Effect of Personal Financial Knowledge on College Students' Credit Card Behavior

Effect of Personal Financial Knowledge on College
Students’ Credit Card Behavior
Cliff A. Robb and Deanna L. Sharpe
Analysis of survey data collected from 6,520 students at a large Midwestern University affirmed that financial
knowledge is a significant factor in the credit card decisions of college students but not entirely in expected ways.
Results of a double hurdle analysis indicated that students with relatively higher levels of financial knowledge
were not significantly different from students with relatively lower levels in terms of the probability of having a
credit card balance. Contrary to expectations, those with higher levels of financial knowledge had significantly
higher credit card balances. Overall, the present findings highlight the complex nature of the relationship between
personal financial knowledge and credit card behavior.
Key Words: college students, credit card use, personal financial knowledge
Introduction
Rather, credit card companies were viewed as enticing in-
In the late 1980s, credit card companies began targeting
experienced and unsuspecting students to sign agreements
college students in an effort to expand market share.
that they did not fully understand, placing them at risk of
Students were encouraged to become credit card customers
overspending and developing financial difficulties. As a
through direct mail promotions, on- and off-campus adver-
result, concerned groups encouraged university and col-
tising, and on-campus recruitment (O’Connell, 1994; Suss-
lege campuses to limit the access that credit card vendors
wein, 1995). A number of researchers have documented
had to their student population (Brobeck, 1992; Davies &
the subsequent rapid expansion of credit card ownership
Lea, 1995).
and use on college campuses from the late 1980s through
the 1990s (Kara, Kaynak, & Kucukemiroglu, 1994; Nellie
Recent research findings suggest that college students
Mae, 2002; Manning & Kirshak, 2005). In 1990, slightly
may not be at risk to the extent initially feared, however.
over half (54%) of all undergraduate students held at least
Although some students do have difficulty with credit,
one credit card. By 2001, over three-quarters (83%) of all
in general, college students are at least as responsible as
undergraduate students had one or more credit cards (Nel-
their age peers in managing credit card use and credit
lie Mae, 2002). These fundamental changes in how and
card debt (Braunsberger, Lucas, & Roach, 2004; Draut &
to whom credit cards are marketed have resulted in credit
Silva, 2004). Commensurate with their low earnings and
cards becoming a way of life for today’s college student
financial inexperience, card limits and balances are rela-
(Lyons, 2004; Manning & Kirshak, 2005).
tively low among college students, usually averaging a few
thousand dollars (U.S. General Accounting Office, 2001).
As the percentage of college students with credit cards
After reviewing several studies of credit card debt levels of
grew, the concern that credit card companies were taking
college students, Lyons (2004) concluded that the majority
unfair advantage of a vulnerable population also increased.
of college students are not amassing excessive debt, and
In essence, the credit market among college students was
over one half of college-aged credit card holders pay their
considered imperfectly competitive. The signed credit
balance in full each month.
contract was not seen as an agreement between equals.
Cliff A. Robb, Ph.D., Assistant Professor, Department of Consumer Sciences, University of Alabama, 304 Adams Hall, Tuscaloosa, AL 35487,

crobb@ches.ua.edu, (205) 348-1867
Deanna L. Sharpe, Ph.D., CFP®, Associate Professor, Personal Financial Planning Department, University of Missouri, 239 Stanley Hall, Columbia, MO

65211, sharped@missouri.edu, (573) 882-9652
© 2009 Association for Financial Counseling and Planning Education®. All rights of reproduction in any form reserved.
25

