Do attitudes matter? attitudes towards debt and graduate student loan debt

PDF

Download

Open document

Flip pages

Summarize

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

DO ATTITUDES MATTER?
ATTITUDES TOWARDS DEBT AND GRADUATE STUDENT LOAN DEBT

by
Emily Chung and Rick Garcia

__________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC ROSSIER SCHOOL OF EDUCATION
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF EDUCATION

May 2015

Copyright 2015 Emily Chung and Rick Garcia

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT ii
Acknowledgements
Rick would like to acknowledge the following individuals:
• Lori and Aidan, you two have made my life magical. You inspire me daily and motivate me
to continue learning and growing.
• Our dissertation committee: Kristan, Tatiana, and Courtney, we could not have pulled this off
without your expert guidance and keen insights. Each critique made our research better and
improved our skills. Thank you!
• My writing partner, Emily. I could not have done this without you. Your perspectives were
vital to the quality of our project. I’d work with you again anytime.
• Derek and Michelle Vergara. Two of the biggest reasons for my success at the University of
La Verne were you two. You both have to know that I would not have survived my freshman
or sophomore years without your assistance. I chose my career path to give to future students
the gifts you gave to me.
• Drs. Campana, Rose, Reed. You three gave me a home in the Department of Philosophy and
Religion and forever changed the way I look at the world.
• Tim Brunold, Noemi Garcia-Tagorda, Joe Sanosa, and Ujjani Sahasrabudhe. Thank you for
helping me to build my career at the University of Southern California, for mentoring me,
and guiding my professional development. I promise to use all that I have learned from you
to build a practice worthy of your names.
• Technical Sergeant Huerta; Colonel Smith; Colonel Hessheimer; Colonel Weskamp; and
Major Hernandez. Thank you giving me a home with the 163d RW and for facilitating my
professional military development. I’ve learned tons about leadership by observing you and
your airmen in action.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT iii
• My cats, Jon Snow and Arya Stark, and dog, Cheyenne. You three spent countless hours
purring in my lap or curled up at my feet as I worked through this dissertation. I owe you
each a basket of treats.

Emily gratefully acknowledges the contributions of those that surrounded her that made
this dissertation and doctoral program possible. First, much love to my husband Gene, baby son
Ren, and my faithful hounds, Isis and Chaz, for sustaining me throughout this journey. In
particular, Ren, your arrival in our world and in the midst of writing has kept me focused on the
finish line as no other could. Kristan Venegas, our committee chair, was instrumental in
shepherding our research, and many thanks to our committee members, Tatiana Melguizo and
Courtney Malloy for their valuable feedback. Rick, I could not have asked for a better writing
collaborator; much gratitude for enriching the usually solitary writing experience! Jerry Lucido
and the USC Center for Enrollment Research, Policy, and Practice—I am very grateful that you
have encouraged this opportunity to truly deepen my learning. And my Tuesday night cohort,
and particularly my reading group, for support and keeping it real!

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT iv
Table of Contents
Acknowledgements ii
List of Tables vi
List of Figures vii
Abstract viii
Chapter One: Overview of the Study 1
Background of the Problem 2
Purpose of the Study 4
Significance of the Study 5
Overview of the Theoretical Framework 7
Overview of the Methodology 9
Note on Author Contributions 12
Chapter Two: Literature Review 13
Effects of Student Debt on the Economy 15
Good Debt and Bad Debt 18
Theoretical Framework 21
Rational Choice Theory 22
Human Capital Theory 25
Behavioral Economics 27
Attitudes and Behaviors towards Debt 30
Attitudes 30
Behaviors 36
Key Background Variables in Graduate Student Enrollment 39
Model of Attitudes and Student Borrowing 47
Chapter Three: Methodology 50
Definition of the Research Approach 51
Site and Sample 54
Instrumentation 55
Data Collection 56
Data Analysis 58
Chapter Four: Data and Findings 60
Full Survey Distribution Data 60
Descriptive Statistics of WUU Overall 60
Descriptive Statistics by School 63
School of Public Policy 63
School of Education 66
School of Social Work 68
Comparing WUU Master’s Student Debt to National Trends 71
Survey and Scale Reliability 73
Answering the Research Questions 78
Research Question 1 78
Research Question 2 80
Chapter Five: Discussion 89
Purpose of the Study and Research Questions 89
Discussion of Findings 90
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT v
Spending Compulsion 91
Debt Aversion 92
Beliefs about Knowledge across Financial Competencies 93
School Affiliation 94
Loan Information Seeking Behavior 95
Limitations 96
Implications for Practice 101
Future Research 108
Conclusion 110
References 112
Appendix A: Survey Instrument 127
Appendix B: Pilot Survey 131
Appendix C: Note on Author Contributions 139

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT vi
List of Tables
Table 1: WUU Respondent Characteristics 61
Table 2: School of Public Policy Respondent Characteristics 65
Table 3: School of Education Respondent Characteristics 67
Table 4: School of Social Work Respondent Characteristics 70
Table 5: WUU Rates of Borrowing Compared to National Rates 71
Table 6: WUU Schools Compared to U.S. Counterparts 73
Table 7: Correlations of DA Independent Variables with Amount Borrowed (DV) 79
Table 8: Correlations of SC Independent Variables with Amount Borrowed (DV) 80
Table 9: KFC Pearson Correlations within Construct Items 82
Table 10: FE Pearson Correlations within Construct Items 82
Table 11: FB Pearson Correlations within Construct Items 83
Table 12: Predicting Amount Borrowed with Multiple Regression (Model 1) 85
Table 13: Predicting Amount Borrowed with Multiple Regression (Model 2) 87
Table 14: Pilot Study Respondent Characteristics 134

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT vii
List of Figures
Figure A: Model of Attitudes and the Relationship to Student Loan Borrowing 48
Figure B: Model of Attitudes and the Relationship to Student Loan Borrowing 74

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT viii
Abstract
This dissertation contributes to a small but growing body of research on graduate student loan
debt. We designed a survey to measure whether debt attitudes were related to the amounts
master’s students at a large, urban, research university borrowed to finance a year of graduate
study. Overall, we found that debt attitudes are related to the amounts borrowed by the
respondents in our study. Our theoretical framework of rational choice, human capital, and
behavioral economic theories revealed that respondents’ borrowing decisions are influenced by
considerations of cost-benefit analyses, long-term returns on investment, and irrational aversions
to debt. The implications of this research are that that institutional revenue streams need to rely
less upon master’s students deficit financing their degree programs, financial aid and admissions
professionals should proactively outreach to prospective borrowers, and that greater transparency
is needed on the topic of graduate student borrowing and career outcomes.

CHAPTER ONE
OVERVIEW OF THE STUDY
The amount of money graduate students are borrowing to pay for master’s programs is a
growing concern in the 21st century. Recent analysis of graduate student debt found that in 2012-
2013 graduate students were less than one fifth of federal loan recipients, but held approximately
41 percent of the debt (McCann, 2013). At the same time, graduate enrollment has been rising.
There were 54% more master’s students enrolled in academic year 2012-2013 than in 1999-2000
and the number of students in master’s programs is expected to increase another 17 percent by
2020 (Snyder & Dillow, 2013). Overall, student loans surpassed credit card debt in 2011
(FinAid, 2014; Lewin, 2011) and the average graduate student owes approximately $57,600
(Bidwell, 2014). The significance of this milestone is that while many United States citizens
have access to credit cards, far fewer obtain access to postsecondary education and far fewer
pursue and complete advanced graduate study (Board of Governors, 2012). Some are worried
that an unchecked growth of student loan debt will negatively impact vital economic sectors like
housing (Brown, Caldwell, & Sutherland, 2014; Neil, 2014; NPR, 2014). Student loan debt and
its perceived effects have become hot topics on internet discussion boards to the point where
pages are dedicated exclusively to this topic (e.g. Reddit.com/r/LostGeneration,
Younginvincibles.org). The growth of master’s student loan debt is the problem we seek to
address in this dissertation.
While an abundance of literature exists on the topic of undergraduate student finance,
scholarships, and indebtedness (Kim & Eyermann, 2006; Malcom & Dowd, 2012; Millet; 2003;
Perna, 1998; Weiler, 1994), little has been written about how graduate students frame their
decisions to borrow. The purpose of this study is to measure master’s students’ attitudes towards
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 2
debt and analyze whether these attitudes influence how much they borrow to pay for their
programs. Master’s students at a large research university in southern California will be surveyed
to determine their attitudes towards debt. The data analysis is informed by rational choice theory
(RCT), human capital theory (HCT), and behavioral economic (BE) theory. This is a team
project consisting of two researchers who employ quantitative methods.
Background of the Problem
Graduate student enrollments are growing and student debt levels have accompanied this
growth. As the number of students completing a bachelor’s degree has steadily increased, so too
has the number of students completing master’s degrees. In 2011-2012 the number of master’s
degrees awarded numbered 756,000 and was approximately 27.56 percent of all degrees awarded
that year (Snyder & Dillow, 2013). Default rates on student loans have also been climbing.
Approximately 240,000 student loan borrowers entering repayment in 2008 defaulted on their
loan(s) (U.S. Department of Education, 2010a, 2010b). More students are seeking advanced
education and more students are failing to pay their student loans. This is a problem of growing
concern that has resulted in various explanations.
One such explanation is that the costs of providing postsecondary education are rising
and this corresponds with increases in tuition prices. The National Commission on the Cost of
Higher Education (NCCHE) finds that public higher education tuition has climbed 57% between
1987 and 1996 and prices for private schools have soared an alarming 99% over the same period
of time (NCCHE, 1998). One narrative for explaining the steady climb of college prices and
graduate student debt is that higher education is a costly enterprise and has been losing
government support over time (Archibald & Feldman, 2011; Ehrenberg, 2000; Vedder, 2004).
The higher education price index (HEPI; Commonfund, 2013), a popular measure of the cost of
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 3
providing a postsecondary education, has charted an average positive increase of 3.025 percent
each year from 2002 to 2013. Higher education is an expensive undertaking requiring multiple
sources of funding. Revenues are derived, in proportions varying by institution, from three
primary domains: tuition and fees, commercial activities, and donative support (Winston, 1999).
As the costs of an industry that is heavily dependent upon highly educated professionals
increases, revenue shortfalls in one domain must be made up in others. Often these shortfalls end
up funded either by government subsidy (in the form of grants and tax breaks), by charitable
donations, and/ or by the students themselves (Archibald & Feldman, 2011; Ehrenberg, 2000;
Vedder, 2004). Yet it has also been argued that an increase in cost sharing—tuition hikes that
must be funded by the student—can explain some of increase in student loan debt (Johnstone,
2003). In light of this it is unsurprising that in 2011-2012, many master’s degree students held
debt ranging from $17,500 to $33,299 (Radwin et al., 2013). The increased demand for advanced
education (Snyder & Dillow, 2013) is interacting with prevailing economic conditions and a
general trend towards sharing the costs of producing postsecondary education with students
(Johnstone, 2003) might explain why student loan balances are increasing.
Higher education has historically been framed as a vehicle into the middle class. The
increasing amounts of debt students are assuming are challenging this history. Some argue this
increase in debt amounts to 21st century indentured servitude (Williams, 2008) and a foreclosure
of the youthful exemption from marketplace conscription wherein young citizens must pay to
learn how to fully participate in American democracy (Williams, 2006). Others have found
markers of economic segregation of access to higher education (Lillis, 2008) and economically
segregated patterns of persistence and success (Dowd & Coury, 2006). Lastly, it is being debated
whether the student loan debt is stunting the national recovery from the recession (Akers, 2014;
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 4
Akers & Chingos, 2014; Brown et al., 2014; Heller, 2014 a, b; Korkki, 2014; Neil, 2014; NPR,
2014; Ungarino, 2014). The effects of student debt on the economy are difficult to measure
(Akers, 2014b), but it is reasonable to assume that as one’s debt totals increase one’s full
participation in economic activity becomes increasingly foreclosed (Williams, 2006; 2008). We
recognize the difficulty inherent in attempting to measure students’ attitudes and determine
whether these attitudes influence students’ borrowing behaviors. This does not diminish the
importance of attempting to increase our understanding of the processes by which graduate
students approach borrowing and rationalize their borrowing decisions.
Purpose of the Study
The goal of this inquiry is to document whether graduate students’ attitudes towards debt
influence their borrowing behaviors. The two research questions for this project are:
1. Is there a relationship between master’s students’ attitudes towards debt and the
amounts they borrow in student loans?
2. If attitudes predict borrowing patterns, what factors are associated with increased
borrowing and decreased borrowing?
To answer these research questions we developed a survey instrument that allows us to measure
students’ attitudes about debt and test whether these attitudes affect borrowing behaviors. Trying
to answer the questions of whether and how attitudes affect borrowing is inherently difficult
because there are many exogenous and endogenous variables that are not measured (Dowd,
2008). This should not discourage any effort to increase our understanding of the factors that
affect students’ borrowing decisions. As such, we plan to contribute to the literature by
examining a specific subset of factors believed to affect graduate students’ borrowing behaviors:
attitudes. Specifically, we wish to determine if attitudes towards debt predict the amounts that
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 5
students borrow. If attitudes are found to affect borrowing patterns, we would like to determine
whether certain factors are more salient predictors of borrowing. Answers to these research
questions can yield insights into how financial aid professionals and enrollment managers can
influence students’ borrowing decisions and improve outcomes for students and their institutions.
Significance of the Study
While the nation’s collective student loan debt now exceeds $1.2 trillion (FinAid, 2014),
an amount greater than home mortgages or credit card debt, most of the attention has been
focused on undergraduate student loan debt (Kim & Eyermann, 2006; Malcom & Dowd, 2012;
Millet; 2003; Perna, 1998; Weiler, 1994). Analysis of student indebtedness suggests that the
attention paid to undergraduates may be misplaced. For instance, it has been found that graduate
students hold nearly half of all the student loan debt in the country despite being constituting 1/5
of all U.S. postsecondary enrollments (McCann, 2013). Therefore it is important that this study
examines not only a problem of great financial magnitude, but delineates the various issues
surrounding graduate student loan debt. In fact, the conflation of undergraduate and graduate
student loan debt obscures the differences between undergraduate and graduate student loans,
both at the individual student level and the policy level. For example, graduate students do not
receive Pell Grants to finance their education, and face no limits on their student loan borrowing
(McCann, 2013). Conversely undergraduate students are heavily subsidized by taxpayers through
government grants, subsidized loans, and are held to limits to the amount of student loans taken.
In comparison to the abundance of literature on undergraduate student college choice and
financial aid, there is little research on graduate students that examines enrollment choice or
financial aid issues. Specifically, there is a dearth of literature on the factors, financial as well as
other, that contribute to the decision of graduate students to enroll in an institution (Mertz,
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 6
Eckman, & Strayhorn, 2012), which leads to the gap of knowledge on graduate student
motivations and expectations. For example, Malcom and Dowd (2012) point to the lack of
research on how student loan debt affects underrepresented ethnic groups in their decisions to
enroll in STEM fields of study in graduate school. Callender and Jackson (2005) also raise the
critique that many quantitative inquiries on student debt have ignored measuring students’
attitudes towards consumer and educational debt. Hence our focus on measuring attitudes and
testing for whether there is a relationship between attitudes and student borrowing.
Postsecondary education confers private and public benefits that justify public support.
Completion of the bachelor’s degree facilitates middle-class attainment and the procurement of
advanced degrees yields lifetime increases to wage premiums. Increased purchasing power
notwithstanding, investments in human capital development also accrue non-monetary benefits
including greater job satisfaction, health, civic engagement, and gains to cognitive and non-
cognitive development (Astin, 1993; Becker, 1993; Jaeger & Page, 1996).
Likewise, graduate degrees are held to foster the mastery of new technologies and
production techniques that are vital to increasing economic productivity and decreasing
production costs (Archibald & Feldman, 2011). Graduate and professional degrees are vital in
the more lucrative STEM fields, for example, to increase parity in wages between
underrepresented ethnic groups and Whites (Malcom & Dowd, 2012). Postsecondary education
represents a significant investment, by individual students and taxpayers, in national human
capital development, but the rise of cost sharing appears to be driving more students into debt
(Johnstone, 2003). The rise of student debt also appears to affect consumer activity across
multiple sectors (Akers, 2014; Akers & Chingos, 2014; Brown et al., 2014; Heller, 2014 a, b;
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 7
Korkki, 2014; Neil, 2014; NPR, 2014). It is important, therefore, to find ways to help students
not only access graduate educational opportunities, but to help them to borrow responsibly.
Finally, this inquiry is significant because it holds potential to help practitioners to
improve their practice and affect positive outcomes for students. We know that access to
financial aid information mediates students’ access to and use of financial aid (Venegas, 2003).
We also know that environmental factors, including the availability of knowledgeable financial
aid professionals, similarly affects whether students attempt to access higher education or use
financial aid (Tierney & Venegas, 2009). It is plausible that if financial aid professionals and
enrollment managers knew whether and how master’s students’ attitudes influenced their
borrowing decisions, these professionals could improve their practice by focusing their
interventions on those attitudes most related to riskier, or excessive, borrowing. Our goal is to
facilitate improvements in graduate student financial aid policy and student loan counseling. In
particular, for those institutions that seek to increase the number of underrepresented ethnic
students to their graduate programs, for example, it is useful to understand what factors affect
students’ financing decisions.
Overview of the Theoretical Framework
The process by which graduate students make decisions about how to finance their
studies is explored using principles of rational choice and behavioral economics. Research on
student loan and financial aid naturally lends itself to these theories as students are expected to
decide on the investment of postsecondary education after weighing the various costs and
benefits of the decision (Goldrick-Rab et al., 2009; Perna, 2006a). Specifically under the
umbrella of rational choice theories, we draw upon human capital theory (HCT) for its
applicability to the social sciences and in examining individuals’ decision making. Becker (1993)
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 8
describes human capital investments as rational behavior in which the resources are expended
with the goal of increasing one’s knowledge, skill, and long-term socioeconomic viability. Such
investments, however, are made on the basis of limited information and knowledge in
determining the associated costs and benefits. In the context of borrowing to fund graduate
education, there are direct (monetary) costs like tuition and fees, and indirect (non-monetary)
costs like opportunity cost, that comprise the cost-benefit analysis individuals make when
deciding when to borrow and how much to borrow.
HCT is useful for our analysis because it accounts for variances in the social contexts of
each individual. HCT maintains that individuals with access greater financial capital or access to
individuals with greater human capital (e.g. years of formal education and/ or holding an
advanced degree) are more apt to pursue higher education. HCT is related to rational choice
theory (RCT) by virtue of positing that students’ decisions whether to pursue advanced education
are framed by variance in the amounts of information to which they have access. While variance
in access to information affects educational outcomes (Astin, 1993; Venegas, 2006), it is
theorized that individuals nonetheless engage in some form of cost-benefit analysis wherein the
economic and non-economic short-term costs and long-term benefits of higher education (Perna,
2006a). Thus, it is rational to invest in advanced studies insofar as such investments yield more
benefits than costs. Despite our personal beliefs in rationality, there is a body of literature that
reveals the limits of human rationality and explains why individuals sometimes make irrational
decisions.
The study of irrational economic decision-making is known as behavioral economics.
Whereas rational choice (Hogarth & Reder, 1987) helps us to understand how individuals
accomplish goals (utility maximization) via resource expenditure, behavioral economics clarifies
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 9
the mechanisms that interfere with rational cost-benefit analyses (Ariely, 2009). In short,
behavioral economics provides another lens to interpret human behavior.
For instance, Avery and Hoxby (2004) found that if students were admitted to their top-
choice schools, they would respond excessively to loans irrespective of a lack of differences in
institutional spending on instruction and academic support. Here we see students’ maximizing
their utility by accepting the offer of admission to their preferred school. A behavioral economist
would explain this behavior as a function of students’ expectations being anchored by the
manipulations of others (Ariely, 2009; Lee, Frederick, & Ariely, 2006). A similar phenomenon
occurs when institutions maximize their utility by providing handsome tuition discounts to
highly desirable students (e.g. high GPA or high test scores, student athletes) and lower tuition
discounts to less desirable students (Ehrenberg & Sherman, 1984). Such behavior is irrational
from an economic perspective because the purpose of capital investments is presumably to
increase the overall stock of human capital via resource investment in instruction and student
support. However, the behavioral economist counterargument is that prestige and brand
recognition create emotional attachments that are strengthened by expectations of quality and
prior experiences (Ariely, 2009; Lee et al., 2006). As such, our inquiry leverages insights from
RCT, HCT, and behavioral economics to determine whether students’ attitudes towards debt
frame their borrowing decisions.
Overview of the Methodology
We designed a survey instrument to facilitate our understanding of graduate students’
borrowing decisions. This instrument was piloted using a sample of volunteers contacted via
social media and email. Afterwards we surveyed the master’s student population of a large
research university in southern California. Our sampling strategy was to obtain a convenience
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 10
sample of no less than 365 respondents, as this will yield sufficient power and a 5% error rate
(Raosoft, 2004). This calculation was based on a population size of 7,000 individuals. We
obtained responses from 587 respondents for the survey, providing us more with a more than
sufficient sample for analysis.
Our unit of measurement is the student, but we designed the instrument to capture
demographic and nominal data to permit the analysis of differences between and within groups.
Data was analyzed to determine whether students’ attitudes towards debt are predictive of their
borrowing decisions. Lastly, we determined whether there is a factor structure to students’ debt
attitudes. An implication for this research is that if we do find that attitudes influence borrowing
decisions, financial aid professionals and enrollment managers will be able to modify their
counseling practices to encourage students to borrow responsibly.
Here we have laid the foundation for our study of master’s students’ attitudes towards
borrowing federal loans. Our research design makes use of insights from RCT, HCT, and
behavioral economics to determine the extent to which students’ attitudes affect their borrowing
decisions. Our theoretical framework facilitates the interpretation of the data by recognizing that
students’ borrowing decisions are likely motivated by beliefs about human capital investment
and informed by varying levels of cost-benefit analysis. Measurements of these domains will
hopefully help us to answer the question: Where does rational borrowing behavior begin and
end?
This study is important for multiple reasons. First, our goal is improve our knowledge of
the problem of the growth of master’s student debt. Second, we hope to contribute to solving the
problem of growing master’s student debt. Lastly, our study examines the problem by filling a
gap in our knowledge about master’s student attitudes towards debt and whether these attitudes
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 11
affect borrowing. We study this problem by developing an instrument that borrows from existing
studies of financial literacy and student borrowing behavior because an instrument designed
specifically for measuring master’s students’ attitudes towards debt does not presently exist.
While undergraduates’ borrowing decisions have been extensively studied, we are filling
a gap in our knowledge by examining how attitudes influence master’s student borrowing
decisions. This study employs scales and items from prior studies of undergraduate attitudes
towards debt because no such research exists that describes whether graduate student borrowing
is similarly affect by attitudes. It follows that since we do not yet know whether attitudes
influence graduate students’ borrowing; we also do not know whether the attitudes that affect
undergraduate students’ borrowing decisions will be the same as those that might affect graduate
students’ borrowing decisions.
By piloting a survey instrument with masters’ students from two different types of higher
education institutions, this study addresses two research questions pertaining to whether there is
a relationship between graduate students’ attitudes towards borrowing and the amount they
borrowed to pay for graduate school expenses. In Chapter 2, we examine literature related to
student loans and economic decision-making. Chapter 2 also describes how we synthesize RCT,
HCT, and behavioral economic theory to understand the complexities of graduate student
borrowing decisions. Chapter 3 provides the methodology for our study, defining the quantitative
research approach using our economic theoretical framework and describing the sample of
graduate student participants and the process for data collection and data analysis that we will
use for this quantitative study. We also cover the limitations of our study in this chapter.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 12
Note on Author Contributions

