to acquire higher l for females was $23,878 and males had an average debt level of $22,340. The mean difference
90 indicates that on average female 2007 hrough student loans accrued $1,538 more in loan debt than their male counterparts. Table 4 3 t test of mean cumulative level of student loan debt (n=16,333) df t Mean difference Std. Error difference p value Males vs. females 16331 5.793 1538.22 26 5.53 .000 *** A one way ANOVA was used to examine if there were statistically significant mean (i.e. White, Black, Latino, Asian) and their cumulative level of student loan debt. In particu lar, this analysis involved a multiple comparisons The re sults of the ANOVA (see Table 4 4 .001). Tabl e 4 4 (n=15,730) Sum of squares df Mean square F p value Level of student loan debt Between groups 4.742 3 1.581 57.404 .000*** Within groups 4.330 15725 2.754 Total 4.378 15728 With regards to the post hoc comparisons between each racial/ethnic group (see Table 4 5) for all groups except for the comparison between Latino and Asian students. Among 2007 degree graduates who borrowed, Black students had the highest average level of student loan debt ($26,438) and Asian students had the lowest average level of debt ($19,486).
91 Table 4 5 hoc tests of mean differences for levels of student loan debt by (n=15,730) (I) Race/ethnicity (group mean) (J) Race/ethnicity Mean difference (I J) Std. error White (23 384.18) Black 3,054.23*** 399.18 Latino 2,805.15*** 425.99 Asian 3,898.18*** 554.86 Black (26 438.41) White 3,054.23*** 399.18 Latino 5,859.39*** 538.96 Asian 6,952.42*** 645.66 Latino (20 579.02) White 2,805.15*** 425.99 Black 5,859.39 *** 538.96 Asian 1,093.03 662.57 Asian (19 485.99) White 3,898.18 *** 554.86 Black 6,952.42 *** 645.66 Latino 1,093.03 662.57 The results of the one way ANOVA that examined differences in student loan debt levels with dependents) are presented in Table 4 6 The overall model indicates there are statistically Table 4 6 (n=16,333) Sum of squares df Mean s quare F p value Level of student loan debt Between groups 7.512 2 3.756 136.461 .000*** Within groups 4.494 16329 2.752 Total 4.570 16331 Post hoc comparisons reveal that independent students, both with out an d with dependents, have higher levels of student loan debt on average than dependent students (see Table 4 7) The mean difference between independent students without dependents and independent students with dependents was not significant. Overall, depend ent students had an average student loan debt of $20,945, while the average debt levels among independent students without dependents ($25,327) and with dependents ($25,953) were significantly higher.
92 Table 4 7 s for levels of student loan debt by (n=16,333) (I) Dependency status (group mean) (J) Dependency status Mean difference (I J) Std. error Dependent (21,292.02) Independent with dependent s 4,034.49*** 304.77 Independent witho ut dependent s 4,660.83*** 350.67 Independent without dependents (25,326.51) Dependent 4,034.49*** 392.42 Independent with dependent s 626.34 305.24 Independent with dependents (25,952.85) Dependent 4,660.83*** 346.08 Independent without dependent s 6 26.34 387.55 Next, a one way ANOVA was conducted to determine if between group differences low, mid high, high) with regards to their cumulative level of student loan debt. The results of this ANOVA are presented in Table 4 8 .001). Table 4 8 (n=16,33 3) Sum of squares df Mean square F p value Level of student loan debt Between groups 1.284 3 4.281 15.338 .000*** Within groups 4.557 16328 2.791 Total 4.570 16331 Mean differences between each of the four i ncome classifications show that on average the cumulative level of student loan debt increases as family income decreases (see Table 4 9) T he post hoc comparisons reveal that students in the highest income quartile had significantly lower levels of debt w hen compared to students belongi mid low and mid .01) quartiles. In addition, students in the low income quartile had higher average levels of loan debt than students in the mid high quartile.
93 Table 4 9 c tests of mean differences for levels of student loan debt by (n=16,333) (I) Family income status (group mean) (J) Family income status Mean difference (I J) Std. error Lowest Quartile (24,162.16) Mid Low Quartile 604.50 34 8.06 Mid High Quartile 1,084.43 358.68 Highest Quartile 2,569.28 *** 388.52 Mid Low Quartile (23,557.67) Lowest Quartile 604.50 348.06 Mid High Quartile 479.93 360.78 Highest Quartile 1,964.78 *** 390.46 Mid High Quartile (23,077.73) Lowest Quart ile 1,084.43 358.68 Mid Low Quartile 479.93 360.78 Highest Quartile 1,484.85 ** 399.96 Highest Quartile ( 21,592.89 ) Lowest Quartile 2,569.27 *** 388.52 Mid Low Quartile 1,964.78 *** 390.46 Mid High Quartile 1,484.85 ** 399.96 The final one way ANOVA examined differences in cumulative levels of student loan debt by the type of postsecondary institution (i.e. public four year, private non profit four year private for profit four year ) the 2007 08 academic year. The results of see Table 4 10 ) indicate significant mean differences by institutional type. Table 4 10 ANOVA for level of student loan debt by type of postsecondary institution (n=14,988) Sum of squares df Mean square F p value Level of student loan debt Between groups 2.366 2 1.183 443.545 .000*** Within groups 3.996 14985 2.667 Total 4.233 Furthermore, a ll three between group post hoc comparisons w ere statistically significant (p ; see Table 4 11 ). The average cumulative level of student loan debt was significantly different between public four year ($20,442), private non profit four year ($27,441), and private for profit four year ($33,121) i nstitutions.
94 Table 4 11 of postsecondary institution (n=14,988) (I) Institution type (group mean) (J) Institution type Mean difference (I J) Std. error Public four y ear (20,442.31) Private non profit four year 6998.32*** 288.20 Private for profit four year 12678.72*** 603.13 Private non profit four year (27,440.63) Public four year 6998.32*** 624.01 Private for profit four year 5680.40*** 315.75 Private for p rofit four year (33,121.03) Public four year 12678.72*** 571.52 Private non profit four year 5680.40*** 617.38 Examination of the results from the preliminary data analysis suggest there are differences in loan borrow status, and postsecondary institution type. Statistically significant findings from the chi squares, t tests, and ANOVAs provided the rationale for conducting for more sophist icated statistical procedures. These statistical procedures are discussed in the subsequent section. Advanced Data Analysis This section will present results from the advanced data analysis used to address the two research questions guiding this study. F irst, a binary logistic multivariate regression model was multivariate regression model was used to examine the level of student loan indebtedness among the 2007 0 are discussed separately in the following subsections. Logistic Multivariate Regression: The Probability of Borrowing through Student Loans A logistic multivariate re gression was used to address the first research question guiding this study. The dependent variable of interest for this question was whether or not the student had ever borrowed through any type of student loan (i.e. federal, private, state, institutional ) in route The overall model was significant (p the decision to borrow through student loans correctly for 7 1% of the students (see Table 4 12 ).
95 The results from the logistic regression model will be discussed in terms of the final block, which controlled for all independent variab les used in this study (see Table 4 13 ) Coefficients are discussed in terms of the expected which were converted into percentage s (using the exponential function) for ease of interpretation Table 4 12 B inary logistic multivariate regression mo del measures for the probability of borrowing through student loans (n=25,671) Chi square df Sig. Nagelkerke R Square % predicted correctly Probability of borrowing 4644.320 47 .000 *** .22 5 7 0.8 % Table 4 13 Results for binary logistic multivariate regression model for the probability of borrowing through student loans Variable name (reference group) Beta Exp(B) Std. Error Wald Demographic Characteristics Fema le (male) .143** 1.154 .050 8.116 Black (W hite) .490*** 1.632 .096 25.862 Latino (White) .016 1.016 .094 .030 Asian (White) .032 1.033 .102 .100 Independent with out dependents (dependent) .216** 1.241 .084 6.638 Independent with dependents (dependent) .115 1.122 .086 1.802 Social and Cultural Capital (advanced degree) .370*** 1.448 .057 42.795 (advanced degree) .206*** 1.229 .061 11.534 English as primary languag e (yes) .461*** .631 .094 24.179 Discussed financial aid with parents (yes) .462*** .630 .055 71.007 Discussed financial aid with financial aid counselor or staff (yes) .744*** .475 .049 234.301 Demand for Higher Education ACT score 18 or less (2 6 or higher ) .331*** 1.392 .085 14.996 ACT score 19 to 25 (26 or higher) .288*** 1.334 .062 21.678 College GPA less than 3.0 (3.0 or higher) .183*** 1.200 .051 12.740 Supply of Resources Low family income ( high income ) .467*** 1.595 .101 21. 376 Mid low family income (high income) .419*** 1.521 .082 26.241 Mid high family income (high income) .337*** 1.400 .074 20.794 Low u nmet need after grants (no need) .632*** 1.882 .075 71.618 Mid low unmet need after grants (no need) .650*** 1.916 .08 0 66.438
96 Table 4 13. Continued Mid high unmet need after grants (no need) .824*** 2.279 .078 111.780 High unmet need after grants (no need) .437*** 1.548 .090 23.321 Expected Benefits Humanities (business/ management) .042 1.042 .079 .27 7 Social/behavioral science (business/management) .105 1.110 .081 1.661 Life/Physical Science (business/management) .127 1.135 .105 1.475 Math/Computer Science/Engineering (business/management) .002 .998 .089 .001 Education (business/management) .028 1.028 .092 .091 Health (business/management) .111 1.117 .100 1.217 Vocational/Technical/Professional (business/management) .100 1.105 .088 1.290 .150* .861 .069 4.722 ee (advanced degree) .010 .990 .063 .024 Expected Costs Attended private non for profit institution (public four year) .399*** 1.490 .078 25.794 Attended private for profit institution (public four year) 2.374*** 10.736 .250 90.221 Institutional Characteristics Attended a private high school (public) .269 *** .764 .064 17.521 Attended a foreign high school (public) 1.218 *** .296 .134 82.257 Institution location suburb (city) .080 1.084 .076 1.109 Institution location town (city) .222 ** 1.2 48 .081 7.518 Institution location rural (city) .097 1.101 .171 .320 Institution region Great Lakes/Plains (New England/Mideast) .309 *** 1.362 .088 12.398 Institution region Southeast (New England/Mideast) .095 .910 .088 1.152 Institution region South west (New England/Mideast) .207 .813 .128 2.620 Institution region Rocky Mountains/Far West (New England/Mideast) .337 *** .714 .101 11.018 The first block of variables examined the probability that students would borrow through student loans as a function of the following demographic characteristics: gender, race/ethnicity, and dependency status. Controllin g for all other variables, female students were 15% more likely to borrow through student loans than male students. There were no statistically significant differences in the probability of borrow for Latino and Asian students when compa red to White students. H owever, Black students were 63% more likely than
97 s without dependents were 24% more likely to borrow than dependent students. The second block In this study, social capital was measured by two variables that indicated whethe r the student discussed their financial aid options with their parents, and/or with a financial aid counselor. Controlling for all other variables, the results indicated that students who discussed financial aid with their parents ( = ) and with financial aid counselors ( = ) were significantly less likely to borrow through student loans. Cultural capital was measured in this study by the level of education attained by the en language. Compared to students whose were 23% students whose parents had earned re 45% more likely to borrow through loans than students whose parents had advanced degrees Students for whom English was not the primary language spoken at home were 36% less likely (= to borrow than students who wer e native English speakers. The third block of variables gave attention to students demand for higher education as measured by their academic achievement. S cores on college entrance exams (i.e. an ACT/SAT derived score) and college grade point average were achievement in this study Compared to students whose test scores were in the top quartile (i.e. 26 or above), students scoring between 19 and 25 were 33% more likely to borrow (= .001) and students scoring 18 or lower were 39% more likely to borrow through student loans In addition, c ollege GPA was also significant ly associated with the
98 probability of borrowing. Students whose GPA was less than 3.0 we re 20% more likely (= .183, to borrow than students with a 3.0 GPA or higher. of financial resources. Compared to students in the highest family income qua rtile (i.e. $93,644 or higher) students belonging to the three other income quartiles (i.e. low, mid low, mid high) were significantly more like ly to borrowing through student loans T he lower the s of family income the more likely he/she wa s to borrow and students from the lowest income quartile (i.e. $18,788 or less) were 60% more likely ( income students remaining unmet financial need after all grant aid was also used a measure of his/her supply of financial resources. In comparison to students with no remaining unmet need, students with any amount of unmet need (i.e. low, mid low, mid high, high) were significantly more lik ely to borrow through student loans. Students in the category of mid high unmet need (i.e. $6,894 to $12,862) had the greatest probability of borrowing and were 1 27% more likely (= .824 the expected benefits of borrowing to achieve their academic goals These variables in and the high est lev el of education expected Compared to business/management majors, none of the other seven classifications of academic majors yielded statistically significant results regarding the probability of borrowing In terms of educational expectations, students th at 14% less likely (= .150 to borrow than students who expected to earn their doctorat e or first professional degree. The likelihood of compared to those who expected to earn their doctorate or first professional.