---

Questions remain, however, as to what factors enable col-
Parental income is a key indicator of a student’s accus-
lege students to manage credit card use despite their rela-
tomed lifestyle, social class, resources and opportunity
tive inexperience in the credit market. Economic theory
to learn about management of money. According to one
proposes that consumers require knowledge to make util-
report, about 14% of students came from families with an
ity maximizing choices. The purpose of this study was to
annual household income under $50,000 (Draut & Silva,
examine the role that knowledge of personal finance con-
2004). Draut & Silva (2004) found that students from
cepts and principles may play in college students’ decision
lower income households were more likely to develop
to revolve a credit card balance and in the level of balance
relatively high credit card balances ($7,000 or more) as
revolved. In this study, credit card revolvers were defined
compared with their peers. These findings suggest that
as respondents who did not pay their credit card balance
perhaps such students did not have as much experience in
in full at the end of the month. Study findings can broaden
financial markets as their peers from middle- and high-in-
the understanding of factors influencing student credit card
come families.
use and may be useful for consumer educators and policy
makers that are interested in helping college students learn
Several studies have linked attitudes toward credit with
how to manage credit effectively.
credit behavior in several studies. Higher affective credit

attitude scores (using measures such as “my credit card
Review of Literature
makes me feel happy,” or “I like using my credit card”)
Credit Card Usage Among College Students
have been associated with students carrying an outstanding
Demographic Trends:
balance on multiple cards (Hayhoe, Leach, Turner, Bruin,
A number of recent studies have examined the character-
& Lawrence, 2000). Similarly, Xiao, Noring, and Ander-
istics, attitudes, and behaviors of college students who use
son (1995) and Joo, Grable, and Bagwell (2003) found that
credit cards (see Lawrence, Christofferson, Nester, Moser,
a positive attitude toward credit cards was associated with
Tucker, & Lyons, 2003 as well as Lyons, 2004 for a re-
card ownership and use. Chien and DeVaney (2001) noted
view of this research). These studies reveal some general
a positive connection between attitudes towards credit
trends about college students and credit card use. Gender
and the likelihood of carrying a balance. After examin-
differences in credit card use exist. Female students were
ing credit card attitudes among undergraduates in Britain
more likely than male students to have a credit card (Arm-
and America, Yang, James, and Lester (2005) concluded
strong & Craven, 1993; Lawrence, et al., 2003). Typically,
that affective and behavioral attitude scores were the
they also held more debt than male students (Micomo-
strongest predictors of the number of credit cards owned.
naco, 2003). In prior research, females have tended to
Interestingly, they noted that those who had more positive
display lower scores than males on measures of personal
attitudes toward money in general also exhibited greater
financial knowledge (Chen & Volpe, 1998, 2002; Jones,
obsession with money.
2005; Lusardi & Mitchell, 2005; Borden, Lee, Serido, &
Collins, 2008).
Hayhoe, Leach, and Turner (1999) developed a scale
measure of money attitudes using survey participant’s
Evidence suggests that there is little difference in terms of
responses to statements about feelings, knowledge, and
credit card ownership based on college students’ ethnic-
behavior related to credit cards and debt. Evaluating the re-
ity. In a study of Louisiana college students, Lawrence et
lationship between this measure and college student credit
al. (2003) noted that 45% of card holders were Caucasian,
card behavior, they found that students’ scores regarding
23% were African American, 19% were Asian, about 6%
money attitudes of obsession and retention and affective
were Hispanic and the remainder were Native American
credit attitudes distinguished between the students who did
or other race and ethnicities. These percentages reflected
and did not have credit cards (Hayhoe et al., 1999). Attitu-
the race and ethnic distribution in the overall student body,
dinal scores also distinguished between students who had
suggesting that ethnicity was not a factor in distribution
less than three credit cards and those with four or more and
of credit card holders on that campus. There is some prior
were significant predictors of who, among students with
research, however, that indicated that minority students are
cards, would carry four or more credit cards.
more likely to be financially at risk when compared with
other students (Lyons, 2004).
There seems to be some “class rank” effects in credit card
behavior. A study by Nellie Mae (2002), a nonprofit stu-
26
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009