This dissertation was collaboratively conceptualized and written by the two authors. Each
author contributed original writing for sections in each chapter, and both authors’ writing was
edited and revised by each other in a collaborative manner. Please see Appendix C for details on
each chapter’s specific contributions.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 13
CHAPTER TWO
LITERATURE REVIEW
As Chapter 1 illustrated, while the costs and prices of higher education have been steadily
rising, resulting in increasing student debt, very little is known about the impact of student debt
or financial aid on graduate choice and enrollment. A cursory literature search shows there is
vast and rapidly growing body of research on undergraduate student college choice, enrollment,
and its relationship to student debt and financial debt. Amidst the increasing attention focused on
student loan debt and the affordability of higher education, it is surprising that there is not equal
attention paid to graduate students, who hold almost forty percent of the $1 trillion student loan
debt in the US (Delisle, 2014).
Perhaps contributing to the relative neglect of graduate student debt is the perception,
fueled by a media that sensationalizes undergraduate debt and the cost of undergraduate
education, is a belief that graduate students either do not borrow as much as undergraduates, or
do so with greater understanding of what benefits and consequences the borrowing entails.
Nevertheless, not only is graduate student debt important in its proportion of total debt, but
recent analysis shows that graduate indebtedness has been surging in recent years and continues
to grow. The median level of student indebtedness for an M.A. degree in 2004 was $38,000; for
the same degree in 2012 the median level had jumped to $59,000, an increase of 155% within a
span of eight years (Delisle, 2014).
Furthermore, graduate students are an important part of the higher education landscape;
there has been an explosion in the number of students pursuing and completing graduate
education. In recent years (1960-1977), the number of graduate degrees awarded has been more
than quadrupled, outstripping the growth in the number of undergraduate degrees awarded
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 14
(Mullen, Goyette, & Soares, 2003). In fact, graduate students comprise a third of the degrees
awarded in higher education, yet there is little research available on graduate education,
especially in comparison to the wealth of literature on undergraduate education (Mullen et al,
2003).
While graduate students are an important and growing part of the higher education
landscape, there is a dearth of literature about graduate students and how they make their choices
to enroll in post-baccalaureate education, especially in contrast to the vast body of literature on
undergraduate college choice and enrollment (Mertz, Strayhorn, & Eckman, 2012). Of the
research that has been conducted on undergraduate education, there is a body of research on the
relationship between undergraduate student loan debt and graduate enrollment, although the
relationship between accumulated student loan debt and graduate enrollment has proven
inconclusive (Malcom & Dowd, 2012; Millet, 2003; Mullen et al, 2003; Zhang, 2010).
There are many important aspects of graduate school choice and enrollment that remain
unexplored. One area for further exploration is to better understand the manner in which students
make the choice to attend graduate school. Furthermore, very few studies are designed to capture
this type of decision-making information as it relates to student debt and enrollment directly
(Malcom & Dowd, 2013; Mertz et al, 2012). Contextualization is important, given the fact that
students both receive and interpret information differently, impacting their individual decisions
(Malcom & Dowd, 2012). Our study seeks to shed more light on the context in which students
make decisions about graduate school by exploring attitudes towards borrowing for financing
their education.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 15
The purpose of this chapter is to provide a comprehensive overview of the literature
about graduate student enrollment in relation to student loans, contextualizing the background
for answering the two research questions guiding our study:
1. Is there a relationship between master’s students’ attitudes towards debt and the
amounts they borrow in student loans?
2. If attitudes predict borrowing patterns, what factors are associated with increased
borrowing and decreased borrowing?
First, we consider the significance of student debt on the economy. Then we move to an
exploration of the key variables influencing graduate student enrollment, including external and
institutional factors, as well as student level factors. We discuss our theoretical framework for
exploring graduate student enrollment and student loans, which integrates important tenets of
human capital theory, rational choice theory, and behavioral economic theory. Finally, we
discuss the literature on attitudes and behaviors towards debt overall and student loans in
particular to better understand how individual attitudes may impact graduate school decisions.
Effects of Student Debt on the Economy
As a policy for education funding, is deficit financing the human capital development of
our nation working? If we consider some of the effects of student borrowing on the economy, it
appears student lending/ borrowing is hindering economic development (Brown, Caldwell, &
Sutherland, 2014; Neil, 2014; NPR, 2014). A major indicator of economic health, besides GDP,
is the housing market. Considering the many ways housing can contribute to national economic
prosperity—architects and engineers designing homes, contractors and crews building homes,
interior furnishing and major appliance purchases, and new state and municipal tax revenues—
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 16
the observation that holders of student debt are less likely to seek or obtain home loans (Brown
et al., 2014; Neil, 2014, Ungarino, 2014) is an indicator to which attention must be paid.
Student debt appears to prevent graduates from starting their own businesses. Korkki
(2014) cites data indicating that as students expend their credit capacity on baccalaureate and
post-baccalaureate programs are less likely to finance start-up companies. This finding dovetails
with the concerns raised over student debt and homeownership. The theme here is that student
loans constrain consumer credit resulting in students’ consumer activities becoming limited.
Insofar as postsecondary enrollments have increased over the last several decades and the
aggregate student loan balance has topped $1.11 trillion, the implication seems to be that we
should expect to observe continued declines in homeownership and business ownership among
graduates as a function of student loan debt.
There are some researchers who maintain that the student debt problem is not really a
problem or, at least not as big a problem as perceived by the public. Akers and Chingos’ (2014)
analysis of data from the Federal Reserve’s Survey of Consumer Finances (SCF) argues student
debt burdens are equal to or less than those faced by borrowers two decades ago. The authors
also argue that the circumstances for student loan borrowers are not worse now than before and
support this claim with evidence that the average lifetime earnings of college degree holders is,
on average, greater than those who attended some college and those who never attended college.
It follows that if student debt is not as big of a problem, then the argument that homeownership
is stunted for holders of student loan debt is weakened. Akers (2014), again using SCF data,
finds that for every year after the recession of 2001, holders of student loan debt have owned
homes (held mortgage debt) at rates above those of who do not hold student loan debt. Akers’
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 17
proposed solution to the problem, however large or small, is to improve income based repayment
plans and extended repayment terms.
A rational analysis of the student loan debt problem has to consider the evidence to
determine the best solutions. Heller (2014 a, b) cautions against subscribing to media
sensationalism of student loan debt on the basis that such stories convey an unintended message:
student loan debt is bad. Heller (2014 a, b) employs a cost-benefit analysis in his argument that
students with college degrees and advanced degrees will always be better off than those who do
not have these qualifications. The problem presented by analyses like those provided by Akers
(2014) and Akers and Chingos (2014) is that they do not disaggregate their data to look for
patterns within their data. The problem may not be that student loan debt is bad overall, but
rather that student loan debt is a problem for select sub-groups. While completing a college
degree program may in fact produce substantial financial gains overall, the distribution of the
financial gains may be disproportionate for some groups at the expense of others.
With this discussion of the significance of student loan debt from primarily an economic
perspective, we move to the next section, which provides greater depth on the theoretical
framework we draw upon for our study. As we are trying to understand attitudes and behaviors
related to graduate student borrowing, our theoretical framework is primarily informed by
economic theories. Given the multidimensionality of graduate student loan debt ranging from
student demographic factors, to institutional type, to undergraduate performance, to
undergraduate debt, we decided upon a theoretical framework that draws upon three concepts
important in economic theory of higher education. Rational choice theory informs us of the
decision calculus that students make when deciding that a graduate degree is worth undertaking
student loans. Human capital theory contributes the idea that higher education is a smart and
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 18
worthwhile investment of limited resources. Lastly, behavioral economic theory helps us
understand the borrowing decisions that appear irrational, but are in fact outcomes arising from
human information processing limitations.
Good Debt and Bad Debt
Not all debt is equal. Some debts are good and others are bad (Bankrate, 2014). The
conceptualization of debt as either good or bad helps is a useful framework for categorizing
student loan debt. Extending the idea of good and bad debt to individuals suggest that there is
also a scale upon which to judge the “goodness” or “badness” of debtors’ decisions to borrow.
The idea that debt and debtors are good, bad, or somewhere in between is a useful one for this
dissertation. Our goal is determine whether students’ debt attitudes are related to their student
loan amounts. If such a relationship is found, it would be helpful to identify which attitudes are
more likely to result in good student loan debt versus bad student loan debt. Similarly, were a
relationship found between students’ attitudes towards debt and being good or bad debtors, we
are better positioned to create interventions that can spare students the challenges associated with
bad debt and bad borrowing behaviors. This section distinguishes good debt from bad, good
debtors from bad, and highlights the potential utility of measuring whether students’ attitudes
towards debt predispose them to good or bad debt.
The simplest conceptualization of debt is that debts are either good or bad. Good debts
generate returns over time and bad debts incur extensive costs over time (Bankrate, 2014). For
example, credit card purchases of disposable items are regarded as bad debts. These purchases
result in net losses over time because once disposable items are taken from the retailer they are
both worth less than what was paid and the interest paid on the good actually increases the actual
cost of the item. Good debts, on the other hand, are investments that create value (Bankrate,
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 19
2014). Examples of good debts include mortgage and student loans. A mortgage generates
returns in the form of increased home value. Similarly, student loans are held to increase debtors’
human capital (Avery & Turner, 2012; Becker, 1993; Dynarski, 2008; Volkwein et al., 1998).
Alternately, student loans may be considered a bad debt if the money that is borrowed is used for
non-educational expenditures on disposable items that are unrelated to one’s education. Further,
categorizing student loan debt as good using increased financial returns as the sole criterion is
problematic because there are many confounding variables affecting postsecondary educational
outcomes (Pascarella & Terenzini, 2005) up to an including income. While it seems a simple
matter of describing debt as good or bad as function of increasing one’s financial worth, it is
much more complicated. Nonetheless, the idea that debt is good or bad as a function of
increasing one’s net worth, or facilitating gains to net worth, is useful.
Whereas debt is cognizable as occurring on a scale of bad to good, debtors are similarly
cognizable as being somewhere on a scale from good to bad. A qualitative study of 27 White,
middle-class consumers found that debtors vary from managing debt (bad) to using debt the right
way (good; Penaloza & Barnhart, 2011). Those held to be managing their debt obtained their
credit from others (cosigners), had curtailed access to credit instruments, indulged in excessive
consumption due to actual or perceived peer pressure, and treated debt as an emergency source
of funds (Penaloza & Barnhart, 2011). Conversely, debtors who using debt and credit the right
way were individuals who sought financial independence from parents and significant others,
had good credit scores and access over minimum lines of credit, reported high levels of self-
discipline, and were wary of the negative repercussions of credit misuse (Penaloza & Barnhart,
2011). Apart from these conceptualizations of debtors as either good, bad, or somewhere in
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 20
between, another key finding was that debt and credit has become normal facets of participation
in American consumer society.
The concepts of good versus bad debt and good debtors versus bad debtors are useful
constructs for approaching this study of students’ attitudes towards debt. Our goal is to determine
whether there is a relationship between students’ attitudes towards debt and whether these
attitudes predict the amounts that students borrow. We also want to determine whether students’
attitudes are related to certain spending behaviors. If we can predict students’ borrowing as a
function of their attitudes, we may also be able to predict whether certain debtors are more likely
to be good or bad debtors by analyzing spending behaviors through the rubric of creating value
or incurring increased long-term expenses.
Lastly, it is useful to consider the ways in which student loan debt differs debts framed as
good and bad. Student loan debt can be a good debt, or those produce net financial gains over
time (Bankrate, 2014), if it facilitates increased financial gains for the debtor. In a bad economy
there are fewer guarantees that holding a graduate degree will yield increased earnings in the
short term. Human capital investment theory and rational choice theory hold that educational
investment decisions are generally good decisions on average (Avery & Turner, 2012; Dynarski,
2008; Volkwein et al., 1998). Yet, it has also been argued that student loan debts are impairing
graduates’ ability to obtain home mortgages (Brown et al., 2014; Korkki, 2014; Neil, 2014;
Ungarino, 2014), despite mortgage being considered a good debt (Bankrate, 2014). It would
seem, therefore, that good student loan debts facilitate increased earnings and subsequent access
to home mortgages. Conversely, bad student loan debt would either not result in increased
income, or preclude home ownership, or worse: both! It is beyond the ken of this dissertation to
determine the long-term effects of student loan debt on both respondents’ incomes and rates of
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 21
homeownership, but this dissertation seeks data on attitudes to explain how much students
borrow and what they do with the money they borrow.
Theoretical Framework
To capture the complexity of the literature as it relates to graduate student debt, we use a
theoretical framework that integrates multiple lenses to better understand graduate student
decision-making: rational choice theory (RCT), human capital theory (HCT), and behavioral
economic (BE) theory. The first lens, rational choice theory (RCT), contributes to our
understanding of student debt by framing borrowing behavior as a coherent pattern of behavior
intended to yield more profitable ends. Put another way, RCT conceptualizes human behavior as
a series of mathematical transactions wherein the output must be equal to or greater than the sum
of the individual parts (Gilboa, 2010). The second lens of human capital theory (HCT) informs
us that borrowing decisions are framed as investments in long-term personal productivity.
Investments in human capital like graduate school are intended to produce positive
changes in what someone knows and these changes are supposed to facilitate novel and more
complex behaviors (Becker, 1962, Coleman, 1988). Lastly, behavioral economics (BE) concedes
that human behavior is predictably irrational (Ariely, 2009). BE theorists assume that human
subjects have limited informational intake abilities and similarly limited cognitive processing
faculties. These limitations affect decision-making and outcomes. Together these lenses provide
slightly different perspectives on human behavior and expand the horizon upon which we hope
to find discernable patterns in students’ borrowing decisions.
Financial assistance for undergraduate and graduate study is provided to reduce the
overall price students pay for their education. Financial aid for higher education goes back to
beginning of American higher education and has evolved with higher education (Heller, 2011).
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 22
As the costs of higher education have increased and the contributions of the state and federal
government have flattened, more students have been pushed towards loans (Archibald &
Feldman, 2011; Ehrenberg, 2000; Vedder, 2004). Some argue that students are borrowing too
much (Avery & Turner, 2012; Baum & Steele, 20101, Dwyer, Hodson, & McCloud, 2012;
Malcom & Dowd, 2012; Sunstein, 2006). Some have gone so far as to characterize current
student lending and borrowing as 21
st
century indentured servitude wherein students learn that
higher education is nothing if not yet another consumer service (Williams, 2008a,b). Lastly,
there is evidence that higher cost institutions ward off students and families from the lower
income quartiles through increased reliance on student loans (Lillis, 2008). If there is one thing
upon which the literature agrees is that students are borrowing more and more frequently; but
why?
Rational Choice Theory
Rational choice theory (RCT) is best known for framing various human behaviors as the
outcomes of cost-benefit analyses (Gilboa, 2010; Hogarth & Reder, 1987). While RCT scholars
differ in their assumptions of human rationality (Gilboa, 2010; Hogarth & Reder, 1987; Wittek et
al., 2013), the core idea behind RCT is that human behavior arises from goals and some form of
calculation regarding whether said goal is possible and if so, whether the goal represents a
quantitative and/or qualitative increase in benefit. This section briefly considers the core
concepts and debates within the RCT literature before turning to an analysis of how RCT has
been employed to understand students’ borrowing behaviors. First, a distinction between strong
and bounded rationality is made. Next, this distinction is couched within the context of the
primacy of individual goals in RCT. This section concludes with the argument that RCT
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 23
enhances our understanding of student borrowing behavior, but offers an incomplete perspective
insofar as the human behavior is known to deviate from the core assumptions of RCT.
RCT researchers vary in whether they assume subjects have complete knowledge of costs
and potential benefits (Wittek et al., 2013). RCT researchers are further divided regarding
whether subjects have limited or unlimited cognitive processing capabilities (Wittek et al., 2013).
The concept of rationality has been regarded as an assemblage of economic models of personal
choice (Gilboa, 2010; Wittek et al., 2013). A more classical construction of RCT holds that
individuals’ expenditure decisions are cognizable as a utility equation consisting of alternative
resource expenditure arguments. The rational agent in this construction optimizes his benefits
over and against his costs (Hogarth & Reder, 1987). The core idea of RCT is that actors have
coherent goals that are quantifiable and capable of being modeled, tested, and analyzed (Gachter,
2013; Gilboa, 2010).
The idea that actors have cognizable goals is a central theme in RCT, but the process by
which goals are constructed reveals a distinction in RCT assumptions. Whereas Taversky and
Kahnemann (1987) inform us that goal framing is the first step towards determining goal
feasibility, the assumption of whether actors have complete or partial knowledge is an ongoing
debate (Wittek et al., 2013). Feasibility has to do with determining whether the choice to deploy
resources makes sense based on available information (Gilboa, 2010). In most cases the social
and human capital of an actor are two limiting variables affecting whether actors have sufficient
information to properly frame decision-making (Cook & Cheshire, 2013; Einhorn & Hogarth,
1987). On the spectrum of strong versus bounded rationality, the literature here espouses a
conservative bounded rationality approach. That is, boundedly rational actors cannot be assumed
to have either perfect information or unlimited cognitive processing power. As such, goals are
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 24
framed with myriad constraints—some observable, others latent—and identifying whether
students’ attitudes towards debt attenuate their borrowing behavior contributes fresh insights into
goal framing.
Scholarship employing RCT as a framework revealed differences in students’ debt
attitudes, persistence, and completions. For instance, Sunstein (2006) considers the implications
of a boundedly rational framework in relation to student borrowing and holds that excessive
borrowing arises from cost neglect, procrastination, unrealistic optimism, self-control issues, and
problems relating to self-control. This coheres with Gandhi’s (2007, p. 137) observation that
myopic loss aversion explains why low-SES students do not want to borrow for funding higher
education. His conclusion is that this is irrational behavior because the outcome—increased
earnings potential—is supposed to offset short-term loan costs.
Similar findings were reported by Callender and Jackson (2005) wherein students’
attitudes and acceptance of student debt varied as a function of SES; low-SES students more apt
to perceive the price of higher education as greater than the perceived benefits. In these three
cases it appears that the rational choice, to invest in higher education, is made less frequently by
students from lower economic quartiles.
RCT framed studies found differences in students’ rates of persistence and post-
baccalaureate planning. Dowd and Coury (2006) saw a significant difference in the persistence
rates of borrowers and non-borrowers; borrowers have decreased odds of persisting year-to-year.
Further, Malcom and Dowd (2012) observed that certain minority groups were more apt to carry
more student debt and that these differences affected graduate school enrollment. The
implication being that institutions seeking to transition more minority students to graduate study
should take steps to reduce the amount of debt these students accumulate as undergraduates
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 25
(Malcom & Dowd, 2012). The choice to borrow for undergraduate education and whether this
choice is perceived as necessary appears related to students’ decisions while in school and what
students plan to do after graduation. RCT holds that the decision to borrow should not influence
attendance because the investment in higher education produces greater salary potential via gains
in human capital (Avery & Turner, 2012; Dynarski, 2008; Volkwein et al., 1998). Yet, the
evidence in the literature indicates that this is not the case for all students. As such, we turn now
to human capital for additional insight, especially in decisions made pertaining to investments
like education.
Human Capital Theory
HCT has been employed to examine students’ borrowing decisions for education and the
findings indicate that borrowing yields increases in personal knowledge, skill, and productivity
(Avery & Hoxby, 2004; Dynarski, 2008; Gandhi, 2007; Volkwein, Szelest, Cabrera, Napierski-
Prancl, 1998). While human capital is described as the accumulation of changes in knowledge
and skill in individuals that facilitates novel and complex behaviors (Becker, 1962; Coleman,
1988), it has also been analyzed as a function of changes in lifetime earnings potential. As such,
the purpose of using human capital theory in empirical inquiry is to examine the relationship
between resource investments in students and the economic and educational outcomes of
postsecondary enrollment and/or completion.
Findings from studies employing human capital theory show positive relationships
between student borrowing and students’ economic productivity as a function of degree
completion (Avery & Hoxby, 2004; Avery & Turner, 2012; Dynarski, 2008; Gandhi, 2007;
Volkwein et al., 1998). For instance, there were observed differences in salary premiums among
college and graduate school dropouts and completers; the former commanding lower average
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 26
salaries later in life in comparison to the latter (Volkwein et al., 1998). Similar findings revealed
that the salary premiums associated with having a college diploma outstrip the average earnings
potential of a high school diploma (Avery & Turner, 2012). Dynarski’s (2008) finding that
decreased costs facilitated increased rates of completions implies that postsecondary attendance
and completion produces outcomes associated with increased average income potential. An HCT
framework therefore seeks to understand students’ borrowing behaviors in relation to potential
earnings capacity.
HCT proposes that students’ borrowing decisions are cognizable as decisions intended to
increase their earnings potential via increased knowledge and the acquisition of new skills and/or
greater skill mastery (Avery & Turner, 2012; Dynarski, 2008; Volkwein et al., 1998). HCT is
also maintains that students’ borrowing represent investments in the marketplace whereby
students’ increased knowledge and skills gain from school attendance produces increased
marketplace efficiency and effectiveness (Avery & Turner, 2012). Following the logic of these
findings prompts Gandhi (2007) to argue that subsidization of higher education is nonsensical
insofar as completions yield increased on-average earnings potential. This argument proposes
that since postsecondary degree holders earn greater average incomes than non-degree holders it
should not matter whether students borrow at all since the reward for completing a baccalaureate
or post-baccalaureate degree is increased earnings potential. Therefore the decision to seek
postsecondary education makes sense and all financial aid is simply a means to an end.
Something HCT has difficulty explaining are the observed differences in resources
invested in certain groups and disproportionate outcomes. While it appears that investments in
students facilitate access and success (Dynarski, 2008; Volkwein et al., 1998), there is evidence
that some groups bear a disproportionate burden in providing their own funding (Dwyer,
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 27
Hodson, & McCloud, 2013; Malcom & Dowd, 2012). Moreover institutional investments in
students appear to be partially driven socioeconomic status (SES; Weiler, 1994) and privileged
access to certain forms of information (Avery & Turner, 2012). The observed differences in
borrowing, receipt of merit aid, and salary outcomes are critical points for which HCT-based
postsecondary research has been criticized.
HCT’s limitations have led to calls for a more integrated framework in which the
influence of various sociocultural factors are included. Perna (2006b) for example, points out
that human capital theory is unable to account for the differences in college choices made by
individuals from different groups, whether in terms of income or race/ethnicity. Tierney and
Venegas (2007) argue that the human capital framework in research has been more useful for
policy-makers rather than for the students and their families who need and use the aid, neglecting
to “speak with, interview, reflect on, or otherwise address the assumptions, beliefs, and concerns
of low-income students” (p. 368).
While both RCT and HCT allow us to conceptualize students’ borrowing behaviors as the
outcomes of cost-benefit analyses, they do not cover the entire spectrum of borrowing decisions
made. Not all students engage in the cost-benefit analysis assumed by RCT and HCT, and
sometimes, student loan decisions may be perceived as irrational or unpredictable. A discussion
of behavioral economics, which taps into concepts of human psychology, provides greater
insight into knowing why students may make the decisions that previously have seemed obscure
or impenetrable.
Behavioral Economics
The field of behavioral economics combines economics with the psychology of human
behavior to conceptualize behavior that does not make sense in an RCT (or HCT) framework.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 28
Ariely (2009) explains that behavioral economics is concerned with “human irrationality…our
distance from perfection” (p. xxix). Behavioral economists assume that humans are boundedly
rational agents whereby our decision-making processes are thwarted by 1) incomplete
knowledge of all variables and 2) limited cognitive processing power (Sunstein, 2006; Wittek et
al., 2013). These assumptions help behavioral economists to understand behaviors that are
characterized as irrational under an RCT framework. An example of behavioral economics
research is submitted to clarify the means and ends of behavioral economic theory.
When studying college-age populations, two words have been found to encourage
increased participation in social science research: free beer. Frederick and Ariely (2006)
illustrate the predictably irrational behavior of human subjects by showing how humans’
subjective experiences are mediated by expectations. Their experiment tested whether
information shaped subjects’ experiences of free samples of beer. Would being told the free
sample of beer had a “special ingredient” encourage favorable evaluations? The special
ingredient in this experiment was several drops of balsamic vinegar placed in an ordinary
domestic ale. Subjects were exposed to three testing conditions: 1) a blind condition wherein the
vinegar was not mentioned; 2) disclosure of the vinegar prior to tasting; and 3) disclosure of the
vinegar after tasting. Statistically significant preferences for the beer with vinegar were observed
when the disclosure occurred after sampling. The distribution of preferences in the blind
condition was normally distributed and the distribution of preferences in the disclosure prior to
tasting condition suggested people do not like vinegar in their beer. The finding here is that our
expectations shape our experiences thereby exhibiting that utility functions are malleable.
Loss aversion is another behavioral economics concept that has been observed affecting
postsecondary students. Loss aversion describes the phenomenon of individuals fearing loss
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 29
more than valuing gains (Ariely, 2006; Tversky & Kahneman, 1991; Gandhi, 2007). Loss
aversion describes behaviors where actors attempt to keep as many options open as possible,
such as buying computers loaded with unnecessary options or warranties for new goods that are
expensive relative to the cost of the good (Ariely, 2006). In keeping options individuals prioritize
that which they might lose over any possible gains (Shin & Ariely, 2004). Such behaviors are
inscrutable under an RCT framework, but behavioral economics informs us that this kind of
conduct is expected and can be corrected.
Debt aversion is complex insofar as it varies in frequency and intensity as a function of
multiple variables. Burdman (2005) argues that debt aversion presents a dilemma for higher
education because though educational loans open lanes of access to students, some groups of
students are less likely to use loans. Gandhi (2007) counters that the long-term benefits of
completing a baccalaureate degree ought to outweigh the short-term losses accrued by
borrowing. The problem, Gandhi continues, is that students’ valuations of their education shifts
as a function of temporal proximity. That is to say, if one were offered the choice between a
large reward in the distant future and a smaller reward in the near future, preferences shift
towards the smaller reward more often than not (Gandhi, 2007; O’Donoghue & Rabin, 1999).
Debt aversion, as a form of loss aversion, keeps options like working more hours and/or taking
fewer courses per term on the table. The problem for students is that these behaviors are
associated more often with attrition (Astin, 1993;Tinto, 1987) than completion. As such, such
students are not maximizing their utility function of college attendance.
Loss aversion and debt aversion arise from human tendencies to be impatient and not
perceive larger long-term gains. The good news is that increases in information availability tend
to thwart this behavior, but the bad news is that this information tends to be privileged, time-
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 30
consuming, and expensive (Tierney & Venegas, 2009; Venegas, 2006). Dynarski and Scott-
Clayton (2006) synthesize several concepts of behavioral economics to find that the complexity
of the federal loan procurement process is a barrier to certain populations that the program was
designed to help. The point here being that financial aid policy, specifically as it relates to
student loans, predicated on one-size-fits-all assumption misses the points illustrated by
behavioral economic research: human behavior, under certain conditions, is irrational, but
correctable. Following this discussion of the broader theories informing our study, the next
section outlines the literature on relevant attitudes and behaviors towards debt, as well as
specifically student loan debt.
Attitudes and Behaviors Towards Debt
Attitudes
This dissertation examines whether master’s students’ attitudes towards debt are related
to subsequent behaviors such as amounts borrowed. An important distinction needs to be made
regarding the nature of attitudes and their relationship to behavior. This section defines what is
meant by attitude. We consider how attitudes were employed in prior studies of the relationship
between students’ borrowing behaviors and their attitudes. Our goal is to illustrate that attitudes
have distinct features and structures by which they are cognizable and measureable and are
appropriate constructs to measure in this study.
We are measuring attitudes in this dissertation because we want to establish whether
master’s students’ debt attitudes are related to variance in the amounts of federal loans they
borrow and subsequent behaviors after borrowing student loans. Attitudes are theorized to shape
intentions and affect behavior (Ajzen, 1991) and are related to the socio-ecological forces in our
lives (Cooper et al., 2013). If attitudes partially affect what we do and are shaped by forces
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 31
within and beyond our control, there are implications for student debt that extend beyond matters
of graduate school pricing. Let us first settle the definition of an attitude before turning towards a
discussion of whether attitudes and student loan borrowing are related.
Several pieces of scholarship help us to triangulate on the meaning and scope of attitudes.
An attitude is held to be evaluative (Crano, Cooper, & Forgas, 2010) and is an expression of
preferences for or against specified objects (Fishbein & Ajzen, 1974; Johnson & Boynton, 2010).
Eagly and Chaiken (1995) offer the definition that attitudes are evaluative responses to attitude
objects such as a person, a behavior, or an idea. Moreover, attitudes provide individuals with
shortcuts for evaluating social contexts and also function to guide long-term planning
(Ledgerwood & Trope, 2010). In sum, attitudes represent our judgments about specific objects
and appear to provide indicators of our intended short-term and long-term behaviors. Attitudes
are also defined as a function of valence.
Attitude valence refers to the attractiveness of a class of attitude objects or specific
attitude objects within a class (Ledgerwood et al., 2010). Let us consider the concept of valence
using vegetables as an exemplar of valence. For instance, assuming one’s attitude valence
towards vegetables is positive, this person is likely to agree that they think they vegetables, in
general, are good. However, individuals also have specific attitude valences that may be opposite
the general attitude valence. It is perfectly normal for some individuals to have a negative
attitude valence towards tomatoes. They just do not like tomatoes and would likely agree with a
statement of “I find tomatoes unpalatable.” One’s attitude valence for vegetables will likely
predict that he/she eat a given range of vegetables per week, but negative attitude valences for
tomatoes are likely to predict total tomato avoidance. In this way we being to appreciate that
attitudes range from general to specific (Eagly & Chaiken, 1995) and that within this range there
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 32
is room for variance when it comes to attitude valence. Thus, it is held that attitudes can be
useful variables for predicting human behavior, but only to the extent that we reckon the extent
to which attitudes change and how stable attitudes are over time.
Attitude malleability and stability important features of attitude structures and they have
implications for this dissertation. It appears that attitudes can be stable, but they are also
malleable. For example, research on stereotypes has shown that attitudes are malleable
depending upon context (Garcia-Marques, Santos, & Mackie, 2006). Research testing the
malleability and stability of prejudices has revealed that individuals’ prejudices are malleable as
a function of disruptions to groups’ social contexts (Dasgupta & Greenwald, 2001). This coheres
with research findings showing that attitude strength—a function of general attitude strength,
internal consistency, and extremity—predicts attitude stability over time (Prislin, 1996). Our
preferences and prejudices exert measureable effects on our behavior (Walther & Langer, 2010),
but these attitudes can change depending on how they were formed and the conditions that
activated them. This suggests that attitudes can be changed via pedagogy and through stimuli
modifications and changes to individuals’ social settings (Ambrose et al., 2010; Mayer, 2011).
If students’ debt attitudes are stable, they may be better predictors of how much students
borrow and what they purchase with borrowed monies. However, if students’ debt attitudes are
malleable and context specific, it may be that certain triggers result in behavioral responses
whereby borrowing patterns become more variable. Research has shown a relationship between
attitude specificity and behavioral predictions such that the more concrete the attitude
(preference for or prejudice against) towards a specific object, the more likely that attitude will
be predictive of a behavior (Eagly & Chaiken, 1995; Fishbein & Ajzen, 1974). This is echoed in
research on attitudes towards environmentalism finding that attitude-behavior congruence was
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 33
more likely when attitude measures were specific to a behavioral criterion (Weigel & Vernon,
1974). This implies that inquiring about students’ attitudes towards student loans in particular
ought to be a better indicator of students’ loan borrowing behaviors.
This project examines master’s students’ student loan debt attitudes to determine whether
these attitudes are related to how much they borrow. Two prior studies of students’ attitudes and
borrowing behaviors both revealed a relationship between attitudes, borrowing frequencies, and
borrowed amounts (Davies & Lea, 1995; Haultain et al., 2010). These studies treat attitudes
towards debt as preferences for debt or an aversion to debt. Davies and Lea (1995) refer to debt
attitudes as being either pro debt or anti debt (p. 670). Haultain et al. (2010) use similar language
and refer to debt attitudes as “Fear of Debt” or “Tolerance of Debt” (p. 325). Haultain et al.
(2010) found that respondents with low debt fear measurements and higher tolerance of debt
measurements were predictive of students’ educational debt loads. Davies and Lea (1995) found
a similar relationship between pro-debt attitudes and students’ reported debt loads. The evidence
from these two studies implies that debt attitudes are related to students’ debt loads. These
findings support the appropriateness of surveying students’ debt attitudes, specifically students’
attitudes towards federal student loans, and can provide insights into whether they borrow federal
loans and why they borrow in varying amounts.
This dissertation seeks to determine whether students’ attitudes towards debt influences
borrowing behaviors. As such, it follows that we discuss what the literature has to say on this
topic. Students’ attitudes toward debt has been identified as a gap in the literature (Callender &
Jackson, 2005; Dowd, 2008; Cooper, 2013) and prior research has recommended timely and
intentional interventions aimed at providing information to students that have been observed to
be less likely to seek or obtain federal financial assistance (Tierney & Venegas, 2009; Venegas,
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 34
2006). Further, given the econometric focus of HCT, RCT, and behavioral economics it is
important to consider how attitudes factor into students’ borrowing behaviors.
An influential article by Davies and Lea (1995) has been at the fore of the dialogue on
students’ attitudes and student debt. Their Attitudes to Debt scale is widely cited and consists of
a 32-item survey inquiring about students’ general attitudes towards debt and also delves into
specific questions about students’ attitudes towards postsecondary educational debt. A limitation
of the Attitudes to Debt scale is the low Cronbach’s alpha observed in their sample (α=0.79). A
second critique of Davies and Lea’s work has been that their scale is unidimensional and
students’ attitudes may not be unidimensional (Haultain, Kemp, & Chernyshenko, 2010). Davies
and Lea (1995), however, did find that as students’ observed levels of debt increased their
attitudes towards debt become more tolerant.
Haultain et al. (2010) also survey students’ attitudes toward debt with the goal of
expanding upon Davies and Lea’s (1995) work. Their design employed three separate studies of
three distinct samples. Study one examined the debt attitudes of 1,252 New Zealand high school
seniors using exploratory factor analysis to test whether students’ debt attitudes are
unidimensional or multidimensional. Findings from study one indicated a multidimensional
factor structure that accounted for 43% of the observed variance in the survey and neither factor
was correlated (Haultain et al. 2010, p. 325). Study two is a longitudinal analysis of the debt
attitudes of 125 first-year New Zealand college students. Two factors were observed to explain
over 30% of the variance and neither factor was correlated. The final study followed-up with 452
of the original 1,252 seniors surveyed in study number one. This time confirmatory factor
analysis validated the original two-factor scale (RMSEA = .07) and found correlations between
debt attitudes and tertiary educational plans. In sum, this research expanded upon that of Davies
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 35
and Lea’s (1995) by furnishing evidence that students’ attitudes towards debt are better
conceptualized as multidimensional constructs.
Our study also builds upon a previous dissertation examining how students made their
decisions to incur graduate student debt (Cooper et al., 2013). This qualitative study interviewed
over 100 graduate students in primarily Southern California higher education institutions across
different professional graduate programs. The major themes the study revealed were: 1) many of
the graduate student respondents did not know the terms and conditions of their student loans,
including repayment; 2) if anyone else was consulted at all, most of the graduate students sought
guidance on student loans from trusted personal contacts, rather than staff from the financial aid
office or through general research; and 3) a graduate degree was pursued because it was
perceived as a worthy return on investment (Cooper et al., 2013). Our inquiry builds upon the
work of Cooper et al. (2013) by examining student debt using an econometric focus. Whereas
Cooper et al. (2013) framed their analysis using Bronfenbrenner’s social-ecological model, our
framework builds upon their work by incorporating theories of economic decision-making
(namely RCT, HCT, and, BCT). Cooper et al. (2013) informed our research design with findings
that revolved around core economic themes such as the return on investment on education. We
found the Cooper et al. (2003) study helpful in its findings specific to graduate students and
noted that it affirmed the need for our study to be steeped in economic theory to understand
attitudes and behaviors related to student loan borrowing. In our survey instrument (see
Appendix A) outlined in chapter three, we borrow a construct from this study to probe students’
level of student loan knowledge, along with their attitudes towards debt and student loan debt.
The implication for this dissertation, therefore, is that our instrument be designed to
capture multiple dimensions of debt attitudes and behaviors. Davies and Lea (1995) provided an
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 36
early, reliable, and valid measure of student attitudes towards debt. Haultain et al. (2010)
subsequently refined our understanding of the factor structure of students’ debt attitudes via EFA
and CFA on a larger sample with a longitudinal design element. Finally, Callender and Jackson
(2005) have furnished evidence that replicates findings from prior research (Davies & Lea, 1995;
Haultain et al., 2010) and show that debt attitudes are variable as a function of students’ SES.
Cooper et al. (2013) examine how the level of knowledge regarding student loans and various
behaviors related to student loan borrowing may result in differential loan debt. As such, our
instrument must be designed to capture the multidimensionality of students’ debt attitudes and
behaviors while also controlling for observable differences in students’ demographics and
backgrounds.
Behaviors
While primarily examining which attitudes may be linked to student debt, this
dissertation project also necessarily explores which behaviors may have a relationship to
masters’ student debt loads. The preceding section provides the groundwork for why and which
attitudes are probed in relation to student debt; this section outlines which behaviors are included
in our study.
Behavior in our study refers to actions in response to particular situations or stimuli.
Drawing from our predominantly economic theoretical framework, we examine how there are
patterns of human behavior in response to specific situations and stimuli. Furthermore, we
assume that we will find patterns of behaviors in relation to student loan debt. It is obvious that
certain behaviors (e.g. the decision to take out a certain amount in student loans) are important in
our analysis of what factors influence student loan debt. Additionally, it also follows that we
include other related behaviors in our study, drawing from the literature in the field.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 37
In our study, we operationalize behaviors as follows: 1) student loan information, 2)
graduate student loan information seeking, 3) financial behavior, and 4) spending compulsion.
The first factor, student loan information, ascertains whether student loans were taken out for
graduate school and the amount. This is an obvious behavior that establishes the relevance and
level of student loan debt; students who did not take out loans from graduate school are excluded
from our study. Next, the information seeking factor is borrowed from a previous dissertation
study (Cooper et al., 2013) and is designed to find out what sources (institutional, online, and/or
friends/family) and to what extent these sources informed the borrowing decision. In the Cooper
et al. (2013) study, the majority (close to 75% of 116 participants) sought information on student
loans from trusted personal connections, defined as family members, friends, or mentors.
Interestingly, a majority (70%) did not seek information on student loan borrowing from the
institution’s financial aid office, with many participants reporting that the information needed
was not readily accessible, and that institutions were not proactive in contacting them.
Furthermore, almost a third (27%) did not speak with anyone regarding the student loan
borrowing decision. While half of this 27% conducted research online before borrowing, the
other half made the decision to take out student loans with little or no research. First generation
and Latino students who are disadvantaged in their access to knowledge about college in general,
often do not have the information to make the best decisions regarding financial aid and student
loans (Cooper et al., 2013; O’Connor, Hammack & Scott, 2009; Tierney & Venegas, 2009;
Venegas, 2003). Overall, Cooper et al.’s (2013) findings are consistent with prior research with
findings indicating that students who are reluctant to seek out pertinent financial aid information
from institutional agents instead rely upon personal contacts.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 38
Third, the financial behavior factor is informed by literature on financial literacy and the
items taken from the Inceptia (2013) and Money Matters on Campus (2014) surveys. These
behaviors are related to how people manage their everyday financial transactions and how
knowledgeable they are about basic financial concepts like compound interest, credit reports, and
budgeting. Research shows that consumers who are more financially literate and possess greater
knowledge of financial transactions are more likely to exhibit financially prudent behavior (Perry
& Morris, 2005). In an analysis of a national sample of 1,000 Americans on debt literacy,
financial experiences, and level of indebtedness, Lusardi and Tufano (2009) observed that people
with lower levels of debt literacy, even after controlling for demographic factors, felt
overindebted (self-reported) and exhibited negative financial behaviors. These negative
behaviors included paying high credit card fees and finance charges. Some studies have
replicated this finding for college students, with a higher level of financial literacy linked to the
greater likelihood that students will exhibit positive financial behaviors (Chudry, Foxall, &
Pallister, 2011; Gutter & Copur, 2011). For example, data from a study of nearly 16,000 college
students from fifteen campuses across the US suggested that positive financial behaviors were
related to financial well-being (Gutter and Copur, 2011). Additionally, in a literature review of
student loan default, studies show the efficacy of loan counseling and general consumer financial
education programs in reducing student loan default rates (Gross et al., 2009).
Finally, the spending compulsion factor is from the Money Matters on Campus (2014)
survey and captures the spending behavior of the participant. Spending compulsion is the
inability to control purchasing behavior (Brougham et al., 2011). Spending compulsion is a
financial behavior, but one that is considered risky financial behavior (Brougham et al., 2011;
Gutter & Copur, 2011; Roberts & Jones, 2001). Compared to the general population, college
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 39
students are more likely to compulsively spend (Brougham et al., 2011). Furthermore, this
compulsion is linked to negative outcomes, ranging from the psychosocial, including guilt,
anxiety, and problems with family and friends, and excessive debt (Brougham et al, 2011;
Roberts & Jones, 2001).
Our final section looks at such demographic and background variables that may influence
graduate student enrollment and borrowing decisions. These factors, while not the focus of our
study, are included as we must acknowledge the complexity of variables that impact the decision
to enroll in and finance graduate education.
Key Background Variables in Graduate Student Enrollment
To understand why students choose to borrow to finance their graduate education, it is
necessary to determine what factors may contribute to the decision to pursue post-baccalaureate
education. There are exogenous and endogenous factors related to student borrowing for
financing higher education. Our study probes for patterns in borrowing as a function of
endogenous characteristics, seeking to find out whether certain internal personal features affect
the distributions of student borrowing. This is in contrast to students’ exogenous characteristics,
which explains student borrowing as a function of observable and unobservable external
phenomena (Dowd, 2008). The implications of asking who borrows are twofold. First,
approaching student borrowing as an endogenous factor beckons us to disaggregate data to
unearth latent inequalities in borrowing behavior. Second, it encourages us to examine the
factors associated with these observed patterns. Therefore it is necessary to examine the factors
known to be related to patterns in students’ borrowing behavior.
While our study focuses on endogenous rather than exogenous factors contributing to
student loan borrowing for graduate school, it is nonetheless important to acknowledge that
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 40
larger, external factors bear on the decision to enroll and finance post-baccalaureate degrees.
Perhaps the most relevant of these exogenous factors is the general condition of the economy.
Larger external factors like the economy and labor market conditions may influence graduate
student enrollment. When labor market conditions are unfavorable with higher rates of
unemployment, students are more likely to attend graduate school (Zhang, 2010).
Pulling from research primarily conducted on undergraduate students, we know that
student loan borrowing decisions are influenced by myriad factors. Among these are
race/ethnicity (Malcom & Dowd, 2012; Volkwein et al., 1998), sex/gender (Amaury, Barlow, &
Crisp, 2006; Dwyer, Hodson, & McCloud, 2013); SES (Avery & Hoxby, 2004; Burdman, 2005;
Callender & Jackson, 2005; Lillis, 2008), and choice of major and career (Rothstein & Rouse,
2010; Youngclaus, Koehler, Kotlikoff, & Wiecha, 2013). Debt aversion is known to be related
with intersecting statuses such as first generation, minority, and low SES and has been shown to
negatively impact students’ willingness to borrow (Burdman, 2005; Eckel, Johnson,
Montmarquette, & Rojas, 2007). While educational borrowing is affected by many variables,
educational borrowing also influences students’ trajectories into, through, and out of
postsecondary education.
We examine literature specific to graduate students to determine the endogenous factors
relevant to our study of graduate student enrollment and student loans. While generally the same
as those important to undergraduate students, these factors may influence the decision to attend
and finance graduate school in different ways. The factors we analyze for our graduate student
study include student demographic factors, the choice of type of undergraduate institution
attended, undergraduate academic performance, and the amount of accumulated undergraduate
debt.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 41
In a discussion of the demographic variables that are important at the graduate student
level, a consideration of parental educational background and race/ethnicity is necessary. More
recent research seems to indicate that higher levels of parental education positively influence
graduate enrollment. While two foundational studies on graduate students from Mare (1980) and
Stolzenberg (1994) did not find any effects of family background on students’ post-baccalaureate
educational decisions, Mullen et al.’s (2003) study found that as the research shows in
undergraduate enrollment, graduate enrollment indeed was linked to parental levels of education.
The Mullen et al. (2003) study utilized a larger sample than the Mare (1980) or Stolzenberg
(1994) studies. It used a nationally representative sample of over 10,000 students drawn from a
dataset of 1.1 million students from the Baccalaureate and Beyond Longitudinal Study (B&B).
Ultimately, they conclude that parental education levels affects enrollment in post-baccalaureate
programs, albeit indirectly, even controlling for various factors, including standardized test
scores, academic performance, and undergraduate institutional characteristics (Mullen et al.,
2003). In general, students who have highly educated parents are most likely to enroll and
graduate from selective, research, or liberal arts colleges, which in turn, makes these students the
most likely to apply and gain admission into a post-baccalaureate program (Mullen et al., 2003).
In addition, race/ethnicity has been found to influence graduate student enrollment, although the
effects are varied and dependent on the racial/ethnic group analyzed. For example, one study
found that African Americans and Latinos were more likely to pursue graduate school than their
White counterparts. Zhang’s (2010) study of graduate students, utilizing the Baccalaureate and
Beyond Longitudinal Study (B&B: 97) data, found that African Americans were more likely
than Whites to enroll in a post-baccalaureate program. She also found that Latino students from
public colleges more likely than Whites to enroll in a post-baccalaureate program. However, the
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 42
sample she analyzed had very few African Americans and Latinos in the sample (5.3% and 3.7%
respectively), which prohibits greater generalization about these groups and the propensity for
pursuing graduate school. Although data were collected in 1997 to obtain post-college outcomes,
the sample had finished college in the academic year 1992-1993, which the author herself points
out was a different era in which graduate students did not have to rely upon student loans as
greatly as they must now (Zhang, 2010). Using a dataset that oversampled for minorities in order
to find racial/ethnic difference, Malcom and Dowd (2012) found mixed results pertaining to
undergraduate debt of STEM majors and subsequent graduate school enrollment. While overall
they found that undergraduate student debt was negatively linked to graduate school enrollment
for all ethnic groups, they found that of STEM undergraduates with heavy debt loads ($35,001 or
more), Latino and Whites were less likely to enroll in graduate school (although with the
important caveat that these findings were not statistically significant), while African Americans
and Asians were undeterred from enrollment. Ultimately, the mixed findings of various studies
show that more research should be conducted to better understand how different groups react to
student debt.
The undergraduate institution type is a factor in graduate student enrollment. College
graduates from selective, research, and private liberal arts are more likely to attend graduate
school than from non-selective or public institutions (Mullen et al., 2003; Zhang, 2010). Mullen
et al. (2003) show that students who have attended more selective, liberal arts, and research
institutions are more likely to attend graduate school. Mullen et al. (2003) points that often
students from higher SES or highly educated families are steered towards these types of
institutions, which also make them predisposed to be able to afford or aspire to graduate school.
Zhang (2010) found that undergraduates who have attended private school are far likelier than
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 43
their public school counterparts to attend graduate school. This may be in part due to the
students’ greater tolerance of student debt, having a greater likelihood than their public school
counterparts of having already incurred debt, and perhaps in part due to a belief that graduate
school is worthy of investment.
Prior academic performance, including choice of college major, significantly predicts
graduate school enrollment. Higher academic ability, including college GPA and standardized
test scores, are linked to a greater likelihood that students will attend graduate school (Mullen et
al, 2003; Zhang, 2010). Students with arts or science majors as compared to professional majors
are more likely to pursue graduate studies (Mullen et al., 2003). Academic performance is an
indication that the student is most likely better prepared for graduate school, and thus possesses
greater human capital. Compounding this effect is that students with greater human capital are
also more likely to view graduate school as a further investment in their human capital.
Whereas the receipt of merit aid is linked to increased enrollment odds and persistence
(Dynarski, 2003, 2008; Franke, 2014; Hossler et al., 2009; Lillis, 2008; Malcom & Dowd, 2012;
Weiler, 1994), the prospect of having to assume educational debt is known decrease the odds of
enrollment for some groups (Avery & Hoxby, 2004; Dowd & Coury, 2006). From the onset,
people who are debt averse may be winnowed out from the higher education pipeline, creating a
population of students who may be more open to acquiring student loan debt if they opt for
graduate education.
Undergraduate student debt loads may impact college outcomes. As such, we need to
briefly examine the role of undergraduate student debt to undergraduate student persistence. The
research is inconclusive, as the effect of loans on students’ persistence has been shown to be
variable among different groups. For instance, Perna (1998) found that borrowing did not
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 44
promote persistence, but St. John, Kirshtein, and Noell (1991) reported that loans did facilitate
persistence. It must be noted that both of these studies examined samples from the 1980s when
college prices were not at the high levels presently observed. A more contemporary inquiry
found that an increase in year-to-year student loan balances was associated with decreased rates
of persistence (Amaury et al., 2006). Students’ financial need, how this need is met (grants vs.
loans), and how it influences access and persistence is influenced both by endogenous
characteristics and the college experience itself. In the next section, we observe how the effects
of undergraduate student loan debt on post-baccalaureate planning.
We examine the research on undergraduate student debt on graduate enrollment, which
seems to have the largest body of research pertinent to graduate students. Evidence of the effects
of student debt on graduate school plans and post-baccalaureate plans shows that the effects of
debt have implications for students’ plans on the order of years and decades. Debt loads have
been shown to influence students’ post-baccalaureate plans including graduate school and
careers. For example, Rothstein and Rouse (2010) did not observe educational debt affecting
students’ plans to attend graduate school, but they did observe significant shifts in students
career planning during the undergraduate years. This finding is replicated in a sample of medical
school students where the amounts of educational debt were observed to influence career path
planning (Youngclaus et al., 2013). Malcom and Dowd (2012) observed that average and above
average levels of borrowing were negatively correlated with graduate school enrollment. These
observations replicated earlier findings (Weiler, 1994) wherein increased amounts of student
loans were negatively correlated with graduate school aspirations. The evidence implies that over
the long-term, students’ perceptions of the benefits of additional schooling may override their
cost-benefit analyses.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 45
The research on the effects of undergraduate student debt on graduate enrollment is
inconclusive. Depending on the sample of graduate students utilized, the studies find varying
effects in terms of graduate enrollment (Malcom & Dowd, 2012; Millett, 2003). Millett (2003)
analyzes dozens of studies conducted over sixteen years on the relationship of undergraduate
student debt on graduate enrollment and finds no consensus in the research. The lack of
consensus, she believes, is due to flawed research design and methods in which study samples
were not nationally representative and the reliance on secondary data which were not designed to
capture information on debt or access to graduate school (Millett, 2003). Malcom and Dowd
(2012) criticize previous studies in a similar manner, emphasizing that the research methods
often did not address endogeneity and self-selection biases in their samples.
Some studies find that undergraduate indebtedness discourages graduate school
enrollment (Malcom & Dowd, 2012; Millett, 2003; Mullen et al, 2003; Zhang, 2010). The usual
explanation is that students who already have student loan debt are loath to incur further loan
debt for graduate school until their undergraduate loans have been paid down. Millett’s (2003)
own study, based on a sample of almost 2,000 students who completed the Baccalaureate and
Beyond Longitudinal Study (B&B: 92-93), found that undergraduate indebtedness adversely
impacted enrollment in a post-baccalaureate program. These students had initially aspired to
pursue a doctoral degree, but within a year of earning their bachelor’s degree, 41% did not apply
to any post-baccalaureate program, which includes first professional, masters, or doctoral
programs (Millett, 2003). Malcom and Dowd’s (2012) study, which focuses on graduate school
enrollment of STEM undergraduates disaggregated by race/ethnicity, finds that for all
racial/ethnic groups, prior undergraduate debt reduces the changes of attending graduate school.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 46
Other studies show that undergraduate indebtedness has no impact or is even positively
linked to graduate school enrollment (Malcom & Dowd, 2012; Millett, 2003; Mullen et al, 2003;
Zhang, 2010). For example, Kim and Eyermann’s (2006) study indicates that after the Higher
Education Amendments of 1992, which raised federal student loan caps and increased loan
eligibility, student loan debt was positively and significantly linked to graduate school. The
rationale for this finding is that student loan debt has become normalized and is an accepted
reality of increasing one’s human capital, given the rising cost of higher education and the shift
of financial aid from grants to loans.
Furthermore, studies offer mixed results, depending on how the sample is disaggregated.
Zhang (2010), for example, finds differential effects of accumulated undergraduate student debt
on graduate school choices. For those who attended public institutions, accumulated
undergraduate debt had a negative and significant effect on graduate school attendance,
particularly for doctoral, MBA, and first professional programs (Zhang, 2010). The population of
students who have undergraduate debt from attending a public undergraduate institution may be
less likely from the beginning to even enroll in graduate school, meaning students who have
chosen to enroll in a master’s program may be less debt-averse than the general population.
However, for students who graduated from private undergraduate institutions, accumulated
undergraduate debt did not have an effect on graduate school attendance (Zhang, 2010). Upon
closer examination, undergraduate debt had no effect on the choice of a master’s or doctoral
program, but actually had a positive and significant effect on MBA and first professional
program enrollment, which tend to require more time and are more expensive. This finding for
students who attended private undergraduate institutions may be explained by these students’
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 47
initial investment in their undergraduate degree, and the belief that graduate student loans are a
further investment in their human capital (Zhang, 2010).
The literature review on factors in graduate student enrollment tends to illustrate that
there is no conclusive evidence that these factors concretely increase or decrease the odds of
graduate student enrollment. In fact, this highlights in many ways why more studies are needed
to determine what factors may be influential in these decisions. Our study, by controlling for
these factors while exploring what attitudes are influential in student borrowing behaviors, may
help to fill in this gap in the literature.
Model of Attitudes and Student Borrowing