99 borrowing. The type of postsecondary institution students attended strongly influenced their probability of borrowing through stude nt loans Students at private not for profit institutions were 50% more likely (= .399 to borrow than students attending public four year institutions. Furthermore, students attending private for profit institutions were 974% more likely to borrow than students attending public four year institutions. The sev enth and final block of variables entered into the regression model examined characteristics The variables included in this block were high school type, postsecondary institution degree of urbanizat ion, and postsecondary institution geographic region of the United States. Compared to students who attended public high schools, students who attended private high schools were 24% less likely (= .269 to borrow and students who attended foreign high schools were 70% less likely (= to borrow through student loans. Compared to students who attended postsecondary institutions located in cities, students who attended instituti ons located in towns were 25% more likely to borrow through student loans. In addition, results indicated several statistically significant results with regards to borrowing across different geographic regions of the United States. Compa red to students attending postsecondary institutions located in the New England/Mideast region, students in the Great Lakes/Plains region were 36% Mountains/Far West region were 30% le ss likely (= loans. Linear Multivariate Regression: Levels of Student Loan Debt among Borrowers Linear multivariate regression was used to address the second research question guiding this study. The dependent v ariable of interest was the cumulative level of student loan debt
100 among 2007 study explained 12% of the total variance in the level of student loan indebtedness ( df= 16,29 2 ; F= 53.32 ; p .001 ) and six of the seven block entries resulted in a significant change in the model ( see Table 4 14 ). The results from the overall regression model will be discussed in terms of the final block, which controls for all independent variables used in thi s study The unstandardized beta coefficients are reported and discussed since they represent dollar amount differences in loan debt levels among student borrowers. The first category of variables entered into the model examined three demographic cha variables, female students were significantly more likely than White students to have hi gher levels of student loan debt and on average had the highest level of loan debt among all racial/ethnic groups examined in this study. Compared to White students, both Latino (= borrowed w ere more likely to have lower levels of student loan debt. Asian students had the dependents (= was measured in this study by two variables that indicated whether the student discussed financial aid with his/her family, and with a financial aid counselor. Mean differences in loan
101 Table 4 14 Uns ta ndardized beta coefficients for blocked entry regression on the cumulative level of student loan debt (n=16,33 3 ) Variable name (reference group) Block 1 Block 2 Block 3 Block 4 Block 5 Block 6 Block 7 Intercept 20 799.63*** 20 849.84*** 20 774.54*** 18 23 4.61*** 18 374.08*** 15 668.40*** 18 114.16*** Demographic Characteristics Female (male) 1 436.29*** 1 259.12** 1 225.18** 1 362.67*** 1 099.05** 901.43* 835.43* Black (White) 2 176.71*** 1 816.83** 1 654.95** 1 660.28* 1 724.14* 1 643.42* 1 856. 80** Latino (White) 3 344.12*** 3 002.66*** 3 076.86*** 2 847.51*** 2 832.78*** 3 119.61*** 3 232.65*** Asian (White) 3 761.69** 2 795.09* 2 774.30* 3 000.27** 2 892.33** 2 701.36** 2 271.39* Independent w/ no dep (dependent) 4 168.66*** 3 768.90*** 3 570.19*** 5 722.44*** 5 709.50*** 5 247.85*** 5 547.19*** Independent with dep (dependent) 4 250.64*** 3 589.11*** 3 257.64*** 3 775.16*** 3 804.24*** 3 131.30*** 3 537.29*** Social and Cultural Capital ed < (ad vanced) 1 190.83*** 1 935.15*** 2 031.22*** 2 055.55*** 2 177.53*** 2 037.40*** (advanced) 753.70 740.23 700.98 737.82 1021.31 968.42 English as primary language (yes) 2 067.18** 2 130.70** 2 128.10** 2 088.28** 1 993.81* 1 614.15* Discussed Fin Aid with parents (yes) 604.22 543.14 594.83 649.41 409.25 399.61 Discussed Fin Aid with financial aid counselor or staff (yes) 1 929.29 *** 1 913.32*** 1 542.83*** 1 529.73*** 1 047.02* 1 194.61** Demand for Higher Education ACT score 18 or less (26 or higher) 687.51 1 041.33 1 044.53 1 012.75 1 348.94* ACT score 19 to 25 (26 or higher) 47.51 620.92 610.67 974.99 1 141.96* College GPA < 3.0 (3.0 or higher) 8.74 485.57 595.51 1 162.41* 1 113.23* Supply of Resources Low income (high income) 6 522.86*** 6 601.01*** 4 363.70*** 4 098.89*** Mid low income (high income) 4 768.68*** 4 789.39*** 3 311.31*** 3 084.80*** Mid high income (high income) 2 158.39*** 2 177.75 *** 1 264.34* 1 072.45 Low unmet need (no need) 2 189.60*** 2 190.62*** 1 296.40* 1 271.34* Mid low unmet need (no need) 3 324.25*** 3 290.59*** 2 357.75*** 2 246.05*** Mid high unmet need (no need) 6 424.31*** 6 371.49*** 4 902.57*** 4 715.16*** High unmet need (no need) 11 430.88*** 11 392.63*** 7 875.81*** 7 537.53*** Expected Benefits Humanities (business/ management) 930.14 1 408.22 1 323.44
102 Table 4 14. Continued Social/behavioral science (business/management) 523.34 1 418.99 1 427.40* Life/Physical Science (business/management) 43.83 871.37 892.04 Math/Computer Science/Engineering (business/management) 447.26 221.96 242.74 Education (business/management) 1 306.97 2 334. 63 ** 2 061.34** Health (business/management) 1 309.28 2 078.05* 1 888.28* Vocational/Technical/Professional (business/management) 92.94 854.58 664.34 (advanced degree) 346.47 558.20 580.63 Expected to degree) 632.49 599.39 623.63 Expected Costs Private not for profit (public four year) 5 259.08*** 4 871.42*** Private for profit (public four year) 7 858.95*** 7 626.18*** Institutional Characte ristics Private high school (public) 327.57 Foreign high school (public) 4 091.58** Institution location suburb (city) 806.26 Institution location town (city) 740.14 Institution location rural (city) 2 834.44* Great Lakes/Plains (New England/Mideast) 1 128.43* Southeast (New England/Mideast) 3 298.78*** Southwest (New England/Mid East) 1 264.06 Rocky Mtns/Far West (New England/Mid East) 4 122.02*** R .027 .03 5 .035 .08 5 .087 .106 .116 Change in R .027*** .007*** .000 .051*** .001** .020*** .010*** F 86.21*** 56.10*** 43.63*** 74.68*** 52.71*** 62.24*** 53.32***
103 options with their family. However, students who talked with a financ ial aid counselor were more likely (= to have lower levels of student loan debt, on average, than students who did not consult with a counselor. In this study, cultural capital was measured whether was English had significantly higher levels of debt if their parents had earned a 1) In addition, students for whom English was not the primary language had lower average levels of student loan debt than native English speakers (= cated by their academic performance and achievement. test scores (i.e. ACT/SAT derived composite scale) and college grade point averages were used as one measure of academic achievement. Compared to students who scored highly on the ACT (i.e. 26 or above), students who scored 18 or less (= 1348.94, p (= 1141.96, p on the test were likely to have higher levels of student loan debt. college GPA students who had less than 3.0 were more likely to have higher levels of loan debt than studen ts who had earned a 3.0 GPA or higher. resources. Net all independent variables, as the level of family income decreased so too did In comparison to students in the highest quartile of family income, the lower average loan debt levels of students from the low (= 4098.89 p and mid low remaining unmet financial need after receiving all grant aid was used as a second measure of an
104 Statistically with greater levels of unmet need had higher levels of student loan debt than students with no unmet need. Furthermore, the average amount of loan debt a student owed increased as the level of unmet need increas ed. assessment of the benefits of borrowing through student loans These variables in cluded highest level of education ever expec ted When compared to business/management majors, students earning degrees in the following areas were more likely to have higher levels of student loan debt: social/behavioral science education and health (= 18 Results indicated there were no statistically significant differences in loan debt level ighest level of education expected. The sixth block of variables added to the regression model examined ent of the costs of borrowing through loans. With regards to postsecondary institution type, students wh o attended private institutions not for profit and private for profit (= had higher levels of loan debt than st udents who attended public four year institutions. Students attending private for profit institutions had the highest average of student loan debt after controlling for all other independent variables. The seventh and final block examined the following ins titutional characteristics: high school type, postsecondary institution degree of urbanization, and postsecondary institution geographic region of the United States. Students who attended a foreign high school had lower average levels of loan debt (= 409 than students who attended public high schools. The differences in debt levels between students who attended public and private hig h schools
105 were not significant. Compared with students attending postsecondary institutions located in cities, students attending institutions in rural areas (= had lower levels of student loan debt. In addition, students attending postsecondary institutions in the Great Lakes/Plains (= Southeast (= and Ro cky Mountains/Far West (= regions of the United States had lower levels of loan debt on average compared to students attending institutions in the New England/Mideast region. Summary of Results This chapter presented results fro m the preliminary and advanced statistical analyses used to address the two research questions guiding this study. Preliminary data analysis included c hi squares, t tests, and ANOVAs of the two dependent variables as a function of five independent variable s Statistically significant results suggest ed that the probability of borrowing through student loans and the total level of loan debt among borrowers differ ed gender, race/ethnicity, dependency status, family income, and type o f postsecondary institution attended. Overall, results from the preliminary data analysis suggest that females, Blacks, independent students, students with greater unmet financial need and students attending private institutions more likely to borrow and have higher levels of student loan debt. These findings provided the rationale for conducting more advanced statistical procedures. The results from one logistic multivariate regression model and one linear multivariate regression model were presented in the advanced data analysis section. Overall interpretati on of the regression social and cultural capital, supply of financial resources, and postsecondary institution type were strongly associate d useful framework for examining student loan borrowing and indebte dness. The follow chapter
106 will discuss the statistical findings from this study in greater detail and in relation to the existing research literature.