---

dent loan provider, found slightly more than half of fresh-
administrators who were contacted as part of a study on
men (54%) had a credit card. Freshmen also had the lowest
student credit card use commissioned by the General Ac-
average number of cards (2.5) and average debt ($1,533).
counting Office have acknowledged that there may be a
The numbers of cardholders rose with class rank, however.
relationship between various financial concerns, includ-
Ninety-two percent of sophomores had a card; for seniors
ing mishandling of credit, and persistence to graduation
the percentage was ninety-six. From sophomore to senior
(GAO, 2001).
year, the average number of cards held was successively
larger, changing from 3.67 to 4.50 to 6.13, respectively.
Financial Knowledge
Average debt levels were larger for higher levels of class
There are two lines of research on financial knowledge.
rank as well. The average debt of seniors ($3,262) was
In one group of studies, participants answered questions
more than double that of freshmen (Nellie Mae, 2002).
related to general financial knowledge (Markovich & DeV-
aney, 1997; Chen & Volpe, 1998; Avard, Manton, English,
Similar to seniors, 96% of graduate students reported own-
& Walker, 2005; Jones, 2005). The questions used in these
ing at least one card; on average, they had 6 credit cards. At
studies related closely to the topics typically covered in an
$7,831 per student in 2003, the average credit card debt of
introductory personal finance course. The second group of
graduate students was much higher than that of undergrad-
studies used specific financial knowledge as a proxy for
uates. The average level of credit card debt among graduate
financial literacy (Warwick & Mansfield, 2000; Joo et al.,
students in 2003 was almost $3,000 higher than reported
2003; Braunsberger et al., 2004). These studies generally
in 1998 (Nellie Mae, 2007). One in four graduate students
asked individuals to report particular facts about their own
with credit card debt in 2003 had balances between $6,000
credit cards (e.g. APR, fees, etc.). There is strong evidence
and $15,000, about the same proportion observed in 1998.
from both lines of research that suggests, regardless of
Fifteen percent had a balance over $15,000, over twice the
how financial knowledge is operationalized, college stu-
proportion seen in 1998 (Nellie Mae, 2007).
dents do not possess a high degree of financial knowledge
(Markovich & DeVaney, 1997; Chen & Volpe, 1998; War-
Lyons (2004) analyzed responses from a random sample of
wick & Mansfield, 2000; Avard et al., 2005; Jones, 2005).
University of Illinois undergraduate and graduate students
who had completed a survey related to financial issues in
Chen and Volpe (1998) administered a 36 question survey
2001 to determine the probability of being at risk of credit
dealing with various aspects of personal financial knowl-
misuse or mismanagement as measured by four specific
edge to college students. The average score of correct re-
outcomes or behaviors: having $1,000 or more in outstand-
sponses was close to 53%, not a passing score on a typical
ing credit card balances, being late on payments by two
grading scale. They noted significant degree-type and class
months or more, having reached the limit on credit cards,
rank effects. Business majors tended to score better than
and rarely or never paying off credit card balances. Lyons
non-business majors. Students with more years of college
(2004) concluded that financially at-risk students were
had higher financial knowledge scores than students with
more likely than other students to receive need-based fi-
fewer years of college. Other researchers have also found
nancial aid, have $1,000 or more in other outstanding debt,
that college freshmen have low scores on tests of financial
or to have acquired their card by mail, at a retail store or as
knowledge. Avard et al., (2005) found that college fresh-
the result of a campus solicitation.
men were able to answer only about 35% of financial
knowledge questions correctly. Using a six-question scale
Graduating students leave college and university campuses
of credit knowledge to evaluate financial knowledge, Jones
with an average debt burden of $20,402 for education and
(2005) reported that, on average, incoming freshmen gave
credit card debt combined (Nellie Mae, 2002). Financial
correct answers only 56% of the time.
experts have expressed concern that credit card debt cou-
pled with student loan debt could create serious financial
Among existing studies, the ability of a cardholder to
burdens for college students near and post graduation
report his or her annual percentage rate (APR) is one of
(Bianco & Bosco, 2002; College Board, 2005). These
the most commonly used measures of specific financial
concerns have only intensified in recent years as rising tui-
knowledge. The federal law mandating reporting of the
tion costs have consistently outpaced increases in financial
APR was passed in 1968. Since that time, awareness of
aid available per student (College Board, 2005). University
APRs has grown considerably (Durkin, 2000; Hogarth
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009
27