The conceptual model below (see Figure A) visualizes how the theoretical framework
informs the survey instrument created to capture the attitudes and behaviors that may influence
student loan debt. Rational choice, human capital, and behavioral economic theories explain the
formation of debt attitudes, which we posit are key to understanding borrowing decisions. These
relevant attitudes and behaviors drawn from these theories are measured by the constructs in our
survey instrument: Beliefs about Knowledge across Financial Competencies (KFC), Loan
Information Seeking behaviors (LIS), Financial Efficacy (FE), Financial Behaviors (FB), Debt
Aversion (DA), Spending Compulsion (SC), along with demographic and other individual
factors (Demo in the conceptual model) that are salient in the student loan decision, as show in
this literature review. This conceptual model is explained in further detail in Chapter 4.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 48

Figure A. Model of attitudes and the relationship to student loan borrowing. This figure
illustrates our theoretical framework and how it relates to the multiple constructs of the survey
instrument and ultimately student loan amount.

Next, Chapter 3 discusses the research design and how it will help answer our two
research questions. First, we want to determine whether there is a relationship between graduate
students’ attitudes towards borrowing and how much they report borrowing to fund their
graduate education. Second, given a relationship, we seek to ascertain what factors affect
Debt
Attitude
Formation
Rational
Choice
Theory
Human
Capital
Theory
Behavioral
Economic
Theory
KFC LIS FE FB DA
SC Demo
Student Loan
Amount
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 49
increased or decreased borrowing. Our research design blends together elements from three
seemingly disparate frameworks to ascertain the extent to which students’ attitudes are a function
of rational human capital investments and irrational behaviors. It is hypothesized that differences
in students’ attitudes are explained both by rational cost-benefit analyses and irrational decisions
that are precipitated by limits to human cognition and information processing. The research
design helps answer these research questions and establish the validity and reliability of our
measurement scales.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 50
CHAPTER THREE
METHODOLOGY
This study was designed to examine the attitudes that graduate students have towards
borrowing for financing their education. Chapter 2 provided a review of the literature that
revealed that there is a need to better understand how graduate students approach their decisions
in borrowing for their graduate education. While there is existing research on attitudes towards
student debt and decision-making, there is not enough known about a significant sector of
debtors: graduate students. Therefore this study aims to contribute to the literature on graduate
student debt by further probing into the processes of their decision-making to borrow.
A quantitative approach was used for this study in the attempt to answer our research
questions. The quantitative approach is most appropriate, given that we sought to identify the
factors that influence outcomes (Creswell, 2009), i.e. graduate school financing decisions.
Furthermore, a quantitative approach also provides a way to further extend the qualitative study
conducted by a previous research team that focused on understanding graduate student attitudes
towards student loans (Cooper et al., 2013). This qualitative study provided insight into three
themes that influenced students’ decisions when incurring graduate school debt: 1) student
understanding of loan debt and borrowing decisions; 2) advice-seeking on financial graduate
education; and 3) perceptions of return on investment (Cooper et al., 2013). Our study builds
upon this initial research into the factors that influence student decisions on graduate school
financing with a quantitative study that deepens an understanding of which factors most
influence graduate student borrowing.
As an overview of Chapter 3, we will first define the research approach, positioning it in
terms of our research questions and our theoretical framework. Next we will define the sample,
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 51
including the site and specific masters programs represented by our participants. We will then
articulate the survey instrument, which was created from existing survey instruments and
adapted for our study. Lastly, we cover data collection and data analysis.
Definition of the Research Approach
The purpose of this inquiry was to determine whether differences in master’s students’
attitudes affects whether they borrow federal student loans and, if so, how much. The two
research questions that guide this study are as follows:
1. Is there a relationship between master’s students’ attitudes towards debt and the
amounts they borrow in student loans?
2. If attitudes predict borrowing patterns, what factors are associated with increased
borrowing and decreased borrowing?
The rising costs of higher education (Archibald & Feldman, 2011; Ehrenberg, 2000;
Vedder, 2004) have been linked to increases in the amounts that undergraduate students are
borrowing. Yet others have noted that graduate students are borrowing at rates that are
disproportionately high in comparison to their overall participation in U.S. higher education
institutions (McCann, 2013). Our research endeavors to fill a hole in our knowledge of the
factors affecting master’s students’ decisions to finance their graduate programs using federal
student loans. In so doing we hope to provide financial aid professionals and other enrollment
managers with insights into why master’s students borrow and why some groups are more apt to
borrow than others. Such knowledge will be useful for improving financial aid and pre-
enrollment counseling and hopefully produce benefits for institutions and borrowers alike.
Considering that most of the literature on student borrowing focuses on undergraduate
students, we felt it important to study master’s students’ borrowing patterns. Prior research has
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 52
established that race/ethnicity (Malcom & Dowd, 2012; Volkwein et al., 1998), sex/gender
(Amaury, Barlow, & Crisp, 2006; Dwyer, Hodson, & McCloud, 2013), SES (Avery & Hoxby,
2004; Burdman, 2005; Callender & Jackson, 2005; Lillis, 2008), and choice of major and career
(Rothstein & Rouse, 2010; Youngclaus, Koehler, Kotlikoff, & Wiecha, 2013) influence
undergraduate students’ borrowing frequency and the amounts they borrow. We hypothesize that
many of these same factors influence master’s students’ borrowing decisions, but our approach
tries to explain any observed variance in borrowing as a function of three interacting theoretical
constructs.
Our theoretical framework consists of three distinct bodies of work that when combined
can help explain master’s students’ borrowing decisions. First, human capital theory (HCT;
Zhang, 2010) holds that student borrowing decisions are influenced by students’ beliefs that the
long-term benefits of deficit financing a graduate degree outstrip the short-term costs of loan
repayment and opportunity costs of attending graduate school. Prior research using HCT
concepts has generally held that insofar as students borrow money to fund their education they
generally reap increased financial rewards over time compared to those that do not borrow
(Avery & Hoxby, 2004; Avery & Turner, 2012; Dynarski, 2008; Gandhi, 2007; Volkwein et al.,
1998). One of our goals, therefore, is to measure whether master’s students’ borrowing decisions
are influenced by such considerations and to establish the magnitude of human capital
investments on borrowing.
Secondly, we employ rational choice theory (RCT) to determine whether students
calculate the costs and benefits of borrowing federal loans for their master’s programs. RCT
informs our methodology by examining students’ cost-benefit analyses. It follows that if one of
the benefits of a graduate degree is increased salary potential, students’ borrowing will be
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 53
influenced by perceived gains over and against their present financial situations. As such our
instrument captures data about respondents’ income, sources of income, and information seeking
behaviors. RCT maintains that individuals’ decisions are quantifiable cost-benefit arguments
(Hogarth & Reder, 1987) where the ends justify the means (Gachter, 2013; Gilboa, 2010).
Lastly, we expect that certain amounts of variance will not be explained by human capital
investment decisions or rational cost-benefit analyses. As such, we incorporate the behavioral
economics concept of loss aversion into our survey design. Myopic loss aversion (Ghandi, 2007)
is defined as the avoidance of short-term costs despite the long-term average gains. Put another
way, individuals tend to prefer short-term security because of cognitive limitations that affect our
abilities to discern benefits far into the future. Conversely, it may also be that cognitive
limitations entice individuals to over-estimate the returns on their investments (Sunstein, 2006).
We include items in our instrument that attempt to measure respondents’ attitudes towards debt
and levels of debt aversion. These items are included to supplement the HCT and RCT items and
broaden our scope.
In sum, our study is designed to employ quantitative methods and blends together strands
from three distinct theoretical paradigms to determine whether master’s students’ federal
borrowing can be predicted. Further, if we can predict some of the variance in master’s students’
patterns of borrowing, we wanted to determine whether there are distinct factors that affect these
patterns. Our research begins with recognition that most research on student borrowing has been
limited to studies of undergraduates. Yet, there have been studies on undergraduate student
borrowing across all three theoretical families (HCT, RCT, and BE). Our design blends insights
from each and combines scales drawn from prior studies of loan borrowing behavior. If we find
patterns in master’s student borrowing and if these patterns have a coherent structure, the
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 54
implication of this research is that financial aid professionals and enrollment managers will have
a tool by which to improve student loan counseling services and affect students’ access,
persistence, and success. The following section details our site selection and sampling protocol.
Site and Sample
The site for our study is the Western Urban University (WUU), an elite private research
institution located in Los Angeles, California. WUU offers over 400 graduate degrees programs
the lead to certificates, PhDs, professional doctoral and master’s degrees. WUU’s graduate
student population of approximately 22,000 students is approximately 55% of its total student
enrollment (About WUU, 2013). We used multiple data sources to estimate the total WUU
master’s student population. First we used data from WUU’s Enrollment Services Division to
determine the number of admitted master’s students who accepted WUU’s offer of admission.
WUU received 6,562 acceptances of admission offers for 2010-2011 (ESD, 2011). Next, we
used IPEDS data to determine that in 2011-2012 WUU awarded 6,136 master’s degrees (IPEDS,
2013). It is therefore safe to assume that the master’s student population of WUU is somewhere
in the range of 6,000 to 7,000 students enrolled per year. This makes sense considering that we
are excluding all Ph.D., Ed.D., J.D., and other professional doctoral programs from our analysis.
Given the size of its graduate student population and the variety of graduate degree programs
offered, WUU was chosen as a site that would yield: 1) an adequate sample of graduate students
who had borrowed to finance their graduate degree; 2) a greater range in terms of student debt
loads than graduate students at public institutions, helping to find differences in student
borrowing; 3) graduate students enrolled in diverse programs, allowing for greater generalization
to the overall graduate student population.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 55
Our sampling strategy involved a convenience sample. Respondents were recruited via
coordination with program directors in charge of various professional masters’ programs
throughout WUU to distribute our electronic survey. Using an online sample size calculator we
estimated that we need a sample of 378 completed surveys. A sample of 378 was determined to
provide us with a 5% margin of error with a 95% confidence level (Raosoft, 2004); we were able
to obtain 587 responses.
To ensure that graduate students were in programs of comparable tuition cost, we limited
our sample to graduate students enrolled in professional master’s degree program. Doctoral
(Ph.D.) students are excluded from the study as the cost of the degree program is often far less
than masters program, given the subsidization of many Ph.D. programs at WUU. As well,
doctoral programs are a greater commitment than masters programs in terms of time to degree
completion, and the cost-benefit analysis may be different, given the academic trajectory of
many of the doctoral programs at WUU. Furthermore, we excluded certain professional master’s
degree programs like the master’s in business administration (MBA), as it is highly selective in
admission and the expected salary after completing the degree is much higher than other
professional master’s degrees.
Instrumentation
The survey instrument for our study was created specifically for our study on graduate
student debt. First, to our knowledge, there is no survey designed to capture information from
graduate students on their student loan decisions. Our chapter two establishes the dearth of
literature on graduate student debt and their decision-making processes. Keeping this in mind,
we created our own survey instrument, but utilized portions of prior reliable and validated
surveys. Constructs from Inceptia’s National Financial Aptitude Analysis (Inceptia, 2013),
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 56
Money Matters on Campus (2014), Young Invincible’s (Whitsett & Mishory, 2012) survey of
student loan borrowers, and EverFi’s (2014) survey of college students on financial attitudes and
behaviors were utilized intact, with minor revisions in wording for applicability to graduate
students, as needed to reflect the level of understanding of the proposed audience. Attitudinal
constructs included financial efficacy, beliefs of knowledge across financial competencies, debt
aversion, and spending compulsion. Behavioral constructs included financial behaviors and
student loan information seeking.
Furthermore, we used the findings from a recent dissertation on graduate student debt
(Cooper et al., 2013) to include constructs that would further tap into their qualitative findings on
the lack of graduate students’ knowledge regarding their loans. One such construct delved into
behaviors: sources of information influencing loan decisions. We also included items specific to
actual loan amounts, as well as demographic information on age, gender, race/ethnicity, income,
etc. to control for external factors.
Data Collection
Data for this cross-sectional research project was collected in two stages. The first stage
of data collection was a pilot test of the instrument. This pilot collection was a convenience
sampling of current and former master’s students who identified themselves as having borrowed
federal student loans for their master’s degree programs. Our goal in the first stage was to
establish the average length of time it takes respondents to complete the survey, determine
whether there are problems with words or phrases in the instrument, and to obtain initial
measures of reliability and criterion validity (Creswell, 2009; Kurpius & Stafford, 2006; Salkind
2011). Lastly, the pilot established the construct validity of the instrument. Master’s students’
attitudes towards debt and the effects thereof have not been extensively studied, but
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 57
undergraduate students’ attitudes have been (Davies & Lea, 1995; Haultain et al., 2010). It is yet
unknown whether any of the factors associated with undergraduates’ borrowing are also
predictors of graduate students’ borrowing behaviors.
The sampling procedure for the pilot relied upon a convenience sampling approach. Pilot
study respondents from WUU were recruited using Facebook pages for WUU master’s
programs. Our intent was to gather 35 responses or more, but due to time constraints this number
was not met. Nonetheless, this approach provided a pilot sample of sufficient size to fine tune the
final instrument design ahead of our principal data collection. The pilot was distributed online
using Qualtrics and respondents were provided with a link to the survey. The window for the
pilot test was open for two weeks to allow for data collection.
The second stage was our main data collection. We used Qualtrics to distribute our
instrument, gather data, and export data for analysis in SPSS. The full instrument is located in
Appendix A. Our survey distribution strategy was to contact the program leads for WUU’s
master’s programs and ask them to help distribute the web link for our instrument. We identified
program leads at three WUU schools that would provide more than an ample number of
responses: the School of Public Policy, the School of Education, and the School of Social Work.
Through the program leads, respondents were contacted via email and provided with a link to our
online instrument. The response window was open for two months to maximize data collection
within our time constraints. However, there were problems with survey distribution, which
becomes evident in the next chapter, given the much larger subsample of respondents from social
work in comparison to respondents from public policy and education. While the survey was able
to be distributed through the program lead at social work in a timely manner at the very
beginning of the two month window, survey distribution was delayed for the schools of
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 58
education and policy, due to some unexpected bureaucratic hurdles at the school level. Thus,
after obtaining authorization needed, the surveys for education and public policy were distributed
after the winter break, which impacted the response rate accordingly from master’s students at
these two schools. However, we did collect more than an adequate number of complete surveys
to examine attitudes towards debt and the relationship to student loan amount.
This study examined master’s students’ attitudes towards debt using an instrument pieced
together using items from other instruments. The rationale for this approach is that the literature
on student debt is almost entirely focused on undergraduate students. Instruments that have been
used in the past (Davies & Lea, 1995; Haultain et al., 2010) have established reliability and
validity. A problem for our inquiry is that these instruments have not been used to study whether
master’s students’ borrowing behaviors can be explained by their attitudes towards debt. As
such, our goal was to determine whether prior approaches to undergraduate students are reliable
and valid for the purpose of studying master’s students. After piloting our instrument design in
stage one of our data collection, the instrument was further refined to increase its reliability and
validity before being deployed in stage two for data collection. In sum, this data collection was
an excursion into new territory using extant instrument items to determine whether master’s
students’ borrowing of federal loans vary as a function of their attitudes.
Data Analysis
The first step in our data analysis procedure was to obtain estimates of our scale
reliability and construct validity. As outlined above, we conducted a small-scale pilot of our
instrument to obtain Cronbach’s alpha measurements for each scale. Results obtained from
Qualtrics were exported for analysis using Statistics Package for Social Science (SPSS). We also
used the pilot to determine whether respondents understood each item. This was accomplished
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 59
by incorporating open-response items at the end of the pilot survey. We determined that several
scale items needed to be reverse-coded, reworded, or removed from the final draft of the
instrument. A full treatment of the pilot survey data and analysis is located in Appendix B.
After distributing the refined survey instrument to WUU master’s students via Qualtrics,
we exported the main collection data to SPSS for data analysis. Missing data was not imputed, as
we needed complete survey data to run our multiple regression analysis. We again conducted
initial scale analysis to determine if our constructs were valid, and used correlational and
multiple linear regression analyses to answer our research questions. The next chapter delves into
the descriptive statistics and findings from our main data collection.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 60
CHAPTER FOUR
DATA AND FINDINGS
The primary goal of our research was to determine whether attitudes were related to
students’ decisions to borrow for their master’s programs. If a relationship was found, we also
wanted to discern the explanatory power of our variables on the amounts our respondents
reported borrowing. There were a number of practical considerations (primarily, time and
resources) that impacted our data collection. Despite this, however, we did find that attitudes are
related students’ borrowing decisions. Moreover, we found that these relationships were in
directions that we would expect. If one scores high on being debt averse, one is apt to borrow
less and vice versa. Likewise, we found that pro-debt attitudes (e.g. spending compulsion)
predicted a greater likelihood of borrowing for than individuals with low pro-debt attitudes. The
following chapter describes our findings and analysis from our main data collection.
Full Survey Distribution Data
Descriptive Statistics of WUU Overall
The survey gathered data from three of WUU’s graduate and professional schools: the
School of Public Policy, the School of Education, and the School of Social Work. Our final
dataset yielded 589 cases. 49 respondents were from the School of Public Policy, 99 were from
the School of Education, and 441 responses were from the School of Social Work. Females were
overrepresented in the sample and comprised 61.3% of the dataset (n = 361). Males were the
minority in our sample with n = 104 observations total. The average age in the sample was 32
years of age (n = 465, Mdn = 29, SD = 9.52) and was positively skewed. Overall, WUU’s racial
diversity was well represented in our sample. The largest racial group represented in the sample
identified as White (n = 231). African American respondents were the second largest racial
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 61
group in the sample (n = 50). We asked respondents to identify whether they were of Hispanic,
Latino, or Spanish origin. 23% of the 467 valid responses to this question indicated that
respondents were of Hispanic, Latino, or Spanish origin (n = 135). This compares to WUU’s
population as a whole where 33% of students identify as White, 5% as African American, and
12% as Hispanic. A complete breakdown of WUU’s descriptive statistics is presented in Table 1.

Table 1
WUU Respondent Characteristics
CHARACTERISTIC n CHARACTERISTIC Amount
GENDER

INCOME

Male 104 25th percentile $15,393
Female 361 50th percentile $40,423
Transgender 1 75th percentile $65,300
Totals 466 Mean $46,204

Median $40,423
RACE/ ETHNICITY

Mode $ 0
African American 50 Standard Deviation $37,959
American Indian 4

Asian 37

Hispanic/ Latino 135

Pacific Islander 11 AMOUNT BORROWED
White 231 10th percentile $ 0

20th percentile $12,177
MARITAL STATUS 30th percentile $20,075
Single 227 40th percentile $22,029
Unmarried, but living with a partner 52 50th percentile $30,000
Married 154 60th percentile $36,875
Divorced 31 70th percentile $46,779
Widow/ Widower 4 80th percentile $59,831

90th percentile $97,374
EMPLOYMENT

Mean $37,194
Full time 154 Median $30,000
Part time 148 Mode $ 0
Unemployed 164 Standard Deviation $20,023

Our survey also inquired into respondents’ domestic affairs. Specifically, we wanted to
measure students’ marital status, employment, and income levels. These data were captured for
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 62
their use as control variables and to also better understand whether certain subpopulations might
be more or less inclined to borrow federal loans for graduate study. Thirty-nine percent of
students reported being single (n = 227) and 26% of respondents indicated that they are married
(n = 154). Small proportions of our sample reported being divorced (5%, n = 31), unmarried and
living with a partner (9%, n = 52), and four respondents indicated having become a widow/
widower (1%). Respondents’ marital status was captured in our survey owing to prior research
that has established relationships between levels of financial anxiety, feelings of financial, and
marital status (Archuleta, Dale, & Spann, 2013). We also inquired about students’ employment
status and found that students were fairly evenly distributed across our three employment
categories (full time, part time, not currently employed). Twenty-six percent indicated being
employed full time (n = 154), 25% indicated being employed part-time (n = 148), and 28%
indicated being presently unemployed (n = 164). We also captured data regarding respondents’
estimated levels of personal income. This question received 454 valid responses and ranged from
$0 to a maximum of $150,000. Mean income for all respondents was observed to be $46,204 (SD
= $37,959) and the distribution was also observed to have a positive skew with reported income
tapering off substantially after crossing the $50,000/year mark. In terms of respondents’
domestic affairs our sample was fairly diverse and evenly distributed.
We also inquired whether students identified as international students. Since international
students (students who require a visa to study in the U.S.) are barred from federal loan programs
and because WUU has a large international student enrollment (About WUU, 2013) this question
helped us filter out respondents who did not borrow (n = 16).
Lastly, our dependent variable measured students’ debt loads in one year of their
program. We asked respondents to provide their best estimate of how much they borrowed to
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 63
finance the current year of graduate education at WUU. There were 468 valid responses to our
question about how much students had borrowed to pay for their graduate education for this
academic year. Respondents indicated borrowing between $0 to $100,000 (M = $37,194, Mdn =
$30,000, SD = $29,923). This distribution was positively skewed, where amounts borrowed
tapered off past around the $50,000 mark.
Descriptive Statistics by School
School of Public Policy
This section examines the descriptive statistics for the sample from the School of Public
Policy. Overall, this school had the lowest response rate to the survey at 8% (n = 49).
Nonetheless, the School of Public Policy is an important sample to study as it is has a large
social science master’s student population (n = 920, School of Public Policy Facts, 2014). The
School of Public Policy offers eleven master’s degree programs of which nine are traditional
campus-based instruction and two are online programs. One of the limitations of our design
worth mentioning here is that we did not include a question asking whether respondents were
pursuing an online degree. There is some evidence pointing to differences in student debt and
repayment as a function of on-campus vs. online study. Indeed, evidence shows that default rates
increase where access to online programs is more open and input characteristics (e.g. GPA and
test scores) decline relative to on-campus peers (Belfield, 2013). Regardless, the idea here is to
examine the descriptive characteristics of each WUU school in our sample ahead of a more
thorough analysis of the data to answer our research questions.
The distribution of males and females in this subset was fairly even. Forty-three percent
of the School of Public Policy respondents identified as male (n = 21) and 55% identified as
female (n = 27). The mean age for this subset was observed to be 32 years (n = 48, SD = 9.84).
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 64
The racial composition was less diverse for the School of Public Policy than for WUU overall.
Sixty-nine percent of Public Policy respondents indicated White as their racial identification (n =
34). Chinese students were the second largest group and comprised 12% of this subset (n = 6).
Table two displays the distribution of selected characteristics for the School of Public Policy.
Table six shows that only one person in our sample identified as Hispanic/ Latino.
The following statistics describe the marital status, employment status, and income levels
of School of Public Policy respondents. Fifty-three percent of respondents identified themselves
as single (n = 26), 22% identified themselves as married (n = 11). Eighteen percent responded
that they were single and living with a partner (n = 9), and there were three respondents who
indicated they were a widow/widower. Most of the respondents indicated that they were
employed in some capacity. Full-time employment was observed at a rate of 43% (n = 21). Part-
time employment was also frequent with a rate of 31% (n = 15). To compare, 27% of
respondents indicated that they were presently unemployed (n = 43). Income measurements for
School of Public Policy respondents revealed that their levels of income were the highest among
the three sample subgroups (n = 48, M = $59,721, SD = $45,353). However, this finding should
be regarded cautiously owing to the paucity of observations from the School of Public Policy.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 65
Table 2
School of Public Policy Respondent Characteristics
CHARACTERISTIC n CHARACTERISTIC Amount
GENDER

INCOME

Male 21 25th percentile $21,278
Female 27 50th percentile $54,453
Transgender 0 75th percentile $88,844
Totals 48 Mean $59,721

Median $54,453
RACE/ ETHNICITY

Mode $ 0
African American 3 Standard Deviation $45,353
American Indian 0

Asian 11

Hispanic/ Latino 0

Pacific Islander 1 AMOUNT BORROWED
White 34 10th percentile $ 0

20th percentile $ 0
MARITAL STATUS 30th percentile $ 0
Single 26 40th percentile $10,102
Unmarried, but living with a partner 9 50th percentile $17,111
Married 11 60th percentile $20,800
Divorced 3 70th percentile $30,116
Widow/ Widower 0 80th percentile $35,476

90th percentile $82,053
EMPLOYMENT

Mean $37,194
Full time 21 Median $30,000
Part time 15 Mode $ 0
Unemployed 13 Standard Deviation $20,023

Lastly, we examined the amounts School of Public Policy respondents reported
borrowing to fund their current year of study. On average, School of Public Policy students
borrowed the least out of all three WUU schools. A caveat, however, is that the School of Public
Policy subset had the fewest observations and those who opted to participate in our survey may
have introduced some sampling bias. Overall, School of Public Policy students in the sample
reported borrowing an average amount of $24,407 (n = 45, SD=$29,379) within a range of $0-
$100,000.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 66
School of Education
This section describes the demographic data for the School of Education. The School of
Education had the second highest survey response rate with 17% of all respondents reporting
being a master’s student (n = 99). The School of Education offers six master’s degrees on-
campus and three online programs (About The School of Education, 2014). The gender
distribution of respondents for the School of Education was primarily female (73%, n = 72). The
racial composition of the School of Education respondents was different from WUU’s overall in
a couple of ways. Most School of Education respondents identified as White (66%, n = 65), but
the second largest group of respondents from the School of Education was African Americans
(16%, n = 16). Table 3 displays a summary of the descriptive statistics and demographic features
for School of Education respondents.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 67
Table 3
School of Education Respondent Characteristics
CHARACTERISTIC n CHARACTERISTIC Amount
GENDER

INCOME

Male 27 25th percentile $15,401
Female 72 50th percentile $38,504
Transgender 0 75th percentile $60,333
Totals 99 Mean $43,892

Median $38,504
RACE/ ETHNICITY

Mode $ 0
African American 16 Standard Deviation $37,023
American Indian 2

Asian 13

Hispanic/ Latino 0

Pacific Islander 3 AMOUNT BORROWED
White 65 10th percentile $ 0

20th percentile $ 0
MARITAL STATUS 30th percentile $11,505
Single 48 40th percentile $16,414
Unmarried, but living with a partner 10 50th percentile $20,378
Married 33 60th percentile $29,982
Divorced 8 70th percentile $34,907
Widow/ Widower 0 80th percentile $49,821

90th percentile $74,597
EMPLOYMENT

Mean $27,533
Full time 40 Median $20,378
Part time 32 Mode $ 0
Unemployed 27 Standard Deviation $26,678