107 CHAPTER 5 D ISCUSSION The purpose of this chapter is to discuss and give context to the statistic al results presented in Chapter 4 First, the purpose statement and research questions guiding this study are revisited Next, t he findings from the previous chapter ar e discussed in relation to the existing research literature and policy reports on studen t loan borrowing and indebtedness. F indings are organized and discussed based upon the major constructs described in the conceptual model guiding this study The final section provides a summary of the highlights presented throughout the chapter. Purpose of the Study Revisited Numerous research studies and policy reports have indicated that more college students are borrowing more money than ever before through education loans in order to pay for college ( see Boushey, 2005; King, 2005; King & Bannon, 2002; Steele & Baum, 2009 ) In 2008, the average level of loan debt among was $23, 242 and approximately 15 % of these graduates had accumulated $40,000 more in loan debt (NPSAS, 2008) The dramatic spike in student lo an borrowing and debt burden has drawn widespread attention in the media and within the public policy arena (Gladieux & Perna, 2005) Furthermore, rising tuition rates and declining financial supp ort for postsecondary education from federal and state gover nments have the potential to exacerbate the student debt problem (Carey & Dillon, 2009). Collectively, these f actors corroborate the need for timely research on student loan borrowing The purpose of this study was t o examine the probability of borrow ing through student loans graduates. The latest national dataset related to s NPSAS:08 ) was used to address the t wo research questions guiding this study :
108 1. What individual, familial, and institutional factors are associated with 2007 degree gradu through student loans? 2. What individual, familial, and ins titutional factors are associated with the total level of student loan debt among 2007 Descriptive and multivariate analysis of the NPSAS:08 data allowed for the identification of individual, familial, and ins to borrowing through student loans and the cumulative level of loan indebtedness among student borrowers. The major findings from this empirical study are discussed more systematically in the following section. Discussion of Results In the following subsections, the findings from the data analyses will be discussed in relation to the existing research literature on student loan borrowing and indebtedness The discussion is organize d and presented based upon the major constructs described in the adapted These constructs include demographic characteristics, social and cultural capital, demand for high er education, supply of financial resources, expected benefits and costs of borrowing, and institutional characteristics. In addition to integrating aspects of the economic theory human capital model gives attention to key sociological factors th at may influence student loan borrowing behavior The model also recognizes that student decision making occurs within a situated context, and may be influenced by exogenous variables like institutional characteristics and public policy (2006a). Overall, P present study within the landscape of the extant research literature on student loan borrowing and indebtedness.
109 Demographic Characteristics bor rowing behavior and debt levels as a function of three demographic characteristics: F emale students were more likely than males to borrow through student loans and have higher levels of loan debt an d this finding held across all blocks for both regression models. In the borrowing behavior. In support of findings from this study, s everal researchers have fou nd that gender was significantly related to student loan borrowing and indebtedness (Kantrowitz, 2009) Conversely, other studies have found no statistically significant differences in the loan borrowing behavior and debt levels of female and male college students ( Harrast, 2004 ). Existing studies that have examined the relationship between gender and student loan borrowing ha ve utilized different research designs, samples, and sources of data. These methodological differences across these s tudies may explain some the mixed findings in the extant research literature. However, it may also be possible that female and male students have different attitudes and/or perceptions regarding the use of student loans to pay for college. These are potent ial gender differences that may be worthy of exploration in future studies of student loan borrowing and indebtedness. this study found that Black students were the most likely to borrow and have the highest leve ls of debt. This confirmed the hypothesis proposed in Chapter 3. This finding is well aligned with existing studies that suggest Black students are at the greatest risk of experiencing financial hardships as a result borrowing through student loan s ( King & Bannon, 2002; Price, 2004a) In this study, the greater probability of borrowing and the higher levels of loan debt were evident for Black students after controlling for family income and level of unmet need after all grant aid These finding s suggest tha t Black college students may be
110 more willing than other racial/ethnic groups to utilize loans to help them achieve their higher Approximately 80% of all Black 2007 route to earning their degree This percentage was significantly higher than any other racial/ethnic group. Black students perceive the benefits of borrowing t hrough student loans to outweigh the costs. Discouragingly Bl ack students also continue to be overrepresented among students with $40,000 or more in loan debt (Kantrowitz, 2009) and among loan defaulters ( Flint, 1997 ). Since Black students comprised only 10% of all 2007 the high debt burden of this student population may be further exacerbating social inequality in American higher education (Price, 2004a). A particularly notable finding from this study is related to the borrowing behavior of Latino and Asian students. Several e xisting studies and reports have suggested these particular racial/ethnic groups may be loan averse and less likely to borrow through student loans than their peers ( Cunningham & Santiago, 2 008 ; EMC Foundation, 2003 ). The loan aversion of Latino and Asian students has been ascribed to held cultural values and generally negative perceptions towards the acquisition of financial debt (Singer & Paulson, 2004) In recent years, however, researcher s have highlighted the importance of examining other key variables (e.g. social/cultural capital, family income) before attributing differences in loan borrowing behavior Controlling for key s ociological and financial variables this study found that Latino and Asian were not significantly less likely to borrow through student loans than White students. However Latino and Asian graduates who borrow ed did ha ve
111 signif icantly lower levels of student loan debt than their White peers. These findings conflict with the hypotheses proposed in Chapter 3 of this study. W hile Latino and Asian students borrowed at rates comparable to other racial/ethnic groups, their cultural va lues and/or perceptions of debt may have influenced the total amount they borrowed through students loans in route to e implies that race/ethnicity is just one of numerous factors within the student and family context (i.e. Layer 1) that can and debt burden. The lower loan debt levels among Latino and Asian student borrowers in this study may support the notion tha t cultural and/or familial values impac t the borrowing behavior of these student populations (Cunningham & Santiago, 2008; EMC Foundation, 2003). However, there may also be a variety of additional factors (e.g. educational choices, psychological variables ) not available in NPSAS:08 that could h elp explain why these 2007 borrowed less than their peers. These findings, therefore, should be considered in conjunction with existing research literature that addresses the loan borrowing behavior of Latino and Asia n colleg e students. In future studies on this topic, qualitative research methods may help provide a better understanding of the more nuanced factors that shape the borrowing decisions of these racial/ethnic groups. tudy found that independent students were more likely to borrow and have higher levels of loan debt than dependent students. Similarly, existing studies have found than older students are more likely to have high levels of debt (Harrast, 2004; Kantrowitz, 2009). Dependent students are typically traditional college aged individuals who still rely on their parents or other family members for financial support to pay for college related expenses. Conversely, independent students are typically 25 years of age o r
112 older may have additional financial obligations beyond those typically held by dependent students. For example, many independent students have mortgages and children to raise and support. ncy status in future studies of student loan borrowing. their college related decisions, including their decision to attend college (Perna, 2006a). Findings from t his study corroborate that demographic variables also play a major role in college students how demographic characteristics are related to other key constructs t hat collectively shape many of the choices that students, and their families, make about attending and financing higher education. Social and Cultural Capital conceptual model to the student of college studen t decision making and behavior is the inclusion of sociological constructs, such as social and cultural capital. These constructs have been given limited attention in previous studies of student loan borrowing that have primarily relied on human capital th eory. E xisting literature suggest that s tudents utilize various forms of social capital, such as parents, teachers, counselors and peers, when making college related decisions ( McDonough, 1997; Perna, 2008; Tierney & Venegas, 2006 ). In this study, two NPSA S:08 variables were used to examine the extent to which 2007 about financial aid for college. These particular measures indicated whether students relied on their parents an d/or financial aid counselors when making financial aid decisions. This study found that 2007 discussed financial aid options with their parents or family members were significantly less likely to borrow through
113 student loans. Existing research suggests that parents can play an important role in helping students make decisions about college (McDonough, 1997). In this study, p arents of many students may have provided them with assistance in applying for grants or scholarsh ips, and therefore reduc ed the need for these students to borrow. related decisions, including choices about borrowing through student loans. Researchers have als o found that parents who have limited or negative experiences using Accordingly, many engaged parents in this study may have been hesitant for their child to use s tudent loans and advised their child to explore other alternatives. For example, s ome parents may have encouraged their child to attend a lower cost postsecondary institution (e.g. the local community college) where the need to borrow through student loans would not be as substantial Controlling for all other variables, in this study discussing financial aid options with a financial aid counselor or staff member was associated with a lower probability of borrowing and lower levels of loan debt among 2007 These discussions could Speaking with a counselor may have helped stude nts identify ways to pay for college other than borrowing through student l oans. Furthermore, a financial aid counselor(s) could have helped these students complete the required paperwork (i.e. FASFA, scholarship application) to become eligible for grant aid. Some student borrowers may have also utilized advice from financial aid counselor s to help limit their overall level of loan debt. Recently, several research studies have drawn attention to the ways i n which financial aid counseling influence s
114 financial aid (Dowd, 2008; McDonough & Calderone 2006 ). Therefore, the se particular findings from this study represent an important and timely contribution to the extant research literature. In this study, a s tudent s c ultural capital was measured by education and whether Engli sh was their primary spoken language. Existing research suggest student borrowing ( Kantrowitz, 2009; Perna, 2008 ) and findings from this study support those results. Students whose parents had earned a bachel were more likely to borrow and have higher levels of loan debt. Since level of educational attainment is strongly correlated with income ( Baum & Ma, 2007 ), it is proba ble that many parents with lower levels of education have lower salaries and therefore do not have the financial means to help their children pay for college. As a result, perhaps a growing number of students are using education loans in order to bridge th financial contribution Students for whom English was not the primary language were less likely to borrow and ha ve lower levels of loan debt than primary English sp eaking students. A significant number of the 2007 for who English is not the primary language may have been immigrants. Existing research suggest that language barriers and inexperience with existing financial institutions pr events many immigrants from using a variety of financial services (Singer & Paulson, 2004). Therefore many students for whom English was not the primary language may have made a conscious decision not to borrow, or to borrow as little as possible, because and/or were unable to navigate the complex loan process.