---

& Hilgert, 2002). Ironically, despite increased awareness
compared with students who had lower scores on the test.
of this measure, research indicates that few consumers
Research among secondary school students has suggested
seem to understand how to use the APR to make effective
that financial education has a positive effect on financial
financial decisions (Lee & Hogarth, 1999). Similar results
competency (Langrehr, 1979; Tennyson & Nguyen, 2001).
have been found for the college population, as Chen and
Among the general population in the United States, strong
Volpe (1998) discovered that 67% of the college students
correlations have been found between a person’s composite
surveyed could not correctly answer a multiple-choice
score of financial knowledge and an index of credit man-
question regarding the APR.
agement behaviors (Hilgert, Hogarth, & Beverly, 2003).
Liebermann and Flint-Goor (1996) suggested that prior
Not all researchers would concur, however, that there is a
knowledge of an issue is one of the most important factors
significant link between financial knowledge and behavior.
influencing information processing. Evidence regarding
Using a six-question scale to measure financial knowledge,
the relationship between financial knowledge and financial
Jones (2005) did not find a significant relationship between
behavior has been mixed, however. Results vary depend-
knowledge and college student credit card debt behavior.
ing on how financial knowledge has been measured, what
Similarly, in research by Borden et al. (2008), no signifi-
behaviors have been studied, and what populations have
cant relationship was found between financial knowledge
been analyzed (Mandell, 2004; Peng, Bartholomae, Fox, &
and effective or risky financial behaviors.
Cravener, 2007).
In summary, inconsistencies from the available literature
Findings of some studies suggest that life-cycle stage
make it difficult to draw strong conclusions regarding the
may influence the perceived salience of personal financial
relationship between financial knowledge and behavior.
instruction. Mandell (2004) noted that having a savings
The present analysis utilized components of previous
account has been associated with higher savings knowl-
knowledge measures and built on the previous research
edge among high school students. Ironically, however,
by directly comparing a theoretical measure of personal
using credit cards has been associated with lower credit
financial knowledge to an observable financial behavior.
knowledge among this age group. Peng et al. (2007) noted
that both high school and college students that completed
Theoretical Framework
a personal finance course displayed improved savings rates
According to the life-cycle hypothesis, individuals strive
following a personal finance course. But, participation in
to have a constant consumption path through life (See
a college level personal finance course was also associated
Ando & Modigliani (1963) for a formal discussion of this
with improved investment knowledge, an effect not noted
hypothesis). In youth and old age when income tends to be
among high school participants. In the workplace, there
limited, dissaving occurs. Saving occurs in midlife when
is evidence that targeted instruction, such as retirement
income is relatively higher. In this context, college stu-
planning education, has a significant influence on financial
dent use of debt instruments, including credit cards, could
behavior (Todd, 2002; Bernheim & Garrett, 2003).
be considered a rational decision given their significant-
ly higher expected earnings path as compared with high
On the basis of their research, Chen and Volpe (1998) ar-
school graduates (Baum & Payea, 2004; Kidwell & Tur-
gued that a person’s level of financial knowledge tends to
risi, 2000; Norvilitis & Santa Maria, 2002).
influence their opinions and affect their financial decisions.
Their study was among the first to establish a link, albeit
Although the life-cycle income hypothesis suggests why
a tenuous one, between knowledge and behavior among
borrowing can be a rational decision, it does not specify
college students. Individuals with higher levels of finan-
the means by which borrowing might occur. Borrow-
cial knowledge were more likely to make good financial
ing can be envisioned as a two-step process. The decision
decisions in a hypothetical situation (Chen & Volpe, 1998).
of whether or not to borrow is the first step in that proc-
Focus group data analyzed by Cude, Lawrence, Lyons,
ess. Once a decision is made to borrow, the next step is to
Metzger, LeJeune, Marks, and Machtmes (2006) suggested
decide how much to borrow, taking the cost of borrowing
that students who scored higher on a financial fitness test
into consideration. Assuming a rational decision-making
were more likely to report paying their balance in full
process, individuals seek to equate the marginal costs with
each month and were less likely to own a credit card as
the marginal benefits of any given decision.
28
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009