The single and married response rates for the School of Education were slightly higher
than for WUU overall; 49% of the School of Education respondents indicated they are single (n
= 48). The School of Education also had a higher rate of married students than WUU overall
with a 33% response rate (n = 33). Next, there was a slightly higher divorce rate in the School of
Education responses (8%, n = 8). Employment among the School of Education students was
higher compared to Public Policy and Social Work, but otherwise fairly evenly distributed. 40%
of the School of Education respondents indicated being employed full time (n = 40). 32%
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 68
reported being employed part time (n = 32) and 27% indicated being unemployed (n = 27).
Employment variables are important control variables for this study as prior research has
established a relationship between employment status and intensity and graduate school
enrollment (Zhang, 2010). Finally, respondents’ income levels were examined. Responses
revealed a lower average income for the School of Education students (n = 97, M = $43,892, SD
= $37,023) compared to Public Policy students.
Our final variable of interest was how much respondents reported borrowing to pay for
their educational costs for the academic year. There were 96 valid responses to our question
about how much students borrowed. The mean amount reported was $27,533 (SD = $26,678)
and the response distribution was positively skewed. We found that 24% (n = 24) of respondents
from this school reported having not borrowed any money at all for their education this year.
School of Social Work
The last school for which we gathered data and analyzed descriptive statistics was the
School of Social Work. Whereas the School of Public Policy and the School of Education both
offer an array of program choices, Social Work is unique among WUU graduate and professional
schools only offering one master’s degree, the master of social work, in both on-campus and
online formats. Owing to the sheer volume of responses from Social Work (n = 441) our lack of
data on whether respondents were in on-campus or online programs is particularly glaring.
Nonetheless, Social Work master’s students are interesting to study owing to the large number of
units required to complete the degree (60 semester units; WUU Social Work, 2014). This is
costly, given the WUU per unit price that is approximately $1,602 (WUU Financial Aid, 2014),
and the average starting salary for new social workers with a master’s degree is relatively low
($42,000, Social Work Salaries, 2015).
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 69
The distribution of the genders and race followed patterns observed across all three sites:
predominantly female and white. Gender frequencies at Social Work were observed to be 82%
female (n = 262) and 17% male (n = 56). There were 185 valid responses to our question about
race/ ethnicity. 71.4% of respondents identified as White (n = 132), 16.8% of respondents
identified as African American (n = 31), and 7% identified as Asian (n = 13). Table 4 provides a
summary of the descriptive statistics for the School of Social Work.
Next, we looked at employment and income levels for respondents from the School of
Social Work. There were 318 valid observations for employment and the rate of unemployment
was highest at Social Work compared to either Public Policy or Education: full time at 29.2% (n
= 93), part time at 31.8% (n = 101), and unemployed at 39% (n = 124). Among those who were
employed we found that the 50
th
percentile for income was approximately $40,000 per year.
Overall, Social Work students had a larger mean salary (M = $44,830, n = 309) compared to
Education students (M = $43.892, n = 97), but made less than Public Policy students (M =
$54,453, n = 48). Table 4 summarizes the descriptive characteristics of the School of Social
Work.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 70
Table 4
School of Social Work Respondent Characteristics
CHARACTERISTIC n CHARACTERISTIC Amount
GENDER

INCOME

Male 56 25th percentile $15,000
Female 262 50th percentile $40,000
Transgender 0 75th percentile $44,830
Totals 318 Mean $40,000

Median $40,423
RACE/ ETHNICITY

Mode $ 0
African American 31 Standard Deviation $36,689
American Indian 2

Asian 13

Hispanic/ Latino 135

Pacific Islander 7 AMOUNT BORROWED
White 132 10th percentile $10,040

20th percentile $20,000
MARITAL STATUS 30th percentile $20,977
Single 153 40th percentile $26,023
Unmarried, but living with a partner 33 50th percentile $34,967
Married 110 60th percentile $40,925
Divorced 20 70th percentile $50,000
Widow/ Widower 4 80th percentile $64,973

90th percentile $100,000
EMPLOYMENT

Mean $41,790
Full time 93 Median $34,790
Part time 101 Mode $100,000
Unemployed 124 Standard Deviation $29,791

Lastly, we wanted to get an initial idea about respondents’ reported amounts borrowed.
There were 327 valid responses to the question about how much students had borrowed for
education this year. The observed range was $100,000. On average, students from the School of
Social Work borrowed the most out of the three schools, reporting $41,790 (n = 327, SD =
$29,791). We noticed that 8% (n = 36) of the respondents from this school reported borrowing
$100,000 for the year. This amount seems high, yet given the number of individuals who
reported borrowing this much it may well be that this amount is accurate. Conversely, this also
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 71
could be explained by measurement error; respondents may have misinterpreted the question and
reported total amounts borrowed from their undergraduate and graduate education.
Comparing WUU Master’s Student Debt to National Trends
We wanted to contextualize our sample and obtain a sense of how our observations of
master’s student borrowing compared to national rates of student borrowing. We obtained
national rates of master’s student borrowing using the PowerStats tool offered by NCES. The
most current data available for comparison was NPSAS 2011 – 2012 data (NPSASb, 2012). We
then recoded our student loan amount variable data to correspond to the NPSAS datalab outputs
to create five new categories of debt. WUU respondents borrowed much more in comparison to
the national master’s student averages. Indeed, the biggest difference between WUU students
and the national rate was when students did not borrow. Whereas just over 10% of WUU
respondents indicated borrowing nothing, nearly 44% of national master’s students reported the
same. Interestingly, WUU master’s students were nearly twice as apt to borrow $60,000 or more
when compared the national rate. Table 5 below summarizes WUU borrowing rates in
comparison to NPSAS borrowing rates.
Table 5
WUU Rates of Borrowing Compared to National Rates
AMOUNT BORROWED % WUU % U.S.
$0 10.4 43.63
$1 – 17,499 9.5 16.61
$17,500 - 33,299 24.1 16.59
$33,300 - 59,999 19.9 14.26
$60,000 or more 15.6 8.91

We also wanted to get a sense of how the schools from which we sampled compared to
national rates of borrowing in related schools. We again used NPSAS (2012b) data to draw these
comparisons. First, we selected the total amount borrowed for graduate school and applied a
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 72
percentage filter to obtain discrete ranges of amounts borrowed. Next, we were able to filter the
national amounts borrowed by degree of interest (public policy, education, and social work). We
replicated these borrowing categories in our dataset and generated a frequency distribution across
our three sampling sites at WUU. There were two key findings that emerged using this strategy.
First, we found that WUU master’s students were more likely to borrow when compared
to peers in similar programs across the country. The national rate of students who reported not
borrowing federal loans was above WUU’s rates for all three schools. Public Policy rates of non-
borrowing were close to the national rate. However, the national rate was nearly double the rate
for the School of Education and four times the rate for the School of Social Work. This means
that WUU students, on average, were more likely to borrow compared to their peers across the
country.
Besides borrowing more often, we also found that WUU students were more apt to
borrow more than their peers in similar programs across the country. This is a troubling finding
because our survey instrument inquired about respondents’ borrowing in the current year only.
The NPSAS data we used was the total amount borrowed for graduate school. While we must
admit for measurement error in our survey (respondents may have reported their total amount
borrowed for all their graduate study and/ or undergraduate and graduate study), the frequency of
students borrowing $60,000 or more at WUU compared to national rates is distressing and in
need of attention. Table 6 provides a complete breakdown of the amounts borrowed by WUU
respondents in comparison to their counterparts across the United States.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 73
Table 6
WUU Schools Compared to U.S. Counterparts
AMOUNT BORROWED
PUBLIC POLICY EDUCATION SOCIAL WORK
%WUU % U.S. % WUU % U.S. %WUU % U.S.
$0 28.6 29.1 24.2 41.2 5.2 21.9
$1 – 17,499 18.4 22.4 18.2 17.2 6.6 21
$17,500 - 33,299 24.5 28.5 23.2 20.2 24.3 30.7
$33,300 - 59,999 8.2 9.4 21.2 16.1 20.9 16.2
$60,000 or more 12.2 10.6 10.1 5.20 17.2 10.1

Survey and Scale Reliability
We next present our survey and scale reliability data. Our survey consisted of six scales
and 41 total items. Figure B below, first introduced in Chapter two, is shown here again to show
how the theoretical framework shaped the actual survey instrument. The actual scales or
constructs of the survey instrument are shown below in the smaller blocks. Beliefs about
Knowledge across Financial Competencies (KFC), Loan Information Seeking behaviors (LIS),
Financial Efficacy (FE), and Financial Behavior (FB), and demographic (Demo) factors were
drawn from rational choice and human capital theory. To tap into behavioral economic theory,
which helps explain why less than rational choices are made, Debt Aversion (DA) and Spending
Compulsion (SC) scales were also included in the survey instrument. These scales, along with
demographic factors as control variables, were hypothesized to help explain the total student
amount of our graduate student respondents.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 74

Figure B. Model of attitudes and the relationship to student loan borrowing. Previously shown in
Chapter 2, this figure illustrates our theoretical framework and how it relates to the multiple
constructs of the survey instrument and ultimately student loan amount.

Having established baseline scale reliability estimates in our pilot study, this section
examines scale reliability after the main data collection was complete. This section provides
summaries of scale reliability for each scale employed in our survey. We used Cronbach’s alpha
coefficients as our measure of scale and item reliability. Generally our scales provided reliable
Debt
Attitude
Formation
Rational
Choice
Theory
Human
Capital
Theory
Behavioral
Economic
Theory
KFC LIS FE FB DA
SC Demo
Student Loan
Amount
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 75
measures, but some scales were more reliable than others. The highest observed alpha coefficient
was for the Beliefs about Knowledge across Financial Competencies scale (α =.894). The lowest
observed Cronbach’s alpha was for the Debt Aversion scale (α =.552). The following paragraphs
present findings pertaining to scale and item response rates and reliability coefficients.
The Beliefs about Knowledge across Financial Competencies (KFC) scale consisted of
eight items. The purpose of the scale was to measure respondents’ beliefs about their knowledge
of various financial tools and processes. A low score (1 = strongly disagree) indicated the
respondent did not believe s/he had good knowledge of a specific financial tool or processes. A
high score (4 = strongly agree) indicated that the respondent believed s/he had good knowledge
of specific financial tools or processes. There were 494 valid responses recorded for KFC items.
Overall, KFC reliability was observed to be highly reliable (α = .894).
The next scale, Loan Information Seeking (LIS), consisted of eight items. Four items
captured nominal data regarding whether respondents sought student loan information from
various sources. Four more items captured data on respondents’ levels of agreement with
statements that the information they received from these sources was influential in their decision
of whether to borrow federal student loans. A high score (4 = strongly agree) indicated that the
respondent believed the information s/he received was influential in his/her student loan
borrowing decision. A low score (1 = strongly disagree) indicated that the respondent believed
the information that s/he received was not influential. Overall, scale reliability for LIS items was
good (α = .592).
We felt it was worthwhile to explore respondents’ patterns of information seeking
because information-seeking behaviors are rational choices wherein borrowers are gathering
information to maximize their benefits and minimize their costs (Hogarth & Reder, 1987). We
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 76
found that 31% (n = 181) of respondents sought information about federal loans from financial
aid professionals, 26% (n = 150) asked an admissions professional, 29% (n = 169) sought help
from a family member or friend, but a majority turned to online resources to access information
about student loans (61%, n = 360). These findings reveal that respondents’ financial information
seeking behaviors were somewhat irrational insofar as the majority reported not seeking
professionals’ advice prior to borrowing federal loans. Despite this seeming irrationality,
however, respondents who did seek information did so using online sources. From a rational
choice perspective, these students were being irrational in only consulting online resources and
not seeking to maximize their sources of information of their student loan loans (Ariely, 2009;
Sunstein, 2006; Wittek et al., 2013). Behavioral economics would frame this as boundedly
rational behavior; seeking online information is better than not seeking any outside information
at all (Goldrick-Rab et al., 2009; Perna, 2006a).
The next scale, Financial Efficacy consisted of three items that measured respondents’
feelings of being in control financially. There were 467 valid responses for all three items
resulting in a response rate of 79.3%. High scores (4 = strongly agree) indicated that respondents
had strong feelings of financial efficacy. Low scores (1 = strongly disagree) indicated that
respondents had weak feelings of financial efficacy. FE items were observed to be reliable with a
high measure of internal consistency (α = .845). In general, respondents expressed agreement
with each of the three items (n = 467, M = 2.95). This suggests respondents felt in control of
their financial situations as measured by these three items. This is a positive finding because
there is evidence that showing a correspondence between problems with borrowing and issues of
self-control (Sunstein, 2006).
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 77
Financial Behaviors (FB) consisted of five items inquiring about respondents’ patterns of
financial behaviors. Questions included items concerned with whether respondents inspected
their quarterly credit reports, regularly deposited money into savings accounts, and compared
terms when borrowing money. These measures were intended to provide control variables
insofar as prior research has shown that financially literate individuals are more apt to be prudent
borrowers (Perry & Morris, 2005). There were 467 valid responses to these items resulting a
response rate of 79%. Scale reliability was (α = .647) fell just short of the ideal α = .70 (Pallant,
2008). Respondents were generally apt to express agreement with each item, but appeared less
inclined to agree that they were checking their credit reports each quarter (n = 459, M = 2.41, SD
= .990)
The Debt Aversion (DA) scale consisted of nine items with marginal internal consistency
(α = .552). This scale measured respondents’ self-reported levels of a key attitude: debt aversion.
Attitudes can be measured as preferences for or aversion to specific behaviors (Fishbein &
Ajzen, 1974; Johnson & Boynton, 2010). In this case we wanted to measure the ways in which
respondents might indicate their aversion to debt. A high score (4 = strongly agree) indicated
that respondents were highly debt averse. Low scores (1 = strongly disagree) indicated low debt
aversion; they did not want to go into debt. Overall, respondents disagreed that “There is no
excuse for borrowing money” (n = 465, M = 1.61, SD = .648). Respondents however expressed
general agreement with the statement: “I worry about debts” (n = 465, M = 3.28, SD =.781).
The Spending Compulsion (SC) scale was also marginally reliability (α = .647). Whereas
the DA scale was intended to measure debt aversion, the SC scale was intended to measure a
preference for spending which is a correlate of increased debt (Brougham et al, 2011; Roberts &
Jones, 2001). The SC scale consisted of nine items and high scores (4 = strongly agree)
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 78
indicated a high compulsion to spend. Low scores (1 = strongly disagree) indicated a low
compulsion to spend. There was general agreement among respondents that debt is an integral
feature of today’s lifestyle (n = 455, M = 2.95, SD = .745). However, the mean score for all items
indicated general disagreement with SC items (n = 455, M = 2.21). In other words, respondents
in our sample scored low on spending compulsion indicating a willingness to refrain from
spending despite recognizing that debt is somehow unavoidable.
Answering the Research Questions
Here we present our answers to our two research questions. Our goal was to determine
whether there was a relationship between our respondents’ debt attitudes and the amounts they
had reported borrowing for the current year of their programs. Next, if the data established such a
relationship, we wanted to ascertain whether certain attitudinal variables predicted the amounts
our respondents had reported borrowing. We used correlation analysis and linear regression to
answer our research questions and now present our main findings.
Research Question 1
Is there a relationship between master’s students’ attitudes towards debt and the amounts
they borrow in student loans? The answer was yes, albeit a qualified yes. After running the
correlations between the dependent variable of student loan amount and the independent
variables that had been trimmed down from model one, we found that with the exception of one
Debt Aversion item (once you are in debt it is very difficult to get out), all of the attitudinal
factors included in model two were significant; however the Pearson correlations were weak
(Table 7). More importantly, we found that the directions of the relationships were in the
directions we expected. That is, as measures of debt aversion increase, the amount borrowed
decreased. And as measures of spending compulsion increased, the amount borrowed also
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 79
increased. We look at the correlations between amount borrowed (our dependent variable) and
our attitudinal variables two different ways.
First we examined the relationship between amount borrowed and the attitudinal
variables overall. In theory, the Debt Aversion (DA) variables should have a negative
relationship with amount borrowed. There were negative relationships between all but one of the
nine DA items. There were two significant, negative relationships (DA1: r(435) = -.146, p = .02;
DA2: r(435) = -.119, p = .013) and there was a third significant, but positive relationship (DA8:
r(434) = .129, p = .007). While the negative relationships generally support the theory that debt
aversion is related to lower amounts borrowed, we did not expect the significant positive
relationship. However, we interpret this finding to mean that as one has borrowed increasing
amounts one is more inclined to express worry about debt. Therefore, worry about debt is not
necessarily prima facie debt aversion, but debt aversion after the fact.

Table 7
Correlations of DA Independent Variables with Amount Borrowed (DV)
INDEPENDENT VARIABLE r Sig. n
DA1: No excuse for borrowing -.146** .002 437
DA2: Owing money is wrong -.119* .013 437
DA8: I worry about debt .129** .007 436
*p < .05. **p < .01.

Next we analyzed the data to determine whether our second attitudinal measure of
Spending Compulsion was related to how our respondents had reported borrowing. Again we
theorized that increased measures of Spending Compulsion should correspond with increased
measures of reported amounts borrowed. Correlation analysis of our SC items and the dependent
variable (DV) of student loan amount revealed positive relationships for all of our SC items,
except SC8 (I prefer saving). SC8 was reverse-coded because saving is, in fact, the opposite of
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 80
spending compulsion. As such, it makes sense that this lone item was negatively related to the
DV. There were four significant, positive relationships among SC items with the DV and one
significant, negative relationship. The relationship between credit card usage and student loan
amount can be explained as a function of borrowers perhaps having ample resources and good
credit. These individuals use their credit cards to pay down their costs of attendance, hence the
lower loan balances. Table 8 summarizes these significant relationships.

Table 8
Correlations of SC Independent Variables with Amount Borrowed (DV)
INDEPENDENT VARIABLE r Sig. n
SC1: Debt is integral .188** .001 436
SC2: Taking out a loan is good .129** .007 435
SC4: Students have to go into debt .229** .001 435
SC5: Have something now, pay later .108* .025 435
SC9: Prefer using credit card -.159** .001 435
*p < .05. **p < .01. ***p < .001

Research Question 2
We answered our second research question using linear regression analysis. If we
established that attitudes were related to borrowing patterns, we wanted to also determine
whether certain variables were predictive of respondents’ reported amounts borrowed. As such,
we answered this question in two steps. First, we imputed all our independent variables to
determine the overall model for our instrument. Next, we culled our variables to those that had
significant predictive strength to form a second model for our instrument. Overall, we succeeded
in predicting 13% of the total variance in our dependent variable (R
2
=.131, F(8,422) = 9.07, p <
.001). Therefore, the answer to our second research question is that we have identified several
items that help us to predict, in a limited way, the amounts master’s students at WUU might
borrow in the future.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 81
The first step to answering research question two was to first establish items’ construct
validity. We accomplished this via correlation analysis. First, correlations were run to see if all
items within a construct were significantly correlated to each other. If all of the items were
significantly correlated with each other, a Cronbach’s alpha statistic was computed to measure
the construct’s internal consistency or reliability. If the construct met the threshold of a strong
Cronbach’s alpha (α > .7), we created a new variable that is the mean of all of the construct items
for inclusion in the multiple regression analysis. As shown in Tables 9 and 10, two constructs,
Beliefs about Knowledge across Financial Competencies (KFC) and Financial Efficacy (FE),
both proved to have items that were significantly correlated with each other and were reliable (α
= .89 and α = .85, respectively). The construct of Financial Behavior (FB) had significant
moderate inter-item correlations (see Table 11). This construct was just below the threshold of
reliability at α =.65, but we created a new variable to represent the mean of all FB items, as all
five of these items were significantly and positively correlated with each other. Also, in our pilot,
we had extensively refined this construct to be a coherent measure of important financial
behaviors, eliminating redundant questions. For the two constructs that did not meet the tests for
correlation or internal reliability, Debt Aversion (DA) and Spending Compulsion (SC)—we did
not treat these as valid constructs and thus did not create new variables to represent the means of
the construct items. However, multiple regression was conducted instead with all of the
individual items in both of these constructs to find out if certain items were predictive of student
loan amount.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 82
Table 9
KFC Pearson Correlations within Construct Items

1.
Budgeting
2. Credit
cards
3.
Insurance
4.
Student
loans
5.
Investing
6.
Compound
interest
7.
Retirement
planning
8. Credit
reporting
bureaus
1. I believe my
knowledge of
budgeting is
good
_ .581
**
.538
**
.562
**
.421
**
.429
**
.477
**
.507
**

2. I believe my
knowledge of
credit cards is
good
.581
**
_ .571
**
.510
**
.392
**
.398
**
.470
**
.583
**

3. I believe my
knowledge of
insurance is
good
.538
**
.571
**
_ .595
**
.538
**
.480
**
.536
**
.570
**

4. I believe my
knowledge of
student loans is
good
.562
**
.510
**
.595
**
_ .427
**
.484
**
.465
**
.505
**

5. I believe my
knowledge of
investing is
good
.421
**
.392
**
.538
**
.427
**
_ .617
**
.678
**
.495
**

6. I believe my
knowledge of
compound
interest is good
.429
**
.398
**
.480
**
.484
**
.617
**
_ .572
**
.519
**

7. I believe my
knowledge of
retirement
planning is good
.477
**
.470
**
.536
**
.465
**
.678
**
.572
**
_ .599
**

8. I believe my
knowledge of
credit reporting
bureaus is good
.507
**
.583
**
.570
**
.505
**
.495
**
.519
**
.599
**
_
**. Correlation is significant at the 0.01 level (2-tailed).

Table 10
FE Pearson Correlations within Construct Items

1. 2. 3.
1. I feel capable of using my future income to achieve my financial goals
_ .714
**
.589
**

2. I feel capable of handling my financial future
.714
**
_ .650
**

3. I feel in control of my financial situation
.589
**
.650
**
_
**. Correlation is significant at the 0.01 level (2-tailed).

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 83
Table 11
FB Pearson Correlations within Construct Items
1. 2. 3. 4. 5.
1. I stick to a monthly budget
_ .341
**
.423
**
.225
**
.282
**

2. I save on a regular basis
.341
**
_ .401
**
.226
**
.128
**

3. I regularly monitor my spending
.423
**
.401
**
_ .294
**
.260
**

4. I compare terms of repayment before I borrow
.225
**
.226
**
.294
**
_ .282
**

5. I check my credit report quarterly
.282
**
.128
**
.260
**
.282
**
_
**. Correlation is significant at the 0.01 level (2-tailed).

In conducting the multiple linear regression, because complete information is needed and
we did not impute missing data, only complete survey responses were utilized (n = 431). Two
models for the multiple linear regression were run, using the general equation

with Y as the predicted value of the dependent variable (student loan amount), b
0
as the value of
Y when all of the independent variables (X
1
through X
p
) are equal to zero, and b
1
through b
p
as
the estimated regression coefficients.
The first model contained the independent variables including the means of three
constructs: 1) Beliefs about Knowledge across Financial Competencies (KFC), 2) Financial
Efficacy (FE), 3) Financial Behavior (FB), all items from the Debt Aversion (DA) and Spending
Compulsion (SC) constructs, and demographic factors including employment status, marital
status, gender, race/ethnicity, WUU graduate school affiliation (public policy, education, and
social work), income, and age. Dummy variables were used to replace categorical variables
including employment status (full time, part time, not currently employed), marital status, gender
(male, female, trans/other), race/ethnicity, and school affiliation (public policy, education, social
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 84
work). Marital status was simplified to single (single, divorced, widowed/widower) and partner
(unmarried but living with partner and married). Race/ethnicity was also simplified to Hispanic,
White, Black/African American, American Indian/Alaska Native, and Asian/Pacific Islander
(Asian Indian, Chinese, Filipino, Japanese, Korean, Vietnamese, Native Hawaiian, Guamanian
or Chamorro, Samoan, Other Asian, Other Pacific Islander).
The results of the regression indicated that the predictors explained 10% of the variance
(adjusted R
2
= .100, F(34,378) = 2.35, p < .001). As shown in Table 12, only six of the
independent variables made a statistically significant contribution to the model: affiliation with
the school of public policy (p = .016) and education (p = .001), and strongly agreeing with the
following statements: “I worry about debts” (p = .029), “Once you are in debt it is very difficult
to get out” (p = .036), “Students have to go into debt” (p = .045), and “I prefer using my credit
card to make purchases” (p = .016). While none of the three constructs were found to
significantly contribute to the model, Beliefs about Knowledge across Financial Competencies
was close to the threshold of significance (p = .06), as was strongly agreeing with the statement
“Debt is an integral part of today’s lifestyle” (p = .063).