11 5 capital impacts their decision to enroll in c ollege (Ellwood & Kane, 2000; Perna & Titus, 2005). However, few studies have used these sociological constructs to explain differences in student loan borrowing among college students. Findings from this study highlight the importance of including measure s of social and cultural capital in future studies of student loan borrowing and indebtedness. (Perna 2006a) access to accurate information and support networks play a key role assessment of the costs and benefits of borrowing through student loans to pay for college. Demand for Higher Education A cademic performance was used in this study as a measure of demand for higher education rmance can shape their as sessment of the costs and benefits of borrowing through student loans (Perna, 2008). This study found that higher academic performance, as indicated by college entrance exam test scores and college GPA, was associated with a lower probability of borrowing and lower levels of student loan debt. Existing studies have found that students who perform better academically are likely to have lower levels of debt (Harrast, 2004) and are less likely to default on their student loans (Flint, 2007; Volkwein & Szelest, 1995; Volkwein et al., 1998). O ne explanation for the findings from this study is that students who perform poorly on college entrance test scores do not typically qualify for financial aid in the form of academic scholarships. F or example, many students who live in states with state merit based scholarship programs are not likely to be eligible for these financial awards if they have lower ACT/SAT test scores These students also may not qualify for academic scholarships made ava ilable through the postsecondary institutions they attend. R eceiving any academic scholarship would have
116 likely reduce d the total amount a student needed to borrow through loans in route to earning their For student loan borrowers, e ach semester enrolled in higher education provides another occasion to borrow more money. Existing research studies suggest that college GPA is strongly associated with degree completion and s tudents with higher GPAs are more likely to complete their bach (see Pascarella & Terenzini, 2005) Therefore, a student borrower with a low er college GPA may experience a delayed time to de gree and thus acquire greater levels of loan debt in order to persist until graduation. Existin associated with their decision to attend college (Ellwood & Kane, 2000; Perna & Titus, 2005). In this study, however, there was a minimal change in ance explained (i.e. R ) when the academic performance variables were added to both regression models. While higher academic performance resulted in a lower probability of borrowing and less loan debt in pears more salient in predicting college enrollment than student loan borrowing behavior. Supply of Resources largest increase in explained variance across both regression models Family income and unmet Findings suggest that 2007 financial resources play ed a central role in their loan bo rrowing behavior T hese findings also through student loans.
117 R esults support existing studies that indicate students fro m lower income families are more likely to borrow through student loans ( Christou & Haliassos, 2006; Cunningham & Santiago, 2008). In this study, the probability of borrowing increased as the le vel of family income decreased. Students from lower income fam ilies may be left with little choice but to use student help their child pay for college. T he fact that lower income students borrow the most frequently may fur ther highlight existing issues of social inequality in American higher education (Price, 2004a) Notably, when controlling for all other variables in this study, the cumulative amount of money lower income students borrowed through student loans was signif icantly less than students from higher income families. While they were the most likely to use student loans to pay for college, it is possible that receiving need based grant aid (e.g. Pell Grants, state and institutional scholarships) helped reduce the t otal amount lower income students needed to Controlling for all other variables, in this study level of unmet financial need after receiving all grant aid was strongly associated with their p robability of borrowing and their total level of loan debt. These findings are in agreement with existing research (Cunningham & and/or limit the amount they borr Many 2007 08 based financial aid, such as Pell Grants. However, findings from this study suggest that need based awards alone may not be sufficient enough to help many students graduate without acquiring high levels of student loan debt.
118 ancial resources. In study after study, findings indicate these variables are strongly associated with student loan borrowing behavior. These results accentuate the need to help lower income students identify strategies (e.g. apply for all available grant aid, attend lower becoming heavily indebted. Furthermore, these findings corroborate how current student financial aid system, which is predominately based on loans hinders those stu dents who possess fewer financial resources (Price, 2004a) aid. With regards to college choice, existing research suggest the type of financial aid received can inf attend (Heller, 1997; Perna & Titus, 2004). Consequently, examining student loan borrowing behavior and debt burden as a function of the specific type(s) of financial a id received (e.g. merit based, need based, work study) would represent an important contribution to the extant research literature. Expected Benefits and the cumulative amount they borrow may be infl uenced by the benefits they expect to receive from pursuing higher education Academic major was used a measure of expected benefits in this study s ince t he level of income a student expects to earn in the future is strongly influence by their career choic e There were no statistically significant differences in the probability of borrowing across the eight categories of academic majors examined in this study. students with degrees in the Social/Behavioral Sciences, Education, and Health related fields had
119 higher levels of loan debt. It is challenging to determine precisely why students pursuing majors within these particular fields have higher average levels of loan debt tha n their peers. However, since these career choices may potentially lead to public service careers with relatively low earning potential (e.g. elementary school teacher, social worker), it is important to understand why students in these academic majors bor Most e xisting research studies have found that academic major is not a particularly strong predictor of loan debt levels (Price, 2004a; Volkwein & Szelest, 1995; Volkwein et al., 1998). However, Harras t (2004) found that some majors (e.g. special education, computer engineering, sociology) did contribute to higher levels of debt among student borrowers. The inconclusive findings in the existing research literature signify that future studies should give greater attention to the relationship between academic m ajor and student loan borrowing. their academic major) can influence their student l oan borrowing behavior. In this study, t he highest level of education ever expected was not associated with different levels of loan debt among students who did borrow. However, 2007 ee or less were significantly less likely to borrow than students w ho expected to earn their doctorate or first professional degree T his finding may suggest that students who planned to attend graduate school expected they would have to borrow in order to achieve their long term academic goals and were willing to incur higher levels of student loan debt Conversely, 2007 expected to earn their may have been more committed to achieving this academic goal without the use of student loans.
120 expected benefits from pursuing higher education. Findings from the present study suggest that expected benefits may play a role in However, additional research on how the expected monetary and non monetary benefits of higher education influence student college r elated decisions (to include decisions about loan borrowing) is necessary to disentangle these relationships. Expected Costs This study used the type of postsecondary institution attended as a measure of the expected cost s associated with student loa n borrowing to pursue higher education As hypothesized this study found that 2007 08 bachelor' s degree graduates who attended private not for profit and private for profit institutions were more likely to borrow and have higher loan debt levels than stud ents attending public four year institutions. Many existing studies have also found that students attending private institutions are more likely to borrow and have higher loan debt levels (Boushey, 2005; Kantrowitz, 2009; Steele & Baum, 2009). Th ese result s are not surprising considering that private postsecondary instituti ons typically charge higher tuition rates than pu blic colleges and universities. Accordingly, students often borrow more through education loans to cover the higher costs of attending a p rivate institution. It should be noted that a private for profit institution is n early synonymous with borrowing for college Specifically 96% of all students who graduated from these post secondary institutions during the 2007 08 academic year had acquired student loan debt. The average level of indebtedness among these borrowers was around $33 ,000 which is approximately $10 ,000 more than the average debt levels of graduates from public po stsecondary institutions S tudents who are considering enrolling in
121 private for profits should be cognizant of the high loan debt levels accrued by many students who attended these proprietary schools This information about borrowing can be particularly v aluable as students, and their families, weigh the costs and benefits of attending a private for profit institution. familial, and institutional factors shape stude higher education. Expected college costs have been examined by many existing studies of college choice, as well as extant research on student loan borrowing. Collecti vely, findings from existing literature and from this study highlight the importance of understanding how expected costs influence the ways in which students and their families pay for college. Institutional Characteristics a varie ty of institutional characteristics can influence student decision making and behavior In this study, students who attended private or foreign high schools were less likely to borrow in route to earning their than student s attending publ ic high schools. Private high schools are typically more expensive than public high schools and the student body at private high schools tends to come from more affluent families. Therefore, it is plausible many students who graduated from private high sch ools do not need to borrow through student loans because their parents help pay for many of their college related expenses. However, the differences in debt levels among loan borrowers who attended private and public high schools were not statistically sig nificant. This study found that 2007 who attended a foreign high school were less likely to borrow than students who attended a public high school. A significant number of students are likely to have been immigrants. Therefo re, t his finding seems to support existing research that suggests immigrants are less likely to use many financial services because
122 (Singer & Paulson, 2004). Furthermore, the avera ge l evel of loan debt among student borrowers who attended a foreign high school was approximately $4,000 less than graduates of public high schools when controlling for all other variables in this study. Even when they do borrow, many immigrant students a ppear to be borrowing at more conservative levels than their non immigrant peers. T here were statistically significant differences in student loan borrowing as a result of the (i.e. city, suburb, to wn, rural) Surprisingly, s tudents attending institutions located in towns were more likely to borrow through loans than students attending institutions in cities This finding seems counterintuitive since the cos t of living is typically higher in cities t han in towns and therefore one would expect students attending college in cities to be more likely to borrow In addition, this study found that student borrowers attending institutions located in rural areas had significantly lower levels of loan debt th an students attending college in cities. This finding likely reflect s differences in the lower cost of living and average tuition rates in rural areas than in larger cities Few studies to date have of urbanization Findings from this study may indicate the need to examine this relationship in greater detail in future research. T he Project on Student Debt ( December, 2009 ) found that debt levels were highe st among students attending higher education in the Northeast region of the United States Conversely, the same report found the lowest debt levels for states in the West. In this study, student borrowing was examined as a function of five U.S. geographic regions The results indicated that 2007 08 likely to borrow than students in the New England/Mid East region. Furthermore, compared to
123 the New England/Mid East r egion, the average debt levels among borrowers were significantly less in all other geographic regions (i.e. Great Lakes/Plains, Southeast, Southwest, Rocky Mountains/Far West ) R esults from this study support findings from the P December, 2009) report. The report suggests differences in debt levels across geographic region may be attributed to several factors. First, the New England/Mid East have more students on average attending private postsecondary institutions, which likely e xplains the higher loan debt levels among graduates in this region. In addition, the report indicates that western states tend to have more students attending public institutions and many of these institutions charge tuition rates that are lower than the n ational average. This would explain why 2007 graduates in the Rocky Mountains/Far West region were the least likely to borrow through student loans. t related decisions. V ariables such as high school and postsecondary institution type, which are germane in studies of college choice, are also relevant when examining student loan borrowing behavior. between key student and family variables (i.e. demographics, social and cultural capital) and the ttend. Summary of Discussion This chapter provided a discussion of the results from this study in relation to the extant research literature on student loan borrowing and indebtedness. The discussion was organized around the major constructs describ ed in an conceptual framewo rk to guide this study. Use of this model not only generated findings that are