---

College students may choose from a variety of debt instru-
sociated with a given decision to use credit, can they make
ments as a means of funding consumption during college.
rational decisions in the credit market?
The options include, but are not limited to, student loans,
bank loans, loans from family members, as well as credit
Economic decision-making theory underscores the impor-
cards. In terms of real costs (as measured by the interest
tance of product knowledge in making effective consumer
rate), credit cards are one of the most expensive borrowing
choices. This study extended prior research by analyzing
alternatives available if a balance is revolved (i.e., carried
the influence that general financial knowledge may have on
over from one month to the next, thus incurring an interest
credit decision-making and behavior. Specifically, this study
charge). So while it may be considered rational for college
examined the role that financial knowledge plays in college
students to utilize debt instruments, it could also be argued
students’ choices to have a credit card balance and in the
that credit cards are a relatively inefficient means of bor-
amount of balance held, controlling for factors that other
rowing given their high interest rates and required mini-
studies of college student credit card use have found to be
mum payments. Given the relative inefficiency of credit
influential. In this study, financial knowledge was defined as
cards, it would be expected that among college students
an individual’s understanding of important concepts related
who have a credit card, those who had greater knowledge
to personal finance, and was operationally defined in the
of the credit market would be less likely to carry a bal-
present analysis as a respondent’s score on six questions
ance on their cards and more likely to use other lower-cost
dealing with different aspects of personal finance.
forms of borrowing.
Theoretically, greater financial knowledge should enhance
Analysis of the borrowing decision is complicated by the
understanding of all costs associated with using credit
need to both have and understand credit market infor-
cards; whereas, a lack of knowledge of financial markets
mation. Difficulties can arise when this information is
and instruments makes it difficult to judge actual costs. In
complex, incomplete, or otherwise not sufficient for mak-
this study, it was hypothesized that:
ing effective market decisions. For example, student loans

H1: A higher level of financial knowledge is nega-
can offer college students a less costly form of borrowing

tively related to whether one carries a revolving
than credit cards. But, total costs of a student loan may be

balance.
rather difficult to ascertain as they occur in the future and
will depend to some extent on future payback behavior as

H2: Among those with a revolving balance,
well as type of loan obtained (e.g., subsidized versus un-

a higher level of financial knowledge is
subsidized). In addition, transaction costs for student loans

associated with a lower reported balance.
are relatively high. College students making a comparison
may conclude that carrying a credit card balance is cheaper
Method
Data and Sample
simply because the current and future cost of credit can be
An invitation to participate in an Internet based survey was
ascertained and the present transaction costs to use a credit
sent via electronic mail to a population of just over 25,000
card are relatively low, overlooking the fact that high inter-
undergraduate, graduate, and professional students at a
est rates can make credit cards the costlier option.
large Midwestern university in the United States. The sur-
vey consisted of 83 questions that gathered information on
Empirical findings suggest that there are significant benefits
credit card attainment and use, general demographic data,
to search in credit markets (Lee & Hogarth, 1998). What
consumer attitudes toward credit, online spending habits,
is not understood, however, is the extent to which college
and labor force participation. A drawing for three $150 gift
students understand these benefits of search. If college stu-
certificates was held as a participation incentive. A total of
dents lack knowledge of the operation of credit markets, it
6,520 students completed the survey, for a response rate of
is also likely that they would not fully understand the costs
roughly 24%. Once the data were cleaned, a usable sample
associated with borrowing via credit cards. Ausubel (1991)
of 3,884 college students was obtained. The drop in the
argued for consumer irrationality in the attainment of credit
number of cases was largely due to incomplete survey
cards, as individuals forgo extensive search based on the
responses. Distribution of demographic characteristics for
belief that the card will only be used as a convenience tool.
the reduced sample roughly mirrored that of the student
But, is this truly irrational behavior, or the result of a lack
population, except that the student sample had more fe-
of full market information? If college students lack key
male respondents than were present in the overall student
information to effectively weigh the costs and benefits as-
body (65.8% vs. 51.5%).
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009
29