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 85
Table 12
Predicting Amount Borrowed with Multiple Regression (Model 1)

INDEPENDENT VARIABLE B
Standard
Error of B beta Sig.
KFC (mean)
5593.58 3060.01 0.12 0.068
FE (mean)
-472.64 2876.77 -0.01 0.870
FB (mean)
-1063.38 3285.68 -0.02 0.746
DA1: No excuse for borrowing
-1789.43 2758.62 -0.04 0.517
DA2: Owing money is wrong
419.15 2523.11 0.01 0.868
DA3: Prefer using my debit card
-329.94 1886.97 -0.01 0.861
DA4: Ok to borrow to buy food
445.79 1914.74 0.01 0.816
DA5: Lower a person’s income, more important to save every month
19.16 2128.27 0.00 0.993
DA6: Should always save up first before buying something
-1726.96 2303.33 -0.04 0.454
DA7: Should stay home rather than borrow to go out for evening on
town
-168.96 2133.40 0.00 0.937
DA8: Worry about debts
4860.30 2211.75 0.13 .029 *
DA9: Once in debt, very difficult to get out
-4264.38 2029.44 -0.12 .036 *
SC1: Debt is integral part of today’s lifestyle
4065.51 2177.74 0.10 0.063
SC2: Taking out a loan is good because it allows you to enjoy life as
student
132.38 2402.71 0.00 0.956
SC3: Prefer spending
-1708.01 2554.62 -0.04 0.504
SC4: Students have to go into debt
3352.98 1665.97 0.11 .045 *
SC5: Better to have something now and pay for it later
3125.66 2522.36 0.07 0.216
SC6: Buying things gives me a lot of pleasure
-1510.08 2252.15 -0.04 0.503
SC7: Enjoy spending money on things that are not practical
1038.71 2495.43 0.03 0.677
SC8: I prefer saving
-252.14 2675.23 -0.01 0.925
SC9: Prefer using credit card
-4781.86 1969.59 -0.15 .016 *
Part time employment
4173.64 3812.80 0.07 0.274
Not employed
4206.95 3672.25 0.07 0.253
Partner
1117.19 3310.22 0.02 0.736
Male
-6001.19 3554.12 -0.08 0.092
Transgender
-9739.83 28743.77 -0.02 0.735
Hispanic
-6283.80 3988.78 -0.10 0.116
Black
-323.46 4915.74 0.00 0.948
American Indian
-16352.61 14634.84 -0.05 0.265
Asian/Pacific Islander
235.75 5125.99 0.00 0.963
School of Public Policy
-13014.87 5364.20 -0.13 .016 *
School of Education
-12489.80 3894.80 -0.18 .001 **
Annual household income
0.00 0.04 0.00 1.000
Age
-106.64 192.39 -0.03 0.580
Notes. B unit is dollar amount of student loan
*p < .05. **p < .01. ***p < .001

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 86
We then trimmed the model by conducting a second linear multiple regression (Model 2)
with eight factors, including the six factors that were found to be significant and the two factors
that were at the threshold of significance. The equation for this model was:
Y = 3202.63 (KFC) + 12529.42 + 4782.73 (DA8) + 4972.76 (DA9) + 4161.94 (SC1) + 5264.30 (SC4) +
-4155.19 (SC9) + -11636.91 (Public Policy) + -12598.50 (Education)
This model was stronger than the first, with the predictors explaining 13% of the variance
(adjusted R
2
=.131, F(8,422) = 9.07, p < .001). Building upon the analysis from model one that
we were able to develop a second model that helped us more precisely answer our second
research question.
Model two revealed that there were seven significant predictors of student debt in our
sample (Table 13). The two most important predictors were affiliation with the education school
(p = .001) and strongly agreeing with the statement, “Students have to go into debt” (p = .001).
The other five independent variables that made a statistically significant contribution to the
model were the public policy school affiliation (p = .012), and strongly agreeing with these
statements: “I worry about debts” (p = .016), “Once you are in debt it is very difficult to get out”
(p = .007), “Debt is an integral part of today’s lifestyle” (p = .027), and “I prefer using my credit
card to make purchases (p = .005). Affiliation with the school of education or public policy had
significant negative weight, indicating that after accounting for all other factors, these students
had lower student loan amounts. This was also in contrast to affiliation with the school of social
work; education students averaged $11,274 (β = -11274) less in student loan debt, and public
policy students averaged $11,670 (β = -11670) less in student loan debt than social work
graduate students. Believing that debt is very difficult to get out of and preferring credit cards for
purchases were also significantly negatively correlated with student loan amount. The three
factors that were significantly positively correlated to student loan amounts were worrying about
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 87
debts, believing that debt is an integral part of today’s lifestyle, and believing that students have
to go into debt. To summarize, our second, trimmed model shows that attitudes in conjunction
with demographic factors explained up to 13% variance in total student loan amount.
Table 13
Predicting Amount Borrowed with Multiple Regression (Model 2)
Independent Variable B
Standard
Error of B beta Sig.
KFC 4050.881 2211.346 .086 .068
DA8: Worry about debts
4757.381 1971.610 .124 .016*
DA9: Once in debt, very difficult to get
out
-4786.824 1754.345 -.137 .007**
SC1: Debt is integral part of today’s
lifestyle
4230.273 1910.110 .107 .027*
SC4: Students have to go into debt
4951.898 1508.813 .158 .001**
SC9: Prefer using credit card
-4322.145 1527.518 -.133 .005**
School of Public Policy
-11670.150 4636.843 -.120 .012*
School of Education
-11274.455 3330.809 -.157 .001***
Notes. B unit is dollar amount of student
loan
*p < .05. **p < .01. ***p < .001