124 consistent with existing studies but it also provided new insights that have not been addressed in the extant research literature. Overall, the resu lts from this study support the applicability of emographic characteristics and supply of financial resources were most salient in identifying the factors associated with their decision to borrow and total level of student loan debt. Black students were the most likely to borrow among all racial/ethnic groups and had the highest average levels of loan debt Latino and Asian studen ts borrowed with the same probability as White students, but borrowed smaller amounts on average than all other racial/ethnic groups. In addition, females and independent students (both with and without dependents) were more likely to borrow and have highe r debt levels. circumstances are strongly associated with their borrowing behavior. The greatest increase in explained variance across both regression models occ urred with the addition of two finance variables: family income and unmet need after all grant aid. The probability of borrowing et need increased. contribution to the extant research literature on student loan borrowing. In this study, 2007 08 ial aid options with their parents and/or a financial aid counselor were less likely to borrow. Students who m et with a financial aid counselor also had lower levels of loan debt on average than of education and whether English was the primary spoken language were used as measures of
125 cultural capital and were significantly related to student loan borrowing. These findings help corroborate the inclusion of sociological perspectives in studies of student financial aid. Wh ile they explained a small portion of the overall variance in each regression model, associated with student loan borrowing. Students with lower ACT/SAT test scores and college GPAs are more likely to borrow and have higher debt levels. In addition, students in several academic majors (i.e. Social/Behavioral Sciences, Education, Health) that may lead naturally into lower paying public service careers were likely to have accumulated the most loan debt. A smaller probability of borrowing was also evident for students who expected to earn no higher These particular findings represent a new contribution to the extant research literature on student lo an borrowing. Findings from this study also corroborate the importance of examining institutional characteristics in studies of student loan borrowing. As expected, students who attended private (both not for profit and for profit) postsecondary i nstitutions were more likely to borrow and have higher debt levels. Students who attended private or foreign high schools were less likely to borrow than students who attended high schools classified as public. In addition, the geographic location of the p borrowing and the cumulative level of debt among borrowers. Few, if any, existing studies of student loan borrowing have examined the institutional variables analyzed in this study. The presented study utilized the latest national dataset related to student financial aid, NPSAS:08, to examine student loan borrowing and indebtedness among 2007 degree graduates. Considering the dramatic increases in student loan borrowing in recent years, this study represents a timely contribution to extant research literature. The subsequent, and final
126 chapter of this study will examine the implications of the se findings for policy and practice and also offer recommendations for future research related to student loan borrowing
127 CHAPTER 6 RECOMMENDATIONS AND FUTURE RESEARCH The final chapter of this study begins by providing recommendat ions for improving public policies and institutional practices related to student loan borrowing and indebtedness. Next, several recommendations are discussed t hat can help strengthen future research studies related to these topics. In closing this chapter provides some final thoughts about the current state of student loan borrowing and the rising leve ls of indebtedness college students There is a direct relationship between the affordability of American higher education and student loan borrowing. The rising costs associated with attending higher education have forced many stude During the last decade, tuition and fees that public four year universities rose at an average annual rate of 4.9%, which was considerably higher than in either of the past two d ecades (College Board, 2009). Therefore, it is apparent that finding ways to curtail the rapid increases in college costs represents an important strategy for reducing the number of college students who experience financial hardships as a result of their l oan debt. Federal and state policymakers must make continual and concerted efforts to ensure that American higher education is affordable. As students search for cost college baccalaureate may become an increasingly popular option for students and their families. A community college baccalaureate colleges or 2 year institutions that are approved to confer baccalaureate degrees in one or more A pproximately 30 community colleges across the country currently offer their own However a growing number of states have granted the two year institutions within their jurisdiction with the authorit y to offer four year degrees in
128 select disciplines ( Floyd & Walker, 200 9 ). Since tuition rates and fees are often considerably lower at community colleges than at public four year institutions, students pursuing their lege may need to borrow less through student loans in order to earn their degree. Consequently, in the future the community college baccalaureate may graduates. Recommendations for Policy and Practice F indings from this research study have numerous implications for policy and practice. In support of existing research, these findings emphasize the importance of providing students with the basic informati on and skills they need to make wise decisions about borrowing for college. Accordingly the first recommendation addresses literacy and money management skills. Findings from this study also suggest that financ ial aid counselors can play a major role Two recom mendations are provided that can aid high school and college financial aid counselors in becoming better resources for their students. Finally results from this study suggest the need students. Therefore, the last recommendation addresses issues surrounding student aid reform. Financial Liter acy Education degree graduates are borrowing more money through student loans than ever before. Nearly 1 5 % of all 2007 08 graduates who borrowed had accumulated $40,000 or more in loan debt. Rising tuition rates, dwindling federal and state fiscal support for higher education, and the declining purchasing power of Pell Grants have all undoubtedly contributed to the increase in student loan borrowing (Heller & Rodgers, 20 06) However, giving attention solely to these types of
129 exogenous factors may overlook a more fundamental issue. The extant research literature on student loan indebtedness has given less attention to the possibility that many students who borrow simply ma y not understand how to manage their money. Several e xisting studies indicate that m any students do not begin their college careers with even the basic financial management skills For example, r esearch suggest s that high school students understand very little about money and only one third of high school students reported feeling confident managing thei r own finances (ASEC, 1999). Many students are confused by interest rates on student loans and tend to underestimate the total amount of interest being a ccrued when they face an extended repayment period (Lewis & van Venrooji, 1995). Furthermore, Norvilitis and Santa Maria (2002) suggests that many students may overestimate their ability to repay all types of debt before they make the decision to assume de bt. This lack of financial literacy is particularly troubling when considered in light of research indicating the majority of college students have favorable attitudes regarding the use of credit (Lyons, 2008; Xiao et al., 1995). It is quite possible that many students who borrow do not fully understand the consequences of acquiring debt through student loans. management skills before they ever enter postsecondary educa tion. Unfortunately, there is not an established policy across states for youth about financial matters ( NICE, 1996). While financial literacy programs are delivered through the public education system s in some states, there are conside rable differences in the services offered and in the program r equirements (Norvilitis & Santa Maria 2002) In addition, one report found 30% of high school students indicated their parents had not discussed financial management strategies with them
130 (ASEC, 1999). These findings may suggest that when beginning their college careers, many students have received little or no guidance about how to manage their personal finances. A public policy requiring each state to provide financial literacy courses and/or programs improve their financial management skills Completing a financial literacy program could help expose to the foundational knowledge and skills they need t o make smarter financial decisions through their lifetime. Consequently, acquiring these financial management skills before attending college c ould help students and their families make better decisions about borrowing through student loans. Financial A id Counseling A growing body of literature suggests that college students and their parents are poorly informed about college costs and the accessibility of financial aid (Horn, Chen, & Chapman, 2003; Long, 2004; Perna, 2006b; Vargas, 2004). In addition r esearchers have found that many students have considerable misperceptions about using student loans and are likely to underestimate the total costs of their loan borrowing (King & Frishberg, 2001). These findings ing of the financial aid process and the ways in which student loans are often embedded within larger financial aid package s could help assuage college graduates High school, and college, financial aid coun selors represent an important resource for students seeking information about borrowing through students loans. Researchers have found that students and their parents view high school financial aid counselors as primary and reliable sources of information about college costs and financial aid (McDonough, 2004). However, in recent years studies have found that many high school financial aid counselors are relatively uninformed about college financial aid (McDonough, 2005). Discouragingly the least informed
131 counselors are often found in low income schools where the student body has very little knowledge about student loan programs (Perna, 2008). Many high school counselors are also hesitant and unsure of how to advise students regarding the amount to borrow t hrough student loans (Perna, 2008). Collectively, these findings suggest that many financial aid counselors need additional training and professional development in order to become reliable and helpful resources for students who have specific questions abo ut using student loans to pay for college. Loan counseling strategies When controlling for all other variables, this study found that 2007 or college financial aid counselor were less likely to borrow and had lower levels of loan debt. This may indicate that personal interactions with financial counselors can help students make better decisions about using loans to pay for college. Counselors may have assisted s tudents in applying for grant aid that eliminated their need to use loans, or helped them decide not to borrow the maximum loan amount each semester. Regardless, this finding further emphasizes the need for counselors to be informed and well trained about college financial aid and student loan policies Results from this study can provide high school and college financial aid counselors with valuable information about loan borrowing they can use to improve the services they provide for students For example this study identified several factors (e.g. race/ethnicity, level of unmet need, academic performance, postsecondary institution type) that are strongly associated with higher levels of student loan debt. Counselors can use this information to better und erstand those develop targeted strategies to help these borrowers. High school counselors may also explain to college bound students the considerable differences i n the levels of loan debt held by graduates
132 of different types of postsecondary institutions (e.g. public versus private). Promisingly, well informed and trained financial aid counselors have the potential to help many student borrowers reduce the total am Financial Aid Reform Findings from this study corroborate the influential role that family income and grant aid btedness. In particular, 2007 08 who received lower lev els of grant aid were more likely to have higher levels of loan debt than their peers. These findings are in agreement with numerous existing studies that suggest level of financial resources is strongly associated with their loan borrowing behavior ( Kantrowitz, 2009; King & Bannon, 2002; Perna, 2008; Price, 2004a ). Lower income students and those who receive little or no grant aid are often forced to borrow in orde r to cover the rising costs associated with pursuing higher education. Existing studies and reports have recommended that policymakers consider increasing the (King & Bannon, 2002; College Board, 2008). Many of these recommendations have focused specifically on increasing Pell Grant award amounts to help reduce the rising debt burden held by borrowers from lower income families. S ome studies have also called fo r a restructuring of the federal student loan programs that would result in cost savings to borrowers ( King & Bannon, 2002; Boushey, 2005). Promisingly, in March 2010 President Bara c k Obama signed into law a new student loan reform bill. The law gives atte ntion to the major recommendations offered in numerous existing studies by increasing the annual Pell Gr ant award amounts and introducing landmark changes to the way federal student loans are disbursed. This new law has considerable implications for futur e research on student loan borrowing and is therefore discussed in greater detail in the following section. However, here it is important