To review and close this chapter, our data was sufficient to answer both of our research
questions. Correlation analysis of our attitudinal variables revealed significant relationships with
amounts borrowed by our respondents in the directions we had theorized. Results from our linear
regression analysis yielded a refined, yet somewhat weak model for predicting our respondents’
borrowing. Most importantly for this study, we found two attitudes that were significant
predictors of lower student loan debt were 1) believing that it is difficult to get out of debt once
in debt and 2) preferring credit cards to make purchases. Higher student loan debt was predicted
by attitudes including 1) worrying about debt, 2) believing debt is an integral part of today’s
lifestyle, and 3) believing that students have to go into debt. The results of the multiple
regression also revealed that while school affiliation was a significant predictor, but as this
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 88
finding was not related to attitudes we will not focus on that here in this chapter. Chapter five
presents an analysis of our findings in light of our conceptual model while also examining the
implications of this research on future inquiry and professional practice.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 89
CHAPTER FIVE
DISCUSSION
This concluding chapter provides a recap of the purpose of the study and research
questions, discussion of the findings, the limitations of our study, and implications for the future.
For this study of graduate student debt, we used a theoretical framework that integrated multiple
lenses to better understand graduate student decision-making: rational choice theory (RCT),
human capital theory (HCT), and behavioral economic (BE) theory. The first lens, rational
choice theory (RCT), helps us frame borrowing as a behavior undertaken after cost-benefit
analysis (Gilboa, 2010; Hogarth & Reder, 1987). The second lens of human capital theory (HCT)
informs us that borrowing decisions, particularly in funding graduate education, are viewed as
investments in long-term personal productivity (Avery & Turner, 2012; Becker, 1962; Coleman,
1988; Dynarski, 2008; Volkwein et al., 1998). Behavioral economics (BE), our third lens, makes
it clear that there are limitations to rational behavior, given our limited information intake
abilities and cognitive processing, and that human behavior is predictably irrational. Our study
used a survey instrument designed to utilize these three lenses to provide a more comprehensive
view of human behavior through which we can better understand students’ borrowing behavior.
Purpose of the Study and Research Questions
The purpose of this study was to better understand graduate students' decision-making
processes in borrowing for their education, and thus proposed to measure master’s students’
attitudes towards debt and related behaviors to analyze whether these attitudes and behaviors
influenced how much they borrowed to pay for their programs. A survey instrument specific to
graduate students was adapted from existing instruments to measure relevant attitudes towards
debt, and data was collected from participating masters' students enrolled at the Western Urban
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 90
University. The survey data was then coded and analyzed using SPSS to answer our research
questions.
Our two research questions were:
1. Is there a relationship between master’s students’ attitudes towards debt and the
amounts they borrow in student loans?
2. If attitudes predict borrowing patterns, what factors are associated with increased
borrowing and decreased borrowing?
Discussion of Findings
This project began with the goal of documenting whether attitudes were related to
master’s students’ decisions to borrow to pay for graduate programs. For our sample of masters’
students at the Western Urban University, we found that in answer to our first research question,
there was a relationship between their attitudes and the amount they borrowed for their graduate
education. Furthermore, in answer to our second research question, after controlling for
exogenous demographic factors, key attitudes towards debt were found to predict patterns of
borrowing.
To the extent that prior research has neither focused on graduate students specifically, nor
has prior research established a relationship between attitudes and borrowing for graduate study,
we find our project to be one of theoretical and practical importance. Our findings revealed that
master’s students’ debt attitudes, defined as a preference for or aversion to debt, are related to
how much they report borrowing. We established that certain attitudes were significant
predictors of students’ self-reported amounts borrowed. While the percentage of variance
explained by our model is modest, it bears upon this research to note that human behavior is
complex.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 91
Overall, our study identified that attitudes helped explain student borrowing decisions.
The attitudes most strongly linked to student loan amount were that of debt aversion and
spending compulsion. In general and in accordance with the literature, attitudes of debt aversion
predicted lower student loan amounts and attitudes of spending compulsion predicted higher
student loan amounts. While we also found that school affiliation was a stronger predictor of
student loan amount than attitudes, this may not be the most robust finding given the
overwhelming majority of our sample came from social work graduate students.
Our data indicated that attitudes of debt aversion and compulsive spending were
predictive of borrowing patterns for graduate students. As shown in the previous chapter, while
the survey instrument’s constructs of Debt Aversion (DA) and Spending Compulsion (SC) did
not meet our tests for construct validity, there were items in both of these constructs that proved
significantly and moderately correlated with student loan amount.
Spending Compulsion
Items from the Spending Compulsion (SC) construct were the most significantly
correlated to student loan amount. This construct was intended to capture students’ risky
financial attitudes and behaviors and thus measure debt tolerance. Within our theoretical
framework, spending compulsion or debt tolerance can be viewed as part of rational choice and
human capital theory. To capitalize on the longer-term benefits of graduate education, students
would need to show some degree of spending compulsion to take on student debt. Specifically
the significant factors linked to student loan amount included the belief that debt is an integral
part of today’s society, that students had to go into debt, and that credit cards were preferable (to
cash or debit cards) for purchases. The literature on spending compulsion predicts that students
who exhibit this type of attitude towards money and spending would be more apt to carry higher
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 92
debt loads (Brougham et al., 2011; Gutter & Copur, 2011; Roberts & Jones, 2001), and our
findings tend to agree. The median student loan amount across our sample was $30,000 for the
year. Both the attitude or belief that debt is an integral part of today’s lifestyle, or that students
have to go into debt, were significantly correlated with higher student debt loads. As shown in
the regression models in the previous chapter, for each degree stronger in agreement with these
attitudes, student loan amounts increased an average of $4,200 to $5,000, respectively.
However, students who preferred using their credit cards for purchases were significantly
correlated with lower student debt loads; for each degree stronger in agreement with this
statement, student loan amounts decreased an average of $4,300. This was one finding that
seemed to contradict other findings regarding spending compulsion. We may be able to explain
this counterintuitive finding with the explanation that given the prevalence of credit cards today
and the assumption that graduate students are more sophisticated financial consumers than the
undergraduates these survey items were originally created for, graduate students who prefer to
use credit cards may indeed prefer to use their credit cards rather than debit cards or cash. In fact,
those with strong credit may have access to credit cards, utilize them frequently, and most likely
pay these credit card balances off in full. In the future, this item should most likely be reworded
to better capture compulsive spending, as the use of credit cards themselves is no longer
necessarily indicative of compulsive spending.
Debt Aversion
We found two items in the Debt Aversion (DA) construct that were significantly
correlated with student loan amount. This construct was designed to measure students’ aversion
to debt. Debt Aversion is best viewed through the theoretical framework lens of behavioral
economics. Despite the long-term benefits of obtaining an advanced education, some students are
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 93
averse to borrowing in an irrational manner. Two Debt Aversion items were significantly
correlated to student loan amount, but the relationships to student loan amounts were mixed.
Again, like spending compulsion, the Debt Aversion construct did not meet our criteria for
construct validity, but because these items represented important attitudes linked to borrowing
behavior in the literature, individual items from the construct were included in the multiple
regression analyses. Consistent with much of the literature on debt aversion, we found that the
stronger the belief that debt is very difficult to get out of, the less the student incurred in total
loan debt. For each degree stronger in agreement with this statement, student loan amounts
decreased an average of $4,800.
However, data analysis revealed that the more the student worried about debt and thus the
greater the debt aversion, the larger the student loan debt. This may be explained by the nature of
our analyses; multiple regression provides correlational but certainly not causal data. Thus we
may interpret our finding with this possible explanation: students may have a large amount of
student debt, which in turn causes greater worry about student loan debt. One way to gain more
precision in this data is to design a longitudinal study in which attitudes towards debt are
assessed just before entering the graduate program and at some point after enrollment and after
graduate debt has been incurred.
Beliefs about Knowledge across Financial Competencies
While not as robust a finding as the Spending Compulsion and Debt Aversion items
discussed above, it is important to discuss the one construct that met our tests for construct
validity and was close to being statistically significant: Beliefs about Knowledge across
Financial Competencies (KFC). This construct was created to measure respondents’ beliefs about
their knowledge of various financial tools and processes, or in essence, to tap into their financial
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 94
literacy. This construct fit into the realms of rational choice and human capital theories, in which
greater knowledge should lead to better decision-making processes. Research in the area of
financial literacy often finds that consumers who are more financially literate and possess greater
knowledge of everyday financial tools and processes are more likely to exhibit financially
prudent behavior (Perry & Morris, 2005), and thus theoretically should be linked to lower
student loan amount. Again, while not statistically significant in either of our regression models,
KFC was linked positively to student loan amount, meaning that the greater the student indicated
his/her knowledge of these basic financial competencies, the greater the student loan amount.
This is yet again another counter-intuitive finding, but may be explained in several ways. One
explanation may be that this construct tapped into students’ beliefs about this knowledge and the
students who rated themselves highly in these competencies believe they are financially literate
but perhaps are not as financially literate as they should be. Another explanation may be simply
that these students did borrow in a rational manner, and found that having a higher student loan
amount was preferable or necessary to the alternative, which may have been higher credit card
debt or a longer time to degree completion. In future research, one way to mitigate this
ambiguous finding is to control for actual financial knowledge and to therefore embed a test for
financial competency. However, this was beyond the scope of our study’s inquiry.
School Affiliation
We had also included control variables in the form of exogenous demographic factors
included in our regression analyses. These independent variables were employment status,
marital status, gender, race/ethnicity, WUU graduate and professional school affiliation (public
policy, education, and social work), income, and age. The only factor that was a significant
predictor in student loan amount was that of school affiliation. In fact, the strongest finding was
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 95
that affiliation with the school of education or public policy was negatively linked to student loan
amount, in contrast to the school of social work. However, as graduate students from the school
of social work made up the majority of our sample, we take care not to generalize too much in
terms of school affiliation being a strong predictor.
Loan Information Seeking Behavior
Finally, while not directly related to our research questions, we would like to briefly
discuss Loan Information Seeking Behavior (LIS). This construct was not included in the
regression analysis as it was not pertinent to our research questions, but we had included this
construct to help us identify students’ sources of information in making their student loan
decisions. In our theoretical framework, LIS fit into rational choice and human capital theories
that humans seek knowledge and information before making an important decision. However,
our finding may better help to explain behavioral economic theory, as students did not always
maximize their information sources. What is perhaps the most illuminating finding from our
survey data is that graduate students who we assume are more sophisticated in their financial
knowledge and decision making processes than undergraduate students, turn more to online
resources (60%), rather than institutional resources (never exceeds 30% for seeking information
from a financial aid or admissions professional). While we cannot judge the quality of the
information on student loans found online, this should be a finding that is of interest to
institutions. There should be much more of an effort by institutions to provide better information
and tools online on student loans. Institutional efforts designed to make admission and financial
aid professionals more relevant and accessible to graduate students in the financial decision-
making process could also be instrumental in reducing average loan balances.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 96
Limitations
There are several limitations that impact our findings and their generalizability. In brief,
these include problems with our sampling site, overrepresentation of the social work subsample
coupled with low response rates from both public policy and education graduate students,
problems with scale reliability, self-reporting of student loan debt levels, and weak correlations
between scale items and the dependent variable. Despite these limitations we maintain that our
findings revealed that attitudes partially affect students’ borrowing decisions. Other significant
predictors of student debt were found to be where one chooses to study and the duration of one’s
course of study. Our findings suggest that while attitudes played a role in our respondents’
borrowing decisions, attitudes-based research on student debt needs further refinement. In light
of this, the following paragraphs consider the limitations of our research.
First, our decision to focus solely upon WUU limits our ability to generalize our findings
beyond this institution. We considered sampling from multiple schools during the research
design process, but made the choice to focus on WUU owing to time and resource constraints.
As such, our ability to generalize beyond WUU is limited. WUU is a highly selective institution
and this affects the admitted student profile in ways that may cause it to differ from less selective
institutions’ profiles. Further, considering WUU’s location and the urban focus of the programs
we sampled, it is plausible that our sample differs in key ways from non-urban schools. Lastly,
WUU’s status as a private institution may cause our sample to differ from those we could have
gathered from public institutions. For example, the national average tuition and fees charged for
graduate programs at public schools was $10,408 per year and private, non-profit schools
charged an average of $23,698 (NCES, 2013). WUU’s tuition and fees ($21, 861; WUU, 2013)
during the same period was double the national average for public schools, yet below the
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 97
national average for private, non-profit schools. As such, some generalizations may be
warranted.
For instance, the WUU School of Social Work respondents were generally more diverse
than their counterparts from the National Association of Social Workers (NASW). Whereas
females comprised 80% of NASW membership (NASW, 2003), WUU Social Work respondents
were 59% female and 13% male. Further, the racial diversity of NASW members was fairly
narrow with 87% of respondents identifying as White. Conversely, WUU Social Work
respondents were more diverse with only 30% identifying as White. It must be noted that over
half of our Social Work sample declined to provide a racial identification. This could be related
to our decision not to include a biracial identification category. Overall, though, the Social Work
subgroup of our sample may be representative of social workers situated in diverse, urban
settings. This has implications for the generalizability of our findings from the Social Work sub-
group to social work students pursuing studies in urban schools.
Moreover, our sample mirrors national demographic trends in graduate education. Using
data from the National Center for Education Statistics (IPEDS, 2013) we compared national
graduate student descriptive statistics with those observed in our sample. Nationally, female
postbaccalaureate enrollment was high (58.5 %) compared to males (41.4%). While the IPEDS
data does not indicate master’s student enrollment, it does illustrate that females are enrolling in
graduate programs at higher rates than males. We observed a similar pattern in our data where
female enrollment in master’s programs was high (61%) in comparison to males (18%). This
finding would benefit from future research using random sampling techniques to mitigate the
problem of self-selection bias that is likely affecting our research.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 98
In terms of racial/ethnic composition, our sample differs in key ways from national
master’s enrollment rates. We used NPSAS (2012) to determine national rates of master’s
program enrollment for racial groups to determine comparability with our sample. Nationally,
the majority of master’s degree enrollments are White students (69%). Our sample, on the other
hand, had a White response rate of 39%. African Americans represented 13% of the national
master’s degree enrollments, but African Americans were observed to be 9% of our sample. Our
response rate for Hispanic/Latino students was high (23%) compared to the national rate of 9%.
While there are some parallels between our sample and national response rates, we do not
believe our findings to be generalizable. A random sampling method would have given us a
generalizable sample (Creswell, 2009), but resource and time constraints compelled us to use
voluntary recruitment for our survey. However, recent research suggests that students’ self-
selection to participate in social science research is neither associated with increased odds of
participation, nor participation frequency (Falk, Meier, & Zehnder, 2013). While these findings
are favorable for our research, we cannot underestimate the effects of self-selection bias. Self-
selection surveys like ours suffer by not relying on actual probability sampling, allowing for the
possibility of allowing a vocal subpopulation to bias survey results (Bethlehem, 2008). As such,
we repeat that future studies of graduate student debt employ random sampling techniques within
an institution and across institutions of similar composition.
Indeed, our social work subsample outstripped the participation rates for both education
and public policy combined two times over. Of our total 589 respondents, were from the school
of social work. This is precisely the kind of bias of which Bethlehem (2008) warns. The master
of social work is a special program at WUU for a few reasons. First, full-time students in the
master of social work program take 17 units in their first semester at a per year tuition and fee
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 99
cost of $23,781. Full-time public policy and education master’s students take half as many units
in their first semester. Second, these public policy and education programs require fewer units to
complete a degree compared to the social work program. For example the public policy school’s
MPA degree requires 41 units and the education school’s Master of Educational Counseling
requires 46 units to complete, but the Master of Social Work requires 60 units. Whereas all three
programs can be completed in two years, per unit costs are substantially higher for part-time
social work students who pay $1,602 per unit ($96,120 total, WUU Tuition, 2014). These
differences in per unit cost could also have effects on generalizability as public schools charge
less for tuition. Thus the disproportionate response rate between social work and the other two
schools impairs the generalizability of our findings.
Another limitation of our research involves the reliability of our scales. Cronbach’s alpha
coefficients of .70 are regarded as acceptable for research purposes (Kurpius & Stafford, 2006;
Pallant, 2007). Only two of our scales met this threshold (KFC α = .894; FE α = .845) and a
couple came close (FB α = .647; SC α =.647). The SC scale (Spending Compulsion) was
intended to measure an attitudinal preference for spending money, which is supposed to be
related to an increased likelihood of carrying increased debt (Brougham et al, 2011; Roberts &
Jones, 2001). Another important attitudinal scale, Debt Aversion (DA), had low reliability (α =
.552) indicating possible bias in our sample. Davies and Lea (1995) —the study from which we
borrowed DA items—had good reliability (α = .790) for their scale of attitude to debt, but we
were unable to replicate this finding in our population. This could be a function of many factors
and we consider that their survey was conducted in the mid-1990s, in England, using
undergraduate students from one school to be plausible explanations of the differences in
observed reliability. Nonetheless, the low reliability coefficients of our scales were problematic
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 100
for our research and points to a need to both refine our scales and obtain a more representative
sample.
Another limitation of our study involves the self-reporting of student debt levels. Our
survey relied upon respondents inputting the amount they borrowed to pay for the current year of
study. Our data revealed that out of 468 valid observations there were 44 individuals who
reported borrowing $100,000 to pay for one year of graduate study. These numbers were not
corroborated with the WUU Office of Financial Aid owing to time and institutional policy
considerations. As such, we cannot be sure whether our respondents’ reports were accurate. One
study concluded that self-reporting errors increase as a function of being a member of an at-risk
group such that these individuals underreport on selected measures (Rowland, 1990). Another
inquiry found that social desirability and social approval are important factors in reporting errors
where respondents may seek to please researchers by reporting data in a particular way (Adams
et al., 2005). Given these findings, future research on this topic could benefit from correlating
respondents’ actual debt levels with their reported amounts borrowed. Further, we must assume
that some of the variance in amounts borrowed is reported in error owing to respondents not
reading or understanding our instructions to report only the amount they had borrowed for the
first year of graduate study. Finally, the issue of self-selection bias is once again salient, as we
cannot rule out that persons with large amounts of debt want to call attention to their worries
through our research. A more precise method of matching attitudes to debt with amounts
borrowed would involve gathering student ID numbers and cross-referencing data with student
financial aid records. Again, time was a limited resource for this study and such a tactic was
infeasible.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 101
Finally, we recognize that the relationships between our independent and dependent
variables were generally weak. The relationship between Debt Aversion (DA) items and the
dependent variable (reported amount borrowed) was significant in three cases out of a possible
nine. We theorized that DA items would be negatively related to our dependent variable and that
was generally what we found. (DA_1: r = -.146, n = 437, p = .002; DA_2: r = -.119, n = 437, p =
.013; DA_8: r = -.129, n = 436, p = .007). While we observed that as one reported higher levels
of debt aversion their corresponding debt levels decreased, these relationships were found to be
weak. Similarly, we theorized that as measures of Spending Compulsion (SC) increased there
would be a corresponding increase in the amounts respondents borrowed. This was observed in
our data, but again the relationships were generally weak (SC_1: r = .188, n = 436, p = .001;
SC_2: r = .129, n = 435, p = .007; SC_5: r = .108, n = 435, p = .025). There was a moderately
positive relationship between the dependent variable and SC_4 (“Students have to go into debt”;
r = .229, n = 435, p = .001). Both SC and DA scales were important attitudinal constructs that
were intended to determine whether preferences for or aversions to debt would be related to the
amounts students borrowed. The weakness of these relationships, though significant, suggests
that further research is warranted on attitudes and that the scales used in this survey need further
refinement.
Implications for Practice
Our limitations notwithstanding we are encouraged by our findings as they suggest that
institutional professionals can take steps to mitigate excessive borrowing. The following
paragraphs comment upon how our findings can be leveraged in professional settings.
Our data revealed that surveyed master’s students reported borrowing less in a year of
their program (M = $37,194, n = 468) than the mean cumulative debt (M = $47,269) observed in
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 102
the most recent NPSAS (2012) data. However, if we multiply out the amount borrowed by two—
this is the length of time it takes to finish most public policy and education master’s programs
and the full-time social work master’s program—then we find that our sample borrowed
approximately $27,000 more than national cumulative average. Whereas the national average
price of attendance for master’s programs in 2011-2012 was observed to be $20,700 (NCES,
2015), WUU cost $21,081 per year (WUU Tuition, 2012). The gap between what respondents
reported borrowing in our survey in comparison to the NCES data is large and in need of
attention. Students borrow to pay their tuition and fees and it is expected that where these
charges are above national averages loan balances will climb accordingly.
Prior studies of the relationship between attitudes and students’ educational borrowing
have documented that attitudes play a role in the decision to borrow or to abstain from borrowing
(Davies & Lea, 1995; Haultain et al., 2010). Our study corroborates these findings and adds that
certain other factors were related to students’ reports of borrowing for graduate study. First, we
found that the school in which one enrolls has an effect on patterns of borrowing. A chi-square
test for independence revealed a significant association between borrowing to pay for graduate
studies this year and the school of social work, where students comprised 62% of the total
students who reported borrowing, χ
2
(2, n = 501) = 26.84, p = .001, Cramer’s V = .231. Not only
did social work students borrow more often, they also borrowed more. An ANOVA revealed that
social work respondents reported borrowing significantly more (M = $41, 790) than either public
policy (M = $24, 407) or education (M = $27, 533) respondents: F (2, 465) = 13.67, p = .001.
These data suggest that program duration (number of units required to complete) and enrollment
intensity are important factors when considering how much master’s students borrow. As such,
institutional staff and faculty, including admissions and financial aid professionals, need to be
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 103
cognizant of their programs’ costs and the implications of full-time versus part-time enrollment
on borrowing decisions.
Our study highlights that WUU stakeholders need to pay attention to the amounts of debt
their students are accumulating. The New York Federal Reserve has been monitoring student
debt alongside other forms of household debt. Their researchers have found that student debt is
the only form of debt that continued to increase during and after the Great Recession (Lee,
2013). If students have expended a majority of their credit capacity in student loans, the
implication is that they are less likely to seek or obtain other forms of credit; most notably auto
loans and home mortgages (Korkki, 2014). New home loan originations declined substantially
during and after the Great Recession and as student debt amounts increase so too do the odds of
default and being denied a home loan (Lee, 2013). If educators are concerned with students’ life
outcomes (career, marriage, home ownership), they should be concerned with how much their
students are financing to attend their institutions.
President Obama’s “gainful employment rule” was implemented to prevent students from
being buried by student loan debt (U.S. Dept. of Education, 2014). Though this rule is intended
only for career colleges, it could theoretically be extended across the postsecondary educational
landscape as debt balances rise along with the consequences for borrowers. Researchers are
finding that educational debts are negatively related to outcomes associated with the “American
Dream” (Brown et al., 2014; Korkki, 2014; Lee, 2013; Neil, 2014; Ungarino, 2014). Educators
and policymakers have to be concerned as much with their students’ learning outcomes as their
students’ career and life outcomes.
Thus another implication for practice is that admissions and financial aid professionals
document their students’ returns on investment as a function of employment and income within
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 104
the first year of graduation. We collected data on respondents’ employment and income to
control for variance in the amounts they reported borrowing. However, it would be
immeasurably helpful—not to mention morally praiseworthy—for schools to provide accurate
and timely data on graduate students’ financial and employment outcomes. There is currently a
move towards greater transparency on outcomes of students through federal initiatives like the
College Scorecard or gainful employment rules for for-profit institutions. However, rather than
propose that we have more federal mandates, it seems that institutions that offer graduate
programs have an obligation to proactively share the outcomes of their students for the
enrichment of their prospective students. We found that respondents’ levels of employment were
not significantly related to mean amounts borrowed: F (2, 433) = 2.77, p = .063. While self-
reported employment and income information is captured and used during the financial aid
processing using the FAFSA, it is useful data to have to determine post-program gains;
especially where we are concerned with decisions made from a human capital investment
perspective.
Advice proffered in online magazines and blogs tends to focus on what students should
do before enrolling while failing to consider the constraints imposed by life circumstances and
geography. For example, Forbes ran an article exhorting individuals to research program costs
and job prospects (Shin, 2014). Another commentator adds that where one chooses to enroll has
an effect on ROI (Capuzzi Simon, 2011). While this is good advice, it erroneously assumes an
even distribution of social capital and an equal opportunity to enroll whereby one maximizes
one’s utility. Such considerations are integral to how students frame their enrollment aspirations
and ultimately move through postsecondary education (Mullen, Goyette, & Soares, 2003; Perna,
2004; Perna & Titus, 2005). Moreover, perceived value is an important variable in decision-
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 105
making processes (Ariely, 2009; Lee, Frederick, & Ariely, 2006) and students have been shown
to respond excessively to loans when admitted to their top choice schools (Avery & Hoxby,
2004). As such, it is not sufficient to simply provide information that students must seek out on
their own. Rather, financial aid and admission professionals need to find ways to actively seek
out and reach their target audiences to provide their expert advice. Considering the important
role these professionals play in the enrollment process and that perceptions of value influence
behavior, it is important that financial aid and admission professionals advise students in a
conscientious and ethical manner. Consider the following examples of how our respondents
sought out information pertaining to student loans.
Our Loan Information Seeking (LIS) items contained categorical variables to determine
whether respondents sought information about student loans from several possible sources. Our
basic finding was that students were not apt to seek information from professionals or family
members and were likelier to use online resources. A chi-square test for each category was
conducted to determine whether there was a relationship between loan information seeking
behavior and borrowing to pay for graduate school. Results for seeking information from a
financial aid professional were significant where the majority of respondents did not seek
information (n = 252), but did in fact borrow: χ
2
= (1, n = 501) = 10.71, p = .001, phi = .152.
Results for seeking information from admissions professionals were significant where 270
respondents indicated not seeking information, but reported borrowing: χ
2
= (1, n = 495) = 17.28,
p = .001, phi = .193. There were also significant findings for seeking information from friends
and family where many respondents did not consult their personal networks (n = 257), but
borrowed nonetheless: χ
2
= (1, n = 493) = 8.13, p = .002, phi = .134. The key difference among
these items was that students were significantly more likely to consult online resources (n = 322)
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 106
when making the decision to borrow: χ
2
= (1, n = 493) = 37.99, p = .001, phi = .284. Our
respondents did not seek information from experts (admissions and financial aid professionals)
and turned to online resources to inform their borrowing decisions. Such behavior is not rational
and has implications for amounts borrowed.
These findings show that students’ rational behavior is limited. They are not willing to
consult the professionals who have the expertise to shape borrowing decisions, but are willing to
consult online resources. As such, while it seems practical to recommend pushing more
information out to students regarding federal student loans, the need for professional guidance
cannot be understated. This is especially true for first-generation and underrepresented student
populations who bring different types and levels of knowledge to bear on matters of
postsecondary enrollment and financial aid (Perna, 2004; Perna & Titus, 2005; Tierney &
Venegas, 2009; Venegas, 2006).
Our research shows that students are willing to borrow as a function of having lower debt
aversion and higher spending compulsion. An implication for practice would be for institutions
to simplify the information collection process and to provide succinct reports on recent
graduates’ employment and salary outcomes. Such recommendations, however, may be risky for
some institutions seeking to secure full classes and its associated revenues.
Recent strides have been made towards linking undergraduate education with
employment and income outcomes. President Obama’s plan to protect undergraduate and
technical/career students from becoming stricken with debt resulted in an accountability plan
designed to strengthen existing rules for student lending while also tightening the linkages
between academia and industry. Our recommendation is for institutions to increase their
transparency when reporting recent master’s graduates’ career outcomes. Prospective students
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 107
should be afforded the opportunity to research their programs of interest and review reports of
graduates’ employment post-completion. Further, admissions officers can provide increased
transparency by making such information available in online and printed media and when
meeting with students in counseling sessions or graduate fairs. Again, we recognize such
recommendations are politically fraught. Nonetheless, we argue that the principles of rational
choice (Gilboa, 2010; Wittek et al., 2013) and behavioral economics (Ariely, 2009, Sunstein,
2006; Wittek et al., 2013) teach us that individuals will make bad decisions if presented with
inaccurate or incomplete information.
To summarize, our research suggests there are two approaches to helping manage
students’ federal debt loads. First, steps can be taken to reduce program costs. Our data revealed
that the amount students borrow is related to their program duration and enrollment intensity. We
recognize that the recommendation that schools cut program costs is infeasible, but insofar as
costs continue to increase so too will debt levels. We recommend increasing the footprint of the
role for financial aid professionals in the graduate admission and enrollment processes. Financial
aid professionals’ expert knowledge of the FAFSA and federal loan processes are huge assets
that can be tapped to ensure students borrow only what they need to fund their program of study.
Lastly, we recommend increased transparency in the admission process regarding recent
graduates success in obtaining gainful employment. To the extent that attitudes affect a portion
of the borrowing decision others have shown that attitudes can be changed as a function of
context (Dasgupta & Greenwald, 2001;Garcia-Marques, Santos, & Mackie, 2006) and how the
borrowing decision is framed (Prislin, 1996). If accurate and timely information is provided to
prospective master’s students, it is plausible that attitudes and attitude strength can be modified
in ways where students borrow at more reasonable levels.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 108
Future Research
Considering our limitations and implications for practice we provide several
recommendations for future research. The first involves a random sampling of master’s students.
Our second recommendation is to sample from more sites. Third, future researchers should take
steps to obtain actual federal loan data. Fourth, future researchers could build upon our scales
and items to further refine them and hopefully provide increased measurement precision. This
short section briefly considers each recommendation.
First, we recognize the problems arising from a convenience sampling technique using
self-selected recruitment and self-reported data. We used a non-random, convenience sampling
method and this was a major limitation in our research. Our sample appears to be different from
national datasets in several domains. This can be attributed to either actual differences between
our sample and national samples or sampling errors related to self-selection bias (Bethlehem,
2008). Future studies of master’s student debt and attitudes would benefit from a design that
recruits respondents using a random sampling technique.
Next, we must acknowledge that our site is somewhat unique among major research
universities in the U.S. WUU’s location in a major urban center, its majority graduate student
enrollment, and vast number of master’s programs compared to undergraduate offerings,
combine to provide a site where the population being studied differs from many other schools.
Our data revealed that WUU students borrow more than the national average cost per year for a
master’s program. We also found differences in the composition of our sample where females
outnumbered male respondents by a ratio of 3:1. We cannot say, however, whether our data
compares to peer institutions, public flagships, or more prestigious private institutions. Attitudes-
based research (Davies & Lea, 1995; Haultain et al., 2010), like ours, has established a
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 109
relationship between preferences and aversions and certain human behaviors (e.g. borrowing
money for school) and, as such, has value. Future research in this domain that samples from
public, private, and more/less selective institutions would broaden our understanding of whether
attitudes influence students’ borrowing decisions.
Future research on student debt should take steps to acquire actual student debt data
whenever practical. We speculate that self-reporting biases affected our dataset. This is
consonant with prior research (Adams et al., 2005; Rowland, 1990) showing that bias occurs
where controls for respondent characteristics are absent or deficient. As such, future research
should take steps to obtain permission from prospective sampling sites to obtain respondents’
federal loan data. We did not take this step owing to time constraints and considerations of
practicality. Yet, where researchers have the time and resources to do so, it would be helpful to
corroborate students’ reports of amounts borrowed with actual debt loads, as well as knowing
how much debt the student acquired while an undergraduate. This would allow researchers to
have accurate dependent variable data and would also facilitate the refinement of scale items
measuring respondents’ attitudes.
Our final recommendation for future research involves building upon prior research
(Davies & Lea; Haultain at al., 2010) and this dissertation to refine scales measuring students’
attitudes towards debt. We found that we had problems with scale reliability where our key
attitudinal measures (Debt Aversion and Spending Compulsion) had low observed Cronbach’s
alpha coefficients. We also found that our attitudinal measures contributed to the observed
variance in our dependent variable in small ways. This study was designed in such a way as to
validate prior claims that attitudes relate to students’ borrowing decisions. In that regard we were
successful in establishing such a relationship. Future research employing exploratory and
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 110
confirmatory factor analysis could help facilitate increased scale reliability by eliminating
extraneous scale items and pinpointing items with good intercorrelations.
Conclusion
This project set out to help solve the problem of rising student debt. Cumulative national
student debt was reported to be approximately $1.11 trillion when we began this project (Korkki,
2014). That number has recently been revised upwards to $1.2 trillion dollars along with
increases in student loan delinquency rates (Starkman, 2015). When we considered that graduate
students were found to borrow $57,000 on average (Bidwell, 2014) it was evident that graduate
student debt was a significant problem in higher education. Our inquiry builds upon recent
findings wherein students held that their decision to pursue graduate study and to incur student
loan debt was based on beliefs that a graduate education was an investment with good returns on
investment (Cooper et al., 2013). We opted for a quantitative design to capture inferential data
from which we hoped to draw useful conclusions about whether students’ debt preferences and
aversions influenced patterns of borrowing. Prior research on attitudes and borrowing (Davies &
Lea, 1995; Haultain et al., 2010) is sparse, but has shown that individuals’ attitudes are related to
patterns of borrowing.
Our data also revealed that attitudes were related to patterns of borrowing, but the
relationships, though significant, were generally weak. Our data corroborated the finding that, in
general, master’s students did not seek information regarding loan terms and conditions (Cooper
et al., 2013). Our data also revealed that graduate program costs are important predictors of
student loan debt. As such, our contribution to solving the problem of student debt consists of
showing that attitudes are a small, but important piece of the student debt problem.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 111
We conclude that attitudes-based research on students’ decisions to borrow for graduate
study is ripe for further inquiry. Our design was flawed in several key areas and addressing these
limitations could result in improved predictive power. The scales we used in this study need
further refinement and our sample would have benefitted from a random recruitment strategy. In
the final analysis, though, attitudes are only one part of the student-borrowing conundrum. The
Great Recession appears to have affected borrowing such that student loan debt was the only
debt to continue climbing after the onset of the recession (Brown, 2013) and student debt
continued to climb through the fourth quarter of 2014 (Sajdak, 2015). We know that decisions to
borrow are related to multiple intersecting social systems (Cooper et al., 2013) and attitude
formation regarding debt certainly falls within these domains. We are hopeful that further
refinement of attitudes scales through future research will result in the development of simple
diagnostic tools that can help graduate enrollment managers and financial aid professionals to
identify individuals who, all other things being equal, may be attitudinally inclined to borrow
more.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 112
References
About the School of Education. (2014). Retrieved from: http://rossier.WUU.edu/about/
Adams, S. A., Matthews, C. E., Ebbeling, C. B., Moore, C. G., Cunningham, J. E., Fulton, J., &
Hebert, J. R. (2005). The effect of social desirability and social approval on self-reports
of physical activity. American journal of epidemiology,161(4), 389-398.
Akers, B. (2014). Reconsidering the conventional wisdom on student loan debt and home
ownership. Retrieved from: http://www.brookings.edu/blogs/brown-center-
chalkboard/posts/2014/05/08-student-loan-debt-and-home-ownership-akers
Akers, B., & Chingos, M. M. (2014). Is a student loan crisis on the horizon? Retrieved from:
http://www.brookings.edu/~/media/research/files/reports/2014/06/24%20student%20loan
%20crisis%20akers%20chingos/is%20a%20student%20loan%20crisis%20on%20the%20
horizon.pdf
Amaury, N., Barlow, L., & Crisp, G. (2006). Examining the tangible and psychosocial benefits
of financial aid with student access, engagement, and degree attainment. American
Behavioral Scientist, 49(12), 1636-1651.
Archibald, R. B., & Feldman, D. H. (2011). Why does college cost so much? New York, NY:
Oxford University Press.
Archuleta, K., Dale, A., & Spann, S. (2013). College students and financial distress: Exploring
debt, financial satisfaction, and financial anxiety. Journal of Financial Counseling and
Planning 24(2): 50-62.
Ariely, D. (2009). Predictably irrational: The hidden forces that shape our decisions. New
York, NY: Harper Collins.
Astin, A. W. (1993). What matters in college: Four critical years revisited. San Francisco, CA:
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 113
Jossey-Bass.
Avery, C., & Hoxby, C. M. (2004). Do and should financial aid packages affect students' college
choices? In C. Hoxby (Ed.), College choices: The economics of where to go, when to go,
and how to pay for it (pp. 239-302). Chicago, Il.: University of Chicago Press.
Avery, C., & Turner, S. (2012). Student loans: Do college students borrow too much or not
enough. Journal of Economic Perspectives, 26(1), 165-192.
Baum, S., & Steele, P. (2010). Who borrows the most? Bachelor’s degree recipients and high
levels of debt. Retrieved from:
http://advocacy.collegeboard.org/sites/default/files/Trends-Who-Borrows-Most-Brief.pdf
Becker, G. S. (1962). Investment in human capital: A theoretical analysis. Journal of Political
Economy, 70(5), 9-49.
Becker, G. S. (1993). Human capital: A theoretical and empirical analysis with special reference
to education (3rd ed.). Chicago: University of Chicago Press.
Belfield, C. R. (2013). Student loans and repayment rates: The role of for-profit
colleges. Research in Higher Education, 54(1), 1-29.
Bethlehem, J. (2008). How accurate are self-selection web surveys? Retrieved from:
http://www.cbs.nl/NR/rdonlyres/EEC0E15B-76B0-4698-9B26-
8FA04D2B3270/0/200814x10pub.pdf
Bidwell, A. (2014). How much loan debt is from grad students? More than you think. U.S. News
and World Report. Retrieved from:
http://www.usnews.com/news/articles/2014/03/25/how-much-outstanding-loan-
debt-is-from-grad-students-more-than-you-think
Board of Governors of the Federal Reserve System. (2012). Research resources: Survey of
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 114
consumer finances. Retrieved from:
http://federalreserve.gov/econresdata/scf/scf_2009p.htm
Brown, M. (2013). Student debt overview. Retrieved from:
http://www.newyorkfed.org/regional/Brown_presentation_GWU_2013Q2.pdf
Brown, M., Caldwell, S., & Sutherland, S. (2014). Young student loan borrowers remained on
the sidelines of the housing market in 2013. Retrieved from:
http://libertystreeteconomics.newyorkfed.org/2014/05/just-released-young-student-loan-
borrowers-remained-on-the-sidelines-of-the-housing-market-in-
2013.html#.U4Osf5RdURX
Burdman, P. (2005). The student debt dilemma: Debt aversion as a barrier to college
access. Center for Studies in Higher Education. Retrieved from:
https://escholarship.org/uc/item/6sp9787j
Callender, C. & Jackson, J. (2005). Does the fear of debt deter students from higher education.
Journal of Social Policy, 34(4), 509-540.
Capuzzi Simon, C. (2011). R.O.I. Retrieved from:
http://www.nytimes.com/2011/07/24/education/edlife/edl-24roi-t.html?pagewanted=all
Commonfund Institute. (2013). 2013 HEPI. Retrieved from:
https://www.commonfund.org/CommonfundInstitute/HEPI/HEPI%20Documents/
2013%20HEPI%20Report.pdf
Connelly, L. M. (2008). Pilot studies. Medsurg Nursing: Official Journal of the Academy of
Medical-Surgical Nurses, 17(6), 411.
Cook, K. S., & Chesire, C. (2013). Social exchange, power, and inequality in networks. In R.
Wittek, T. A. B. Snidjers, & V. Nee (Eds.), The handbook of rational choice social
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 115
research (pp. 185-219). Stanford, CA: Stanford University Press.
Cooper, Y., Dieken, P., Dorado, L., Gotto, M., Kreutzer, C., Morales, I., & Venegas, K.M.
(2013). Decision-making and graduate school debt: A focus on master's degree students
(Doctoral dissertation). Retrieved from University of Southern California Digital
Library. (etd-CooperMicha-2283.pdf)
Creswell, J. W. (2009). Research design: Qualitative, quantitative, and mixed methods
approaches. Los Angeles: Sage Publications.
Cunningham, A. F., Wellman, J. V., Clinedinst, M. E., & Merisotis, J. P. (2001). Study of
college costs and prices, 1988-89 to 1997-98. Retrieved from:
http://nces.ed.gov/pubs2002/2002157.pdf
Davies, E., & Lea, S. E. G. (1995). Student attitudes to student debt. Journal of Economic
Psychology, 16, 663-679.
Deritis, C. (2011). Student lending’s failing grade. Retrieved from:
http://image.exct.net/lib/fefb127575640d/m/2/Student+Lendings+Failing+Grade.pdf
Delisle, J. (2014). Graduate student debt review: The state of graduate student borrowing.
(New America Education Policy Program March 2014). Washington, D.C.: New
America.
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, A. C., & Coury, T. (2006). The effect of loans on the persistence and attainment
of community college students. Research in Higher Education, 47(1), 33-62.
Dwyer, R. E., Hodson, R., & McCloud, L. (2013). Gender, debt, and dropping out of college.
Gender and Society, 27(1), 30-55.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 116
Dynarski, S. (2003). Does aid matter? Measuring the effect of student aid on college attendance
and completion. American Economic Review, 93(1), 279-288.
Dynarski, S. (2008). Building the stock of college-educated labor. Journal of Human Resources,
43(3), 576-610.
Dynarski, S., & Scott-Clayton, S. M. (2006). The cost of complexity in federal student aid:
Lessons from tax theory and behavioral economics. National Tax Journal, 59(2), 319-
356.
Eckel, C. C., Johnson, C., Montmarquette, C., & Rojas, C. (2007). Debt aversion and the demand
for loans for postsecondary education. Public Finance Review, 35(2), 233-262.
Ehrenberg, R. G. (2000). Tuition rising: Why college costs so much. Cambridge, MA: Harvard
University Press.
Ehrenberg, R. G., & Sherman, D. R. (1984). Optimal financial aid policies for a selective
university. Journal of Human Resources, 19(2), 202-230.
Einhorn, H. J., & Hogarth, R. M. (1987). Decision making under ambiguity. In R. M. Hogarth &
M. W. Reder (Eds.), Rational choice: The contrast between economics and psychology
(pp. 41-66). Chicago, IL: University of Chicago Press.
ESD. (2011). Graduate Application Details Report. University of Southern California.
EverFi. (2014). Money matters on campus: How early attitudes and behaviors affect the
financial decisions of first-year college students. Washington, D.C.: EverFi.
Falk, A., Meier, S., & Zehnder, C. (2013). Do lab experiments misrepresent social preferences?
The case of self ‐selected student samples. Journal of the European Economic
Association, 11(4), 839-852.
Franke, R. (2014). Towards the education nation? Revisiting the impact of financial aid, college
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 117
experience, and institutional context on baccalaureate degree attainment for low-income
students. Paper presented at Annual Meeting of the American Educational Research
Association. Retrieved from: http://www.aera.net/Portals/38/Newsroom%20-
%20Recent%20Research/Updated%20Towards%20the%20Education%Nation.pdf
Gachter, S. (2013). Rationality, social preferences, and strategic decision-making from a
behavioral economics perspective. In R. Wittek, T. A. B. Snidjers, & V. Nee (Eds.), The
handbook of rational choice social research (pp. 33-72). Stanford, CA: Stanford
University Press.
Gandhi, S. J. (2007). Understanding students from a behavioral economics perspective: How
accelerating student loan subsidies generates more bang for the buck. Kansas Journal of
Law and Public Policy, 17, 130-167.
Gilboa, I. (2010). Rational choice. Cambridge, MA: MIT Press.
Goldrick-Rab, S., Harris, D., & Trostel, P. (2009). Why financial aid matters (or does not) for
college success: Toward a new interdisciplinary perspective. In J. C. Smart (Ed.), Higher
Education: Handbook of Theory and Research (pp. 1-45). Springer Netherlands.
Haultain, S., Kemp, S. Chernyshenko, O. (2010). The structure of attitudes to student debt.
Journal of Economic Psychology, 31, 322-330.
Heller, D. (2014a). Is the $1 trillion student loan debt really a crisis? Retrieved from:
http://www.washingtonpost.com/blogs/answer-sheet/wp/2014/05/01/is-the-1-trillion-
student-loan-debt-really-a-crisis/
Heller, D. (2014b). Is the $1 trillion student loan debt really a crisis? Part 2. Retrieved from:
http://www.washingtonpost.com/blogs/answer-sheet/wp/2014/05/30/is-the-1-trillion-
student-loan-debt-really-a-crisis-part-2/
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 118
Hill, R. (1998). What sample size is “enough” in internet survey research. Interpersonal
Computing and Technology: An electronic journal for the 21st century, 6(3-4), 1-12.
Hogarth, R. M., & Reder, M. W. (Eds.)(1987). Rational choice: The contrast between
economics and psychology. Chicago, IL: University of Chicago Press.
Inceptia. (2013, January). College students are put to the test: The attitudes, behaviors and
knowledge levels of financial education (Research brief). Lincoln, NE: Trombitas, K. &
Lannan, T.
IPEDS. (2013). IPEDS data feedback report: 2013. Retrieved from:
http://nces.ed.gov/ipeds/DataCenter/DfrFiles/IPEDSDFR2013_123961.pdf
Irwin, N. (2014). How student debt may be stunting the economy. Retrieved from:
http://www.nytimes.com/2014/05/15/upshot/the-role-of-student-debt-in-stunting-the-
recovery.html?smid=fb-
nytimes&WT.z_sma=UP_HSD_20140515&bicmp=AD&bicmlukp=WT.mc_id&bicmst=
1388552400000&bicmet=1420088400000&_r=4
Jaeger, D. A., and Page, M. E. (1996). Degrees matter: New evidence on sheepskin effects in the
returns to education. The Review of Economics and Statistics, 78(4): 733-740.
Johanson, G. A., & Brooks, G. P. (2009). Initial scale development: sample size for pilot
studies. Educational and Psychological Measurement.
Johnstone, B. (2003). Cost sharing in higher education: Tuition, financial assistance, and
accessibility in a comparative perspective. Czech Sociological Review, 39(3), 351-374.
Jones, D. P. (2000). An alternative look at the cost question. Retrieved from:
http://www.ihep.org/%5Cassets%5Cfiles%5C/publications/G-
L/HigherEdCostMeasurement.pdf
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 119
Julious, S. A. (2005). Sample size of 12 per group rule of thumb for a pilot study.
Pharmaceutical Statistics, 4(4), 287-291.
Kim, D., & Eyermann, T. (2006). Undergraduate borrowing and its effects on plans to attend
graduate school prior to and after the 1992 Higher Education Act Amendments. Journal
of Student Financial Aid, 36(2), 5-21.
Korkki, P. (2014). The ripple effects of rising student debt. Retrieved from:
http://www.nytimes.com/2014/05/25/business/the-ripple-effects-of-rising-
student-debt.html?_r=0
Krogstad, J. M., & Cohn, D. (2014). U.S. Census looking at big changes in how it asks about
race and ethnicity. Retrieved from: http://www.pewresearch.org/fact-tank/2014/03/14/u-
s-census-looking-at-big-changes-in-how-it-asks-about-race-and-ethnicity/
Kurpius, S. E. R., & Stafford, M. E. (2006). Testing and measurement: A user-friendly guide.
Thousand Oaks, CA: Sage Publications.
Lee, D. (2013). Household debt and credit: Student debt. Retrieved from:
http://www.newyorkfed.org/newsevents/mediaadvisory/2013/Lee022813.pdf
Lee, L., Frederick, S., & Ariely, D. (2006). Try it, you’ll like it: The influence of expectation,
consumption, and revelation on preferences for beer. Psychological Science, 17(12),
1054-1058.
Lewin, T. (2011, April 11). Burden of College Loans for Graduate Loans. New York Times.
Retrieved from: http://www.nytimes.com/2011/04/12/education/12college.html
Lillis, M. P. (2008). High tuition, high loan financing: Economic segregation in postsecondary
education. Journal of Education Finance, 34(1), 15-30.
Malcom, L. E., & Dowd, A. C. (2012). The impact of undergraduate debt on the graduate school
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 120
enrollment of STEM baccalaureates. Review of Higher Education, 35(2), 265-305.
Mare, R. D. (1980). Social background and school continuation decisions. Journal of the
American Statistical Association, 75(370): 295-305.
McCann, C. (2013). New Pell grant, federal loan data reveal changing tides in financial aid.
Retrieved from:
http://edmoney.newamerica.net/blogposts/2013/new_pell_grant_federal_loan_data_revea
l_changing_tides_in_financial_aid-92009
Mertz, N., Strayhorn, T., & Eckman, E. (2012). Entering student affairs: A comparative study
of graduate school choice. College Student Affairs Journal, 30(2): 1-14.
Metcalf, H. (2005). Paying for university: The impact of increasing costs on student
employment, debt and satisfaction. National Institute Economic Review, 191: 106-117.
Millett, C. M. (2003). How undergraduate loan debt affects application and enrollment in
graduate or first professional school. The Journal of Higher Education, 74(4), 386-427.
Money Matters on Campus. (2014). Retrieved from: http://moneymattersoncampus.org/
Mueller, T. (2013). Changes to the student loan experience: Psychological predictors and
outcomes. Journal of Student Financial Aid, 43(3): 148-164.
Mullen, A. L., Goyette, K. A., & Soares, J. A. (2003). Who goes to graduate school? Social and
academic correlates of educational continuation after college. Sociology of Education,
143-169.
NASW. (2003). Demographics. Retrieved from:
http://www.socialworkers.org/naswprn/surveyTwo/Datagram2.pdf
NCES. (2013). Average graduate tuition and required fees in degree-granting postsecondary
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 121
institutions, by control of institution and percentile: 1989-90 through 2012-13. Retrieved
from: http://nces.ed.gov/programs/digest/d13/tables/dt13_330.50.asp
NCES. (2015). Trends in graduate student financing: Selected years 1995-96 to 2011-12.
Retrieved from: http://nces.ed.gov/pubs2015/2015026.pdf
NPR. (2014). Senator warns of a student loan bubble. Retrieved from:
http://www.npr.org/2014/03/27/294858103/senator-warns-of-a-student-loan-bubble
NPSAS. (2012a). 2011-12 National Postsecondary Student Aid Study: Race/ethnicity by
graduate degree.
NPSAS. (2012b). 2011-12 National Postsecondary Student Aid Study: Cumulative amount
borrowed by graduate degree.
O’Donoghue, T., & Rabin, M. (1999). Doing it now or later. American Economic Review, 89(1),
103-124.
Pallant, J. (2007). SPSS Survival Manual. New York, NY: McGraw Hill.
Perna, L. W. (1998). The contribution of financial aid to undergraduate persistence. Journal of
Student Financial Aid, 28(3), 25-40.
Perna, L. W. (2004). Understanding the decision to enroll in graduate school: Sex and
racial/ethnic group differences. Journal of Higher Education, 487-527.
Perna, L. W., & Titus, M. A. (2005). The relationship between parental involvement as social
capital and college enrollment: An examination of racial/ethnic group
differences. Journal of Higher Education, 485-518.
Perna, L. W. (2006a). Studying college access and choice: A proposed conceptual model. In J.C.
Smart (Ed.), Higher Education: Handbook of theory and research (pp. 99-157). Springer
Netherlands.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 122
Perna, L. W. (2006b). Understanding the relationship between information about college prices
and financial aid and students’ college-related behaviors. American Behavioral Scientist,
49(12), 1620-1635.
Perna, L. W., & Titus, M. A. (2005). The relationship between parental involvement as social
capital and college enrollment: An examination of racial/ethnic group
differences. Journal of Higher Education, 485-518.
Public Policy School Facts. (2014). Retrieved from: http://priceschool.usc.edu/about/
Radwin, D., Wine, J., Siegel, P., & Bryan, M. (2013). 2011-12 National postsecondary student
aid study (NPSAS): Student financial estimates for 2011-12. U.S. Department of
Education. Retrieved from: http://nces.ed.gov/pubs2013/2013165.pdf
Raosoft. (2004). Sample size calculator. Retrieved from:
http://www.raosoft.com/samplesize.html
Rowland, M. L. (1990). Self-reported weight and height. The American journal of clinical
nutrition, 52(6), 1125-1133.
Sajdak, K. (2015). Household debt continues upward climb while student loan delinquencies
worsen. Retrieved from:
http://www.newyorkfed.org/newsevents/news/research/20150/rp150217.html
Salkind, N. J. (2011). Statistics for people who think they hate statistics. Thousand Oaks, CA:
Sage Publications.
Shin, L. (2014). Is Grad School Worth It? 7 Steps To Calculating The ROI. Retrieved from:
http://www.forbes.com/sites/laurashin/2014/09/30/is-grad-school-worth-it-7-steps-to-
calculating-the-roi/
Shin, J., & Ariely, D. (2004). Keeping doors open: The effect of unavailability on incentives to
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 123
keep options viable. Management Science, 50(5), 575-586.
Snyder, T. D., and Dillow, S. A. (2013). Digest of Education Statistics 2012 (NCES 2014-015).
National Center for Education Statistics, Institute of Education Sciences, U.S.
Department of Education. Washington, D.C.
Social Work Salaries. (2015). Retrieved from: http://socialworklicensemap.com/social-worker-
salary/
Starkman, D. (2015). New York Fed: Household debt, student loan delinquencies climb.
Retrieved from
http://www.latimes.com/business/la-fi-ny-fed-household-debt-20150217-story.html
Stolzenberg, R. M. (1994). Educational continuation by college graduates. American Journal of
Sociology, 99:1042-77.
Straight talk about college costs and prices (1998). Retrieved from:
http://www.nyu.edu/classes/jepsen/costreport.html
Sunstein, C. R. (2006). Boundedly rational borrowing. University of Chicago Law Review, 73(1),
249-270.
The Institute for College Access & Success. (2014). Quick facts about student debt. Retrieved
from: http://bit.ly/1lxjskr.
Tierney, W. G., & Venegas, K. M. (2009). Finding money on the table: Information, financial
aid, and access to college. The Journal of Higher Education, 80(4), 363-388.
Tinto, V. (1987). Leaving college: Rethinking the causes and cures of student attrition. Chicago,
IL: University of Chicago Press.
Tversky, A., & Kahneman, D. (1987). Rational choice and the framing of decisions. In R.M.
Hogarth & M. W. Reder (Eds.), Rational choice: The contrast between economics and
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 124
psychology (pp. 67-100). Chicago, IL: University of Chicago Press.
Tversky, A., & Kahneman, D. (1991). Loss aversion in riskless choice: A reference-dependent
model. Quarterly Journal of Economics, 106(4), 1039-1061.
U.S. Department of Education, Office of Student Financial Aid Programs. (2010a). Table:
Institutional Default Rate Comparison of FY 2006, 2007, and 2008 Cohort Default Rates.
Available at: http://www2.ed.gov/offices/OSFAP/ defaultmanagement/instrates.html
U.S. Department of Education, Office of Student Financial Aid Programs. (2010b). National
Student Loan Default Rates. Available at:
www.ed.gov/offices/OSFAP/defaultmanagement/defaultrates.html.
U.S. Department of Education. (2013). Average undergraduate tuition and fees and room and
board rates charged for full-time students in degree-granting institutions, by level and
control of institution: 1969-70 through 2011-12. Retrieved from:
http://nces.ed.gov/programs/digest/d12/tables/dt12_381.asp
U.S. Department of Education (2014). Obama administration announces final rules to protect
students from poor-performing career college programs. Retrieved from:
http://www.ed.gov/news/press-releases/obama-administration-announces-final- rules-
protect-students-poor-performing-career-college-programs
WUU. (2013). Schedule of classes: Tuition and fees. Retrieved from:
http://www.WUU.edu/academics/classes/term_20131/tuition_and_fees.html
WUU Financial Aid. (2014). Retrieved from:
http://www.WUU.edu/admission/fa/applying_receiving/graduates2/costs.html
WUU Social Work. (2014). Retrieved from: http://sowkweb.WUU.edu/document/8576
WUU Tuition. (2012). Retrieved from:
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 125
http://web-app.WUU.edu/ws/soc_archive/soc/term_20121/tuition_and_fees.html
WUU Tuition. (2014). Retrieved from:
http://web-app.usc.edu/ws/soc_archive/soc/term_20141/tuition_and_fees.html
Vedder, R. (2004). Going broke by degree: Why college costs so much. Washington, D.C.: AEI
Press.
Venegas, K.M. (2006). Internet inequalities: Financial aid, the internet, and low-income students.
American Behavioral Scientist, 49, 1652-1669
Volkwein, J. F., Szelest, B. P., Cabrera, A. F., & Napierski-Prancl, M. R. (1998). Factors
associated with student loan default among different racial and ethnic groups. Journal of
Higher Education, 69(2), 206-237.
Wei, C. C. (2010). What is the price of college? Total, net, and out-of-pocket prices in 2007-08.
Retrieved from: http://nces.ed.gov/pubs2011/2011175.pdf
Weiler, W. C. (1994). Expectations, undergraduate debt, and the decision to attend graduate
school: A simultaneous model of student choice. Economics of Education Review, 13(1),
29-41.
Whitsett, H. & Mishory, J. (2012). High debt, low information: A survey of student loan
borrowers. New York, NY: NERA Economic Consulting.
Williams, J. (2006). The pedagogy of debt. College Literature, 33(4), 155-169.
Williams, J. (2008). Student debt and the spirit of indenture. Dissent, 55(4), 73-78.
Winston, G. C. (1999). Subsidy, hierarchy, and peers: The awkward economics of higher
education. Journal of Economic Perspectives, 13(1), 13-36.
Yoon, I. (2012). Debt burdens among MSW graduates: A national cross-sectional study. Journal
of Social Work Education, 48(1): 105-125.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 126
Youngclaus, J. A., Koehler, P. A., Kotlikoff, L. J., & Wiecha, J. M. (2013). Can medical
students afford to choose primary care? An economic analysis of physician education
debt repayment. Academic Medicine, 88(1), 16-25.
Zhang, L. (2010). Effect of college educational debt on graduate school attendance and early
career and lifestyle choices. Retrieved from Social Science Research Network:
http://ssrn.com/abstract=942793