133 to discuss the implications of this new law in light of the findings from this study Consistent with existing researc h this study found that lower income students were more likely to borrow through student loans than high income students In addition, loan debt burden increased as that raising the maximum Pell Grant award amount has the potential to help assuage the cumulative amount that lower income students borrow through loans in route to e degree. In years past, other federal legislation has increased Pe ll Grant award amounts but these increases have failed to keep pace with inflation (Curs, Singell, & Waddell, 2007). An important element of the new law is that Pell Grants are scheduled to increase at the rate of inflation in the years to come. The maximu m Pell Grant award amount is expected to be $5,975 by 2017 (Baker & Herszenhorn, 2010). Larger Pell Grant awards that account for inflation costs can help reduce the level of unmet financial n eed among lower income students and thus reduce the amount they need to borrow through student loans. While this aspect of the new law is not a panacea, it does represent a positive step towards alleviating the high student loan debt burden among borrowers from lower income families The new law also included modifi cations to several relatively new federal student loan forgiveness programs. These programs, which are further discussed in the following section, are intended to help reduce the burden of repayment for college graduates who have outstanding student loan d ebt. Importantly, these programs have the potential to reduce financial hardships among borrowers after they graduate and may help lower s tudent loan default rates However, these programs appear to only address the symptoms of the student debt problem and not the root cause. While providing some degree of financial protection for college graduates who have large
134 amounts of federal student loan debt, these programs do little to help students from acquiring high debt burden in the first place. More preventat ive strategies, such as financial literacy education and financial aid counseling, are needed to help students make wise decisions about borrowing through out their college careers. Moving forward, f ederal and state lawmakers must recognize that s tudent financial aid policies can have a significant influence on student loan borrowing behavior and debt burden. Financial aid program eligibility requirements determine the types of college students who receive funding In addition the level of politic al and fiscal support for existing financial aid programs determines how much money for college is available for students and their families. Consequently, policymakers should closely monitor the impact of the new federal student financial aid policies to determine how they affect student loan borrowing and indebtedness. Recommendations for Future Research T he present study provides timely findings regarding student loan borrowing and debt ever, many im portant research issues surrounding the topic of student loan borrowing remain underexplored in the extant research literature The findings presented in this study illuminate four areas for future research on student borrowing and indebtednes s: federal student conceptual model, credit card debt, and methodological approaches. New Federal Student Financial Aid Policy At the time this research study was completed, President Barack Obama had recently signed a student loan reform bill into law that is certain to bring about significant changes to based system for distributing federal subsidized student loans and essentially removes private lenders from the federal student loan system (Basken, 2010). These changes to the federal student loan
135 programs are intended to help reduce the burden of repayment for college graduates who have outstanding loan debt. In addition, the new l aw will also increase Pell Grant award s at the rate of inflation and invest an additional $2 billion in community colleges over the next four years. Considering the ramifications this new law will have on student borrowing behavior, researchers who plan t o examine student loan borrowing and indebtedness in the future will certainly want to examine the impact of these new public policies. One recommendation for future research is to revisit the two research questions addressed in this study using the next N PSAS dataset when it becomes available in 2012. This methodological design would allow borrowers before and after the law took effect. Future research may also want to examine the ways in which increases in Pell Grant awards influenced student loan borrowing among low income students. In addition to this freshly minted law, two federal programs have been established within the last three years that may also signific antly impact student loan borrowing and indebtedness. These programs are the Income Based Repayment plan and the Public Service Loan Forgiveness program. Since they are both relatively new, researchers have not yet explored how these federal programs influ ence student borrowing. Therefore, future studies of student loan borrowing and debt burden should consider exploring the impact of these two federal initiatives. The Income Based Repayment (IBR) plan was established in July 2009 and is intended to offer financial protection for students who rely on federal loans (Steele & Baum, 2009). The discretionary income. Furthermore, the program ensures that any remaining federal s tudent loan debt will be forgiven after 25 years of qualifying repayments. Future research studies on student
136 loan debt should begin to examine whether this program increases or decreases the cumulative amount students borrow through the federal lending pr ograms. In particular, are some students borrowing more through federal loans because they view this program as a panacea to the problem of excessive debt burden? How does the IBR impact student loan default rates? The Public Service Loan Forgiveness ( PSLF) program was established in 2007 and repayment and eligible employment. In order for a student to qualify, he/she must be employed in some form of public servi ce career for 10 years, such as jobs in government and non profit organizations. The program is designed to remove student loan debt as a disincentive to pursuing a public service career. In future research, important questions for investigation surroundin g the PSLF program may include: Is there an increase in number of college graduates nationwide pursuing public service careers as a direct result of this program? What percentage of borrowers who intended to capitalize on the program reaches the 10 year em ployment requirement? Do students participating in this program borrow more through loans than their peers since they may expect a considerable portion of their loan debt to be forgiven? Conceptual Model Most existing research studies examining is sues related to student loan borrowing and indebtedness have relied heavily on human capital theory. However, within the last decade several researchers have described the limitations of relying exclusively on the rational human capital approach to examine college student decision making (Dowd, 2008; Perna, 2006a; St. college student choice was used as a conceptual framework to guide the present study. This relatively new conceptual model builds upon the strengths of human capital theory, while recognizing that student decision making occurs within a situated context and is influenced by
137 other exogenous factors (e.g. familial, institutional, and policy characteristics). Researchers are examine student loan borrowing and debt burden. Future studies on student loan borrowing should give particular attention to Layer 4 (i.e. the soc is a cross sectional dataset, it was not possible to directly address these variables in the present study. However, research has shown that changes to existing federal s tudent loan policies can significantly impact student borrowing behavior (see King, 1999) and the economic context (e.g. the total amount they borrow. Additiona lly, examining Layer 4 of the model may be particularly important in future studies considering the newly established federal student loan legislation and programs described above Findings from this research study corroborate the relevance of using model in future studies related to student loan borrowing. For example, this study found that attend (key constructs in mode l), can influence their borrowing behavior. Many results from this study also support existing research regarding the factors associated with higher levels of student loan debt among borrowers (see Boushey, 2005; Harrast, 2004; Kantrowitz, 2009; Price, 200 4a). Furthermore, Perna (2008) has demonstrated the applicability of the conceptual model to the study of student loan borrowing by using education loans to pay for college al model in future studies has the potential to provide a more complete understanding of the factors that influence student loan borrowing behavior and debt levels.
138 Credit Card Debt E burden has important implications for pol icy and practice, and can also help assuage financial hardships among student borrowers. However, research on student loan debt may not tell the complete picture overall level of indebtedness Debt accrued through education loan s represents only one source of debt for many esearch suggest that a growing number of students are relying on credit cards and there are concerns about those students who are acquiring burdensome levels of credit card debt wh ile enrolled in college (Hayhoe et al.; Lyons, 2004; Norvilitis et al., 2003) Therefore, f uture studies of student indebtedness should consider examining the total level of debt students owe from both education loans and credit cards The most recent dat a suggest the majority of undergraduates (84%) own at least once credit card and carry an average monthly balance of $3,173 (Sallie Mae, 2009). An overwhelming majority (92%) of undergraduate credit card holders paid for direct education expenses (i.e. tex tbooks, school supplies) with their cards and 30% used a credit card to pay tuition. Overall, the Sallie Mae report found record highs in the percentage of students using credit cards, the average number of cards they carry, and their average balance. Thes e latest data corroborate that Numerous studies and reports have examined the credit card behavior and debt levels of college students ( Grable & Joo, 2006; Lyons, 2008 ; N orvilitis et al., 2006; Norvilitis, Szablicki, & Wilson, 2002 ). However, a review of the extant research literature revealed there are no known empirical total level of debt from both student loans and credit cards. F uture re search that examines and credit card debt could provide valuable insights about the true financial hardships these borrowers face during repayment
139 Methodological Approaches In recent years, r esearchers have drawn attention to an array of methodological problems that plague existing studies of student financial aid ( see Cellini, 2008; Chen, 2008; Dowd, 2008). While every research design has inherent limitations, researchers who plan to conduct studies of student loan borrowing shou ld familiarize themselves with the methodological problems and proposed solutions identified in the research literature. Future research studies that provide an accurate assessment of student loan borrowing and indebtedness are essential to the development of effective public policies and institutional practices. A second methodological recommendation involves issues surrounding federal data collection. Existing research suggest s that c omplex constructs such as social and cultural capital may influence present study used a total of four NPSAS:08 variables related to these constructs and each of these variables did result in statisticall y significant findings. However, the NPSAS:08 dataset and cultural capital in great detail. Dowd (2008) has also suggested that many of the federal education datasets do not collect information on psychol ogical variables like self efficacy or recommendat ion from this study is that the federal education datasets (i.e. NCES and NSF) begin to incorporate additional measures of the se complex constructs in future surveys related to financial aid The availability of these variables in the NCES and NSF datasets would allow researchers to examine with greater precision how key n borrowing behavior. Most existing studies related to student loan borrowing have been conducted using quantitative methods and many have been conducted using national data However, q uantitative
140 studies that utilize federal education datasets represe nt only one approach to the study of loan borrowing and debt burden among college students. As St. John (2006) suggest s qualitative research methods can provide valuable insights about how decision making and behavior. E xisting qualitative studies have contributed to a better loan borrowing and perceptions of their debt burden ( Davies & Lea 1995; Perna, 2008 ). Therefore capable researchers are recommended to consider applying qualitative, or mixed, methods when examining student loan borrowing in future studies Closing Words S tudent loan borrowing is a dynamic and evolving issue in American higher education that deserves serious attent ion from resear chers and policy makers. There is a pressing need for high quality and timely research on this topic that will inform public policy and institutional practice This line of research can help ensure that for thousands of college students who graduated each y excessive amounts of student loan debt.
141 LIST OF REFERENCES Alexander, F. K. (2002). The federal government, direct financial aid, and community college stude nts. Community College Journal of Research and Practice 26, 659 679. American Association of State Colleges and Universities (2006, August). Student debt burden. P olicy Report Volume 3, Number 8. Washington, DC. American Savings Education Council (19 99). Youth and money, 1999. Washington, DC. Baum, S. (2003). leadership role. New York: National Dialogue on Student Financial Aid, College Board. Baum, S. (2007, Winter). Hard heads a nd soft hearts: Balancing equity and efficiency in institutional student aid policy. New Directions for Higher Education, 140, 75 85. Baum, S. & Ma, J. (2007). Education pays: The benefits of higher education for individuals and society. Washington, DC: College Board. College on credit: How borrowers perceive their education. The 2002 National Student Loan Survey. Boston: Nellie Mae Corporation. Baum, S. & Saunders, D. (1998). Life after debt: Results of the national stud ent loan survey. Braintree, MA: Nellie Mae Foundation. Baum, S. & Schwartz, S. (2006). How much debt is too much? Defining benchmarks for manageable student debt. Washington, DC: The College Board. Becker, G.S. (1964 1997 ). Human capital: A theoretica l and empirical analysis, with special reference to education. New York: National Bureau of Economic Research. Block, S. (2006, October 25). Private student loans pose greater risk. USA Today. Retrieved online on May 4, 2009 from http://www.usatoday.com/money/perfi/college/2006 10 24 private student loans usat_x.htm Bloom, D.E., Hartley, M. & Rosovsky, H. (2007). Beyond private gain: The public benefi ts of higher education. In J.F. Forest & P.G. Altbach (Eds.), International Handbook of Higher Education, 293 308. Bourdieu, P. (1986). The forms of capital. In J.G. Richardson (Ed.), Handbook of Theory and Research for the Sociology of Education (pp. 241 258). New York: Greenwood Press. Boushey, H. (2005, September). Student debt: Bigger and bigger. Washington, DC: Center for Economic and Policy Research. Bowen, H.R. (1997). Investment in learning: The individual and social value of American higher educ ation. Baltimore, MD: John Hopkins University Press.
142 Burdman, P. (2005). The student debt dilemma: Debt aversion as a barrier to college access. Berkeley, CA: Center for Studies in Higher Education. Callender, C. & Jackson, J. (2005). Does the fear of de bt deter students from higher education? Journal of Social Policy, 34(4), 509 540. Carey, K. & Dillon, E. (2009, July). Drowning in debt: The emerging student loan crisis. Washington, DC: Education Sector. Cellini, S.R. (2008). Causal inference and omitt ed variable bias in financial aid research: Assessing solutions. The Review of Higher Education, 31(3), 329 354. Chen, R. (2008). Financial aid and student dropout in higher education: A heterogeneous research approach. In J.C. Smart (Ed.), Higher educati on: Handbook of theory and research, Vol. 23, (pp.209 239). New York: Springer. Choy, S.P, & Li, X. (2006). Dealing with debt: 1992 later (NCES 2006 156). U.S. Department of Education. Washington, DC: National Cent er for Education Statistics. Christman, D.E. (2000). Multiple realities: Characteristics of loan defaulters at a two year public institution. Community College Review, 27(16), 16 32. Christou, C. & Haliassos, M. (2006). How do students finance human cap ital accumulation? The choice between borrowing and work. British Journal of Sociology in Education, 24, 621 636. Cofer, J. & Somers, P. (2000). A comparison of the influence of debtload on the persistence of students at public and private colleges. Jour nal of Student Financial Aid, 30(2), 39 58. Cohn, E. & Geske, T.G. (1990). The economics of education. New York: Pergamon Press. College Board (2008). Trends in student aid. Washington, DC: Author. College Board (200 9 ). Trends in student aid. Washington DC: Author. Creech, J.D. & Davis, J.S. (1999). Merit based versus need based aid: The continual issues for policymakers. In J.E. King (Ed.), changing ( pp 120 136) Phoenix, AZ: Oryx Press. Cunningha m, A.F. & Santiago, D.A. (2008, December). Student aversion to borrowing: Who Report by the Institute for Higher Education Policy and Excelencia in Education. Curs, B.R., Singell, L.D., & Waddell, G.R. (2007). The Pell program at thirty years. In J.C. Smart (Ed.), Higher education: Handbook of theory and research, Vol. XXII (pp. 281 334). New York: Springer.
143 Davies, E., & Lea, S. E. G. (1995). Student attitudes to student debt. Journal of Economic Psychology, 16, 663 679. DesJard ins, S.L. & Toutkoushian, R.K. (2005). Are students really rational? The development of rational thought and its application to student choice. In J.C. Smart (Ed.), Higher Education: Handbook of Theory and Research, Vol. XX (pp.191 240). New York: Springer Dowd, A.C. (2006). A research agenda for the study of the effects of borrowing and the going choices. Paper prepared by the New England Resource Center for Higher Education (NERCHE) for the Project on Stud ent Debt. Dowd, A.C. & Coury, T. (2006b). The effect of loans on the persistence and attainment of community college students. Research in Higher Education, 47(1), 33 62. Dowd, A.C. (2008). Dynamic interactions and intersubjectivity: Challenges to causal modeling in studies of college student debt. Review of Educational Research, 78(2), 232 259. Dowd, B. & Town, R. (2002). Does X really cause Y? Report published by AcademyHealth. Retrieved online on February 22, 2009 from http://www.hcfo.net/pdf/xy.pdf Doyle, W.R. (2008) Access, choice and excellence: The competing goals of state student financial aid programs. In, Baum, S. McPherson, M and Steele, P. (Eds.) The Effectiveness of Student Aid Policies: What the Research Tells Us. ( pp. 1 59 188). New York: College Board. Dynarski, M. (1994). Who d efaults on s tudent l oans? Findings from the National Postsecondary Student Aid Study. Economics of Education Review 13(1), 55 68. Ehre nberg, R.G. (2006, January February). The perfect storm and the privatization of higher education. Change, 38(1), 46 53. Flint, T.A. (1997). Predicting student loan defaults. The Journal of Higher Education, 68(3), 322 354. Floyd, D.L. & Walker, K.P. (2 009). The community college baccalaureate: Putting the pieces together. Community College Journal of Research and Practice, 33, 90 124. Fossey, R. (1998). The dizzying growth of the federal student loan program. In R. Fossey & M. Bateman (Eds.) Condemning students to debt: College loans and public policy. New York: Teachers College Press. Geske, T.G. & Cohn, E. (1998). Why is a high school diploma no longer enough? The economic and social benefits of higher education. In R. Fossey & M. Bateman (Eds.), Co ndemning students to debt: College loans and public policy (pp. 19 26). New York: Teachers College Press.
144 Gladieux, L. & King, J. (1995). Challenge and change in the federal role: Preparing for the information needs of the Twenty First Century. New Direct ions for Institutional Research, 85, 21 31. Gladieux, L. & Perna, L. (2005, May). Borrowers who drop out: A neglected aspect of the student loan trend. The National Center for Public Policy and Higher Education, Report #05 2. Washington, DC. Grable, J.E. & Joo, S. (2006). Student racial differences in credit card debt and financial behaviors and stress. College Student Journal, 40(2), 400 408. Gross, J., Cekic, O., Hossler, D., & Hillman, N. (2009). What matters in student loan default: A review of the e xtant research literature. Journal of Student Financial Aid, 38(2), 19 29. Harrast, S.A. (2004). Undergraduate borrowing: A study of debtor students and their ability to retire undergraduate loans. Journal of Student Financial Aid, 34(1), 21 37. Hartle, T.W. (1996). The Republican revolution: What it means for student aid. Educational Record, 77(1), 8 16. Hahs Vaughn, D. (1997). Using NCES national datasets for evaluation of postsecondary issues. Assessment & Evaluation in Higher Education, 32(3), 239 35 4. Hahs Vaughn, D. (2005). A primer for using and understanding weights with national datasets. The Journal of Experimental Education, 73(3), 221 248. Hayhoe, C.R., Leach, L. & Turner, P.R. (1999). Discriminating the number of credit cards held by colleg e students using credit and money attitudes. Journal of Economic Psychology, 20(6), 643 656. Heller, D.E. (2001). Debts and decisions: Student loans and their relationship to graduate school and career choice. Lumina Foundation for Education, New Agenda S eries. Indianapolis, IN. Heller, D.E. & Rogers, K.R. (2006). Shifting the burden: Public and private financing of higher education in the United States and implications for Europe. Tertiary Education and Management, 12, 91 117. Horn, L.J., Chen, X., & C hapman, C. (2003). Getting ready to pay for college: What students and their parents know about the cost of college tuition and what they are doing to find out. Washington, DC: U.S. Department of Education. Huck, S.W. (2004). Reading statistics and resear ch. (4 th ed.). Boston MA: Pearson. Joo, S., Grable, J.E., & Bagwell, D.C. (2003). Credit card attitudes and behaviors of college students. College Student Journal, 37(3), 405 420.
145 Kantrowitz, M. (2007). The financial value of a higher education. Journal of Student Financial Aid, 37(1), 19 27. Kantrowitz, M. (2009, May). Characteristics of borrowers with excessive debt. Student Aid Policy Analysis for FinAid.org. Retrieved online on December 12, 2009 from http://www.finaid.org/educators/20090511excessivedebt.pdf K ing, J.E. (1998). Student borrowing: Is there a crisis? In Student Loan Debt: Problems and prospects. Proceedings from a National Symposium, December 10, 1997. Washington DC. King, J.E. (1999). Crisis or convenience: Why are students borrowing more? In J.E. King, (Ed.), changing, (pp 165 176 ) Phoenix, AZ: Oryx Press. King, J.E. (2005, June). Federal student loan debt: 1993 2004. American Council on Education Issue Brief. Washington, DC. King, T. & Bannon, E. (2002). The burden of borrowing: A report on the rising rates of student loan debt. King, T. & Frishberg, I. (2001). Big loans bigger problems: A report on the sticker shock of student loans. Knapp, L.G. & Seaks, T.G. (1992). An a nalysis of the p robability of d efault on f ederally g uarant eed s tudent l oans. Review of Economics and Statistics, 74(3), 404 411. Lange, C. & Byrd, M. (1998). The relationship between feelings of financial distress and feelings of psychological well being in New Zealand university students. International Journal o f Adolescence and Youth, 7, 193 209. Lewis, A. & van Venrooij, M. (1995). A note on the perceptions of loan duration and repayment. Journal of Economic Psychology, 16(1), 161 168. Long, B.T (2007). The contributions of economics to the study of college access and success. Teachers College Record, 109(10), 2367 2443. Lyons, A.C. (2008). Risky credit card behavior of college students. In J.J. Xiao (ed.), Handbook of Consumer Finance Research, (pp 185 207). New York: Springer. Manski, C.F. & Wise, D.A. ( 1983). College choice in America. Cambridge, MA: Harvard University Press. McDonough, P.M. (1997). Choosing colleges: How social class and schools structure opportunity. Albany: State University of New York Press.