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 127
Appendix A
Survey Instrument

1. Are you an international student?
Yes-1 No-2
[Skip logic to end of survey if Yes]

2. Did you borrow student loans to pay for your graduate education this year?
Yes-1 No-2
[Skip logic to KFC if No]

3. How much did you borrow for your graduate education this year?
Drag the slider until you get close to the amount you borrowed then click the number to
the right and input the actual amount. [Slider from $0 to $100,000]

Beliefs about Knowledge across Financial Competencies (KFC)

Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

KFC1. I believe my knowledge of budgeting is good:
KFC2. I believe my knowledge of credit cards is good:
KFC3. I believe my knowledge of insurance is good:
KFC4. I believe my knowledge of student loans is good:
KFC5. I believe my knowledge of investing is good:
KFC6. I believe my knowledge of compound interest is good:
KFC7. I believe my knowledge of retirement planning is good:
KFC8. I believe my knowledge of credit reporting bureaus is good:

Loan Information Seeking (LIS)
LIS1. Did you seek information about federal student loans from a financial aid professional?
Yes-1 No-2

LIS2. The information I received from a financial aid professional was influential on my federal
loan borrowing decision:
Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

LIS3. Did you seek information about federal student loans from an admission professional?
Yes-1 No-2

LIS4. The information I received from an admission professional was influential on my federal
loan borrowing decision:
Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

LIS5. Did you seek information about federal loans using online resources?
Yes-1 No-2
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 128
LIS6. Information gathered online was influential in my federal loan borrowing decision:
Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

LIS7. Did you seek information about federal student loan repayment from a family member or
friend?
Yes-1 No-2

LIS8. My friends and/or family influenced my federal loan borrowing decision:
Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

Financial Efficacy (FE)

Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

FE1. I feel capable of using my future income to achieve my financial goals:
FE2. I feel capable of handling my financial future:
FE3. I feel in control of my current financial situation:

Financial Behavior (FB)

Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

FB1. I stick to a monthly budget:
FB2. I save on a regular basis
FB3. I regularly monitor my spending:
FB4. I compare terms of repayment before I borrow:
FB5. I check my credit report quarterly:

Debt Aversion (DA)

Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

DA1. There is no excuse for borrowing money:
DA2. Owing money is wrong:
DA3. I prefer using my debit card to make purchases:
DA4. [Reversed] It is okay to borrow money to buy food:
DA5. The lower a person’s income, the more important it is to save every month:
DA6. You should always save up first before buying something:
DA7. You should stay home rather than borrow money to go out for evening on the town:
DA8. I worry about debts:
DA9. Once you are in debt it is very difficult to get out:

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 129
Spending Compulsion (SC)

Strongly disagree-1 Disagree-2 Agree-3 Strongly agree-4

SC1. Debt is an integral part of today’s lifestyle:
SC2. Taking out a loan is a good thing because it allows you to enjoy life as a student:
SC3. I prefer spending:
SC4. Students have to go into debt:
SC5. It is better to have something now and pay for it later:
SC6. Buying things gives me a lot of pleasure:
SC7. I enjoy spending money on things that aren’t practical:
SC8. I prefer saving:
SC9: I prefer using my credit card to make purchases:

Demographic

D1. What is your present employment status?
Full time-1
Part time-2
Not currently employed-3

D2. Please estimate your household income:
Use slider or double click the number to the far right to manually input your annual
income. [Slider from $0 to $100,000]

D3. What is your current marital status?
Single-1
Unmarried, but living with partner-2
Married-3
Divorced-4
Widowed-5

D4. Gender:
Male-1
Female-2
Trans/other-3

D5. Age:
Use slider to input your current age [Slider from 18-100]

D6. Are you of Hispanic, Latino or Spanish origin?
Yes-1 No-2
[Skip logic to D8 if Yes]

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 130

D7. Race/Ethnicity:
White-1
Black, African American-2
American Indian or Alaska Native-3
Asian Indian-4
Chinese-5
Filipino-6
Japanese-7
Korean-8
Vietnamese-9
Native Hawaiian-10
Guamanian or Chamorro-11
Samoan-12
Other Asian-13
Other Pacific Islander-14

D8. Which School at WUU are you enrolled in:
School of Public Policy-1
School of Education-2
School of Social Work-3

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 131
Appendix B
Pilot Survey

We conducted a small-scale pilot test of our survey to obtain estimates of item reliability,
construct validity, and to determine whether certain items made sense to respondents. Data for
the pilot was collected over a two-week period using social media (Facebook) to identify
respondents. The target for the sample size of the pilot study was set at 10% of the target of the
final sample size. This decision was consistent with methodological studies related to
determining appropriate sample sizes (Connelly, 2008; Hill, 1998; Johanson & Brooks, 2009,
Julious, 2005). The pilot study collected data from current and former WUU master’s students
only. This also followed from considerations of the estimated size of the WUU master’s student
population. If the population of master’s students enrolled at WUU is approximately 7,000 and
our sample size goal is set at 378 cases, this is supposed to yield 95% confidence and a 5%
margin of measurement error (Raosoft, 2004). As such, our goal was to obtain at least 38
responses to meet the basic guidelines for pilot study sample size. Due to time and resource
constraints, however, we did not hit this goal.
Besides obtaining reliability estimates and testing the validity of our instrument, the pilot
was intended to guide the main data collection effort of our research. The main idea of any pilot
study is to refine the mechanics of the study and establish its overall feasibility (Connelly, 2008).
One of our early findings during the pilot study was that it took two weeks to gather enough
responses to finish the pilot. This information led to our decision to revise our data collection
approach for the final phase of this study. As such, the pilot was integral to getting our survey
tuned up and helping us plan the optimal strategy for getting enough responses for our final data
collection.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 132
Pilot Study Data
The sampling goal for the pilot study was 38 responses within the two weeks allotted for
this stage of the project. 33 responses to our pilot study survey were collected within that
timeframe and constitute the sample upon which decisions about the main data collection were
predicated. However, nearly half of the 33 responses were missing response data for one or more
variables. For instance, there were 15 valid responses to our question regarding whether
respondents borrowed federal loans. The follow-up question, which gathered our dependent
variable data (“How much did you borrow for graduate education this year?”), only had nine
valid responses. Loan Information Seeking items were also left incomplete more often than not.
For instance, our question “The information I received from a financial aid or admission
professional was influential…” was ignored by 84.8% of respondents (n = 28). As such, we
decided to proceed with the pilot study data analysis despite the fraction of valid responses
available. The following paragraphs outline the general features of the sample collected in the
pilot. This includes descriptive data regarding respondents’ sex, age, race, ethnicity, estimated
amounts of federal loans borrowed, and responses to survey questions.
While there were more female respondents in the sample there were also differences in
the mean and modal ages of males and females. 14 cases were missing responses for gender. As
such, females consisted of 57.9% of the response pool. Conversely, males consisted of 42.1% of
the response pool. The mean (n = 11) and modal (n = 3) age of surveyed females was observed
to be 29 years. Males consisted of 42.1% of the pilot sample and reported a mean age of 34.75
years. Overall, the pilot data revealed that age was positively skewed; 84.2% respondents
reported being between the ages of 24 and 34 years. Females were slightly overrepresented in the
pilot study sample. There were 11 females and 8 males total in the valid response pool.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 133
The racial and ethnic makeup of the sample skewed White and non-Hispanic/ Latino. In
comparison to WUU demographics, overall, the distribution was proportional. Whereas WUU
has a 5% African American population, we recorded a 13.33% response rate for African
Americans. Similarly, WUU has a fairly large Asian population (18%) and we recorded a
response rate of 21.34% for Asian self-identification. Race and ethnicity were divided into
separate categories in the pilot and main survey. This followed the convention set by the U.S.
Census Bureau whereby Hispanic/ Latino ethnicity is distinguished from racial categories
(Krogstad & Cohn, 2014). Four respondents identified as being of Hispanic, Latino, or Spanish
origin. The racial makeup of the sample was evenly divided among the several racial categories.
Table 1 contains a summary of the descriptive data for our pilot study respondents.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 134
Table 14
Pilot Study Respondent Characteristics
CHARACTERISTIC n CHARACTERISTIC Amount
GENDER

INCOME

Male 56 25th percentile $15,000
Female 262 50th percentile $40,000
Transgender 0 75th percentile $44,830
Totals 318 Mean $40,000

Median $40,423
RACE/ ETHNICITY

Mode $ 0
African American 31 Standard Deviation $36,689
American Indian 2

Asian 13

Hispanic/ Latino 135

Pacific Islander 7 AMOUNT BORROWED
White 132 10th percentile $10,040

20th percentile $20,000
MARITAL STATUS 30th percentile $20,977
Single 153 40th percentile $26,023
Unmarried, but living with a partner 33 50th percentile $34,967
Married 110 60th percentile $40,925
Divorced 20 70th percentile $50,000
Widow/ Widower 4 80th percentile $64,973

90th percentile $100,000
EMPLOYMENT

Mean $41,790
Full time 93 Median $34,790
Part time 101 Mode $100,000
Unemployed 124 Standard Deviation $29,791

Measures of the dependent variable (how much did respondents borrow in federal loans
for the current academic year) were varied and a broad range. The maximum observed amount
borrowed was $100,000 and the range was $97,000. The mean amount borrowed was
approximately $40,100 (SD = $27,851).
Analysis of scale items was integral to the final design of our survey. We found items
that needed to be reverse-coded, others that had low reliability, and some items that were
duplicates. For instance, Cautious Financial Attitudes (CFA) had little or no reliability with an
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 135
observed Cronbach’s alpha coefficient of .03. While the low alpha for CFA items suggested that
the scale was not going to be useful for our main data collection, we recognized parallels
between CFA items and items from our Debt Aversion (DA) scale. Whereas being overly
cautious and indecisive is related to loss aversion (Ariely, 2006; Gandhi, 2007; Shin & Ariely,
2004; Tversky & Kahneman, 1991), it followed that our CFA items were actually testing for
measures related to debt aversion. As such, we decided to merge CFA items with Debt Aversion
items.
High scores for items in the Beliefs about Knowledge across Financial Competencies
(KFC) scale are indicative strong beliefs in one’s knowledge of financial instruments.
Cronbach’s alpha coefficients for KFC items was observed at .83 suggesting good reliability.
Financial competence is linked to rational choice insofar as high measures of financial
competence are held to facilitate cost-benefit calculations (Perna, 2006a). As such, KFC items
were retained for inclusion in the final version of our survey.
Loan Information Seeking (LIS) items are intended to categorize respondents as a
function of whether they behaved rationally (i.e. sought information regarding the costs and
benefits of borrowing for graduate school). These categorical questions were then followed with
items asking respondents to rate their level of agreement with statements of the whether these
sources of information were good. Cronbach’s alpha coefficients for LIS items were unusually
high having scored a perfect 1.00. The likely explanation is that this is an artifact of the small
sample size in the pilot study. LIS items were retained for the full data collection to determine
whether the high Cronbach’s alpha was in fact measurement error. LIS items were also retained
to facilitate future analyses where nominal-level data could be used to analyze any differences in
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 136
the frequencies of information seeking behaviors and to analyze differences in the mean level of
agreement with LIS statements.
The next scale, Borrowing for Educational Costs (BEC), was supposed to measure how
respondents used the money they had borrowed for graduate studies. Cronbach’s alpha
coefficients for BEC items was observed at .82 indicating good reliability. These items, however,
were removed from the final survey, as they did not directly address our research questions.
They were also removed because respondents expressed confusion about these items and asked
why these items were in the survey.
Financial Efficacy (FE) was meant to measure the respondents’ confidence in their ability
to control their own financial situation in the present as well as in the future. This scale was left
intact at three items, as the Cronbach’s alpha coefficients for FE items was observed at .84
indicating good reliability.
Our analysis of the pilot data convinced us that certain scales had to be merged to form
one overarching construct. These scales were Financial Behavior (FB) and Financial
Management Behaviors (FMB). These were merged and renamed Financial Behavior (FB) to
reflect our focus on determining whether respondents’ debt attitudes shaped their borrowing
behaviors. The Cronbach’s alpha coefficient for this merged scale was observed at .82 and
indicated good overall reliability. Three items from the FMB scale were eliminated after the pilot
study: FMB 2, 4, and 5. Items were reworded for clarity and applicability to our research
population. FB scale items were retained to represent the behavior side of the attitude-behavior
link. If we accept that intentions and attitudes shape behaviors (Ajzen, 1991), it follows that we
should capture behavioral data with our survey.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 137
Lastly, we measured the construct of Debt Aversion (DA) first using separate scales only
to find that our analysis was improved by merging it with other scales and reverse coding items
that represented pro-debt attitudes. DA as an independent scale had an observed Cronbach’s
alpha coefficient of .25 suggesting low reliability. We conceptualized debt aversion as a negative
debt attitude wherein high measures of debt aversion would be negatively associated with
reported amounts borrowed. Likewise, CFA items represented an overall anti-debt attitude where
one is cautious about one’s spending and deficit financing. As such, CFA items were also
merged with DA items. When all the items were analyzed together the resulting Cronbach’s
alpha was observed to be .51. The gains in reliability are likely a function of the increased
number of items (Pallant, 2007), but this may also be related to the fine-tuning of the overall DA
scale.
We also merged Debt as Necessity (DN) items with Spending Compulsion (SC) items
because acceptance of debt is likely correlated with overall willingness to spend. Research has
shown that high spending compulsion is considered a financially risky behavior (Brougham et
al., 2011; Gutter & Copur, 2011; Roberts & Jones, 2001). We theorized that for our main data
collection, SC items would have a positive relationship with our dependent variable of how
much respondents reported borrowing.
Summary
The pilot study was an integral first step for this research. It helped the research team
identify the strengths and weaknesses of the survey design. We identified items that needed to be
reverse-coded and other items that could be merged to form more conceptually sound scales for
our data collection. Items were removed that did not directly address the research questions and
thereby reduced the overall length of the final survey instrument. The final scales of the survey
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 138
address attitudes in a direct and more concise way. Our concern for this research was to measure
whether individuals’ pro or anti debt attitudes were related to variance in the amounts
respondents indicated they had borrowed to pay for one year of graduate study. This concern was
born from our recognition that attitudes-based research on student debt is sparse and has not
directly addressed our population of interest: master’s students. Items relating to demographics,
employment and employment intensity, and income are important control variables that were
retained in the final survey design. Our pilot study clarified key areas of improvement and areas
in which our design was functional. The pilot study enabled us to proceed with greater
confidence that our survey was measuring debt attitudes and that it also captured data pertaining
to whether these attitudes were related to patterns of student loan borrowing.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 139
Appendix C
Note on Author Contributions

This dissertation was collaboratively conceptualized and written by the two authors. Each
author contributed original writing for sections in each chapter, and both authors’ writing was
edited and revised by each other in a collaborative manner. Please see below for details on each
chapter’s specific contributions.
Chapter One: Overview of the Study
Chapter 1 was jointly conceptualized and executed by Emily Chung and Rick Garcia. An
outline of the chapter and set of research questions were developed together. Rick wrote the
initial draft of the introduction, background of the problem, and purpose of the study. Emily
wrote the initial draft of the overview of the theoretical framework, the overview of the
methodology, and significance of the study. Both authors read, reviewed, and gave feedback on
multiple drafts of this chapter. Rick completed the final bibliography and Emily made the final
round of edits to the dissertation to proofread and incorporate all revisions.
Chapter Two: Literature Review
Chapter 2 was jointly conceptualized and executed by Emily Chung and Rick Garcia with
an outline of the chapter developed together. Rick wrote the initial draft of effects of student debt
on the economy and rational choice and behavioral economics. Emily wrote the initial draft of
the overview of the significance of the study, human capital theory, and factors in graduate
student enrollment. In response to comments from the committee after the dissertation proposal
defense, Emily added an initial section on behaviors and Rick added initial sections on attitudes
and understanding good debt and bad debt. Emily also conceptualized and created the model of
attitudes and student borrowing, which was further refined by Rick. Both authors read, reviewed,
and gave feedback on multiple drafts of this chapter.
GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 140
Chapter Three: Methodology
Chapter 3 was jointly conceptualized and executed by Emily Chung and Rick Garcia. The
team had agreed from the beginning that the research project would be quantitative and use SPSS
for data analysis. Rick created the outline for the chapter and wrote the initial draft of the
definition of the research approach, instrumentation, and data collection. Emily wrote the initial
draft of the introduction to the chapter, site and sample, and data analysis. Both authors read,
reviewed, and gave feedback on multiple drafts of this chapter.
Chapter Four: Data and Findings
Chapter 4 was jointly conceptualized and executed by Emily Chung and Rick Garcia with
an outline of the chapter developed together. After collaborating to launch the full survey with
master’s students at three schools at WUU, both authors ran multiple analyses on SPSS to
interpret the findings and write this chapter. Rick uploaded the coded data from Qualtrics into
SPSS to create the initial data set. After jointly discovering minor errors in reverse coding, Emily
cleaned up the data set with the reverse coded items and created dummy variables for the
categorical variables and scale analysis. Rick wrote the initial draft of the sections on the
descriptive statistics, comparing master’s students at the school to national trends, and survey
and scale reliability. Emily ran the multiple regressions and wrote the initial draft of the sections
on answering research questions one and two. In response to the committee’s comments during
our dissertation defense, chapter four was revised to exclude the bulk of pilot survey data (which
was moved to an appendix). The pilot survey was analyzed and refined together, but the section
on this data was largely written by Rick. Both authors read, reviewed, and gave feedback on
multiple drafts of this chapter.

GRADUATE STUDENTS’ ATTITUDES TOWARDS DEBT 141

Chapter Five: Discussion
Chapter 5 was jointly conceptualized and executed by Emily Chung and Rick Garcia.
Emily created the outline for the chapter and wrote the initial draft of the purpose of the study
and research questions and the discussion of the findings. Rick wrote the initial draft of the
limitations, implications for practice, future research, and conclusion. Emily revised this chapter
extensively in light of the committee’s comments from the dissertation defense. However, both
authors read, reviewed, and gave feedback on multiple drafts of this chapter.

Abstract (if available)

Abstract This dissertation contributes to a small but growing body of research on graduate student loan debt. We designed a survey to measure whether debt attitudes were related to the amounts master’s students at a large, urban, research university borrowed to finance a year of graduate study. Overall, we found that debt attitudes are related to the amounts borrowed by the respondents in our study. Our theoretical framework of rational choice, human capital, and behavioral economic theories revealed that respondents’ borrowing decisions are influenced by considerations of cost‐benefit analyses, long‐term returns on investment, and irrational aversions to debt. The implications of this research are that that institutional revenue streams need to rely less upon master’s students deficit financing their degree programs, financial aid and admissions professionals should proactively outreach to prospective borrowers, and that greater transparency is needed on the topic of graduate student borrowing and career outcomes.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Do attitudes matter? attitudes towards debt and graduate student loan debt