146 McDonough, P.M. & Calderone, S. (2006). The meaning of money: Perceptual differences between college counselors and low income families about college costs and financial aid. American Behavioral Scientist, 49(12), 1703 1718. McKinney, L. & Morris, P.A. (2010). Examining an evolution: A case stu dy of organizational change accompanying the community college baccalaureate. Community College Review, 37(3), 187 208. McPherson, M.S. & Schapiro, M.O. (1998). The student aid game: Meeting need and rewarding talent in American higher education. Princeto n, NJ: Princeton University Press. Mendez, J.P., & Mendoza. (2008). The implications of financial aid packages on African American student retention. National Association of Student Affairs Professionals Journal, 11 (1), 46 65. Millet, C.M. (2003). How und ergraduate loan debt affects application and enrollment in graduate or first professional school. Journal of Higher Education, 74(4), 386 427. Millimet, D. (2001). Retrieved Februa ry 4, 2009 from http://www.stata.com/support/faqs/stat/bias.html Mumper, M. & Vander Ark, P. (1991). Evaluating the Stafford Loan program: Current problems and prospects for reform. The J ournal of Higher Education, 62(1), 62 78. Myers, J.L. & Well, A.D. (2003). Research design and statistical analysis (2 nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates. National Center for Education Statistics (2008, February). Trends in undergraduate bor rowing II: Federal student loans in 1995 96, 1999 2000, and 2003 04. NCES Report 179. U.S. Department of Education. Washington, D.C. National Postsecondary Student Aid Study (NPSAS) 2007 08 (2009, April). Student Financial Aid Estimates for 2007 08. NCE S First Look Report, 2009 116. U.S. Department of Education. Washington, D.C. Norvilitis, J.M. & Santa Maria, P. (2002). Credit card debt on college campuses: Causes, consequences, and solutions. College Student Journal, 36(3), 356 363. Nor vilitis, J.M ., Merwin, M.M., Osberg, T.M., Roehling, P.V., Young, P., & Kamas, M.M. (2006). Personality factors, money attitudes, financial knowledge, and credit card debt in college students. Journal of Applied Social Psychology, 36(6), 1395 1413. Norvilitis, J.M., Szablicki, P.B., & Wilson, S.D. (2003). Factors influencing levels of credit card debt in college students. Journal of Applied Social Psychology, 33(5), 935 947. Osborne, J.W. (2002). Notes on the use of data transformations. Practical Assessment, Researc h, and Evaluation, 8(6). Retrieved April 14, 2010 from http://PAREonline.net/getvn.asp?8&n=6
147 Pascarella, E.T. & Terenzini, P.T. (2005). How college affects students: A third decade of research (Vol. 2). San Francisco, CA: Jossey Bass. Paulsen, M.B. (1998). Recent research on the economics of attending college: Returns on investment and responsiveness to price. Research in Higher Education, 39(4), 471 489. Paulsen, M.B. (2001). The economics of hum an capital and investment in higher education. In M.B. Paulsen & J.C. Smart (Eds.), The finance of higher education: Theory, research, policy, and practice. New York: Agathon. Paulsen, M.B. & Toutkoushian, R.K. (2008). Economic models and policy analysis in higher education: A diagrammatic exposition. In J. Smart (Ed.), Higher education : Handbook of theory and research (Vol. XXIII, pp. 1 48). Springer Perna, L.W. (2006a). Studying college access and choice: A proposed conceptual model. In J.C. Smart (E d.), Higher education: Handbook of theory and research (Vol. XXI, pp. 99 157). Springer. Perna, L.W. (2006b). Understanding the relationship between information about college costs related behaviors. American Behavi oral Scientist, 49, 1620 1635. prices. Research in Higher Education, 49, 589 606. Perna, L.W. & Titus, M. (2005). The relationship between parental involvement a s social capital and college enrollment: An examination of racial/ethnic group differences. The Journal of Higher Education, 76(5), 485 518. P owing. The Wall Street Journal p. D6. Pinto, M.B. & Mansfield P.M. (2006). Financially at risk college students: An exploratory investigation of student loan debt and prioritization of debt repayment. Journal of Student Financial Aid, 35(2), 22 32. Podgursky, M., Ehlert, M., Monroe, R., Watson, D., and Wittstruck J. (2000). Student l oan d efaults and e nrollment p ersistence. Journal of Student Financial Aid 32(3), 27 42. Price, D.V. (2004a). Borrowing inequality: Race, class, and student loans. Boulder, CO: Lynne Rienner Publishers. Price, D.V. (2004b). Education al debt burden among student borrowers: An analysis of the baccalaureate & beyond panel, 1997 follow up. Research in Higher Education, 45(7), 701 737. Project on Student Debt (2009, August). Quick facts about student debt. The Institute for College Access & Success
148 Project on Student Debt (2009, December). Student debt and the class of 2008. The Institute for College Access & Success. Psacharopoulos, G. (2006). The value of investment in education: Theory, practice, and policy. Journal of Education Fina nce, 32(2), 113 136. Redd, K.E. (1999). The changing characteristics of undergraduate borrowers. In J.E. King, (Ed.), changing, (pp 78 97 ) Phoenix, AZ: Oryx Press. Redd, K.E. (2004). Lots of money limited options: College choice and student financial aid. Journal of Student Financial Aid, 34(3), 29 39. Saenz, V.B. & Ponjuan, L. (2009). The vanishing Latino male in higher education. Journal of Hispanic Higher Education, 8(1), 54 89. Sall ie Mae (2 usage rates and trends 2009. Retrieved September 12, 2009 from http://www.salliemae.com/NR/rdonlyres/0BD600F1 9377 46EA AB1F 6061FC763246/10744/SLMCreditCardUsageStudy41309FINAL2.pdf Scherschel, P.M. (1998). Student indebtedness : Are borrowers pushing the limits? Indianapolis, IN: USA Group Foundation. Schultz, T.W. (1961). Investment in human capital. The American Economic Review, 51(1), 1 17. Shavelson, R.J. (1996). Statistical reasoning for the behavioral sciences. (3 rd ed). Needham Heigths, MA: Allyn & Bacon. Singer, A. & Paulson, A. (2004). Financial access for immigrants. Policy Brief: Conference Report #19. Washington, DC: The Brookings Institution. Slaughter, S. & Rhoades, G. (2005). Markets in higher education: Students in the seventies, patents in the eights, copyrights in the n ineties. In P.G. Altbach, R.O. Berdahl, and P.J. Gumport (Eds.) American higher education in the twenty first century: Social, political, and economic challenges (2 nd ed.) pp.486 516 Baltimore, MD: The Johns Hopkins University Press. St. John, E.P. (199 0). Price response in persistence decisions: An analysis of the High School and Beyond senior cohort. Research in Higher Education, 31(4), 387 403. St. John, E.P. & Paulsen, M.B. (2001). The finance of higher education: Implications for theory, research, policy, and practice. In M.B. Paulsen and J. Smart (Eds.) The finance of higher education: Theory, research, policy, and practice pp. 545 568. New York, NY: Agathon Press.
149 St. John, E.P. (2006). Contending with financial inequality: Rethinking the contri butions of qualitative research to the policy discourse on college access. American Behavioral Scientist, 49(12), 1604 1619. Stanton Salazar, R.D. (1997). A social capital framework for understanding the socialization of racial Minority children and youth s. Harvard Education Review, 67(1), 1 40. Steele, P. & Baum, S. (2009, August). How much are college students borrowing? Washington, DC: College Board. Steiner, M. & Teszler, N. (2003). The characteristics associated with student loan default at Texas A& M University. College Station, TX: Texas Guaranteed in association with Texas A&M University. Sweetland, S.R. (1996). Human capital theory: Foundations of a field of inquiry. Review of Educational Research, 66(3), 341 359. Thomas, S.L. & Heck, R.H. (2001 ). Analysis of large scale secondary data in higher education research: Potential perils associated with complex sampling designs. Research in Higher Education, 42(5), 517 540. Thomas, S.L., Heck, R.H., & Bauer, K.W. (2005). Weighting and adjusting for de sign effects in secondary data analysis. New Directions for Institutional Research, Vol. 127. San Francisco: Jossey Bass. Tierney, W.G. & Venegas, K.M. (2006). Fictive kin and social capital: The role of peer groups in applying and paying for college. Th e American Behavioral Scientist, 49(12), 1687 1702. Titus, M.A. (2007). Detecting selection bias, using propensity score matching, and estimating Research in Higher Education, 48(4), 487 521. Trent, W.T., Lee, H.S., & Owens Nicholson, D. (2006). Perceptions of financial aid among students of color. American Behavioral Scientist, 49(12), 1739 1759 Volkwein, J.F & Cabrera, A.F. (1998). Who defaults on student loans? The effects of race, class, and gender on borrower behavior. In R. Fossey & M. Bateman (Eds.) Condemning students to debt: College loans and public policy. New York: Teachers College Press. Volkwein, J. F., & Szelest, B. P. (1995). Individual and c ampus c haracterist ics a ssociated with s tudent l oan d efault. Research in Higher Education 36(1), 41 72. Volkwein, J.F., Szelset, B.P., Cabrera, A.F., & Napierski Prancl, M.R. (1998). Factors associated with student loan default among different racial and ethnic groups. Jou rnal of Higher Education, 69(2), 206 237. Wei, C.C., Berkner, L., He. S., Lew, S., Cominole, M., & Siegel, P. (2009). 2007 08 National Postsecondary Student Aid Study (NPSAS:08): Student Financial Aid Estimates for 2007
150 08: First Look (NCES 2009 166). Nat ional Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education. Washington, D.C. Wilms, W.W., Moore, R.W., & Bolus, R.E. (1987). Whose fault is default? A study of the impact of student characteristics and institution al practices on guaranteed student loan default rates in California. Educational Evaluation and Policy Analysis, 9(1), 41 54. Woo, J. H. (2002). Factors a ffecting the p robability of d efault: Student l oans in California. Journal of Student Financial Aid 3 2(2), 5 23. Woolridge, J.M. (2002). Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press. cards. Journal of Consumer Studies, 19, 155 1 74.
151 BIOGRAPHICAL SKETCH Lyle McKinney was born and raised in Cordele, Georgia. He holds a BFA in Speech Communication from Valdosta State University and a MSM from Troy University. Before attending the University of Florida, Lyle served as Director of th e AndrewServes Servant Leadership Program and an instructor at Andrew College in Cuthbert, Georgia. Upon completion of his Ph.D. in Higher Education Administration, he will assume the role of Assistant Professor of Higher Education at the University of Hou ston