---

Dependent Variables
the Money Attitude Scale (MAS) introduced by Yamauchi
The decision to revolve a balance is a two-step process.
and Templer (1982). Following the example of Roberts
First, students decide whether or not to revolve a bal-
and Jones (2001), Yamauchi and Templer’s (1982) time-
ance, which may be modeled as a simple yes-no decision.
retention dimension was not used in this study since the
Thus, the first dependent variable in this analysis was
sample was limited to college students. Student responses
dichotomous, set equal to 1 if revolve, 0 otherwise.Second,
to 20 separate questions about personal financial attitudes
respondents that choose to revolve a balance must decide
were factor analyzed. Similar to research by Roberts and
how much to revolve. Consequently, the second dependent
Jones (2001), and more recently Norum (2008), results of
variable was continuous and conditional on having a bal-
the factor analysis indicated four underlying factors ex-
ance. This two-step process is best modeled by a double-
isted: power-prestige (? = .87), distrust (? = .78), anxiety
hurdle approach (Cragg, 1971). Assumptions of ordinary
(? = .72), and second guess (? = .61). Scores for each of
least squares regression are violated in both equations. The
the attitudinal measures were reverse coded so that higher
dependent variable in the first equation is dichotomous
scores indicated stronger attitudes (e.g., a higher score on
rather than continuous. The dependent variable in the sec-
the power-prestige measure indicated an individual who
ond equation contains a substantial number of zero cases.
was more likely to view money as a source of power or
Tobit can be used in such situations, but this procedure
prestige). The results of the factor analysis are included
forces parameter signs to be the same at each stage. The
in Tables B.1-B.4 in the Appendix. The definitions and
double-hurdle model, in contrast, allows signs and signifi-
coding of the remaining independent variables used in this
cance of the independent variables to differ at each stage
study are outlined in Table C in the Appendix. These vari-
of the analysis.
ables were selected based on evidence of their importance
in previous research.
Similar to previous studies of college student credit card
usage (Reynolds, Hogarth, & Taylor, 2006), a large per-
Results
centage of the sample did not carry a balance (identified as
Descriptive statistics for the entire sample are reported
non-revolvers in this study), and observations among those
in Table 1. Roughly 66% of respondents had at least one
that did carry a balance were not normally distributed. For
credit card, with a reported average of about 1.4 cards per
this reason, the log of the credit card balance was utilized
respondent among cardholders. Respondents were asked to
in this study.
report only those credit cards that were used on a regular
basis. Thus, an individual who owned four credit cards
Independent Variables
but only used two of them on a regular basis should have
Financial knowledge was measured using six questions
reported using two credit cards. The present study was
dealing with general financial practices (see Table A in
concerned with credit card use rather than prevalence of
the Appendix for list of questions). Each question was
cards among college students.
designed to measure a different aspect of personal financial
knowledge. The six questions were drawn from the 2006
About a third (38%) of card holders reported having a
Jump$tart Survey and from research conducted by Chen &
revolving balance. Among those respondents that had and
Volpe (1998). Questions were selected with the intention
used credit cards, the average balance carried was $848.05.
of serving as a reflection of the issues that college stu-
Median balance carried was $0.00 due to the high percent-
dents might be faced with in a general course on personal
age of individuals who did not revolve a balance. Among
finance. The knowledge measure served as part of a larger
those who did revolve a balance, the average amount
survey analyzing numerous student financial behaviors
carried was $2,238.46, with a median amount owed of
and attitudes. Thus, an effort was made to keep the survey
$1,000.00. Average monthly spending on all cards was
at a reasonable length. Individuals’ scores ranged between
$298.93, with a median of $75.00. Slightly less than half
0 and 6, depending on the number of correct responses
(44.93%) of the sample was able to answer 4 or more of
that a participant provided. The total number of correct
the financial knowledge questions correctly. Mean re-
responses was summed to create the independent variable,
sponse on the financial knowledge questions was 3.14 on a
financial knowledge.
scale of 0 to 6.
The attitudinal variables - power, anxiety, second guess,
Average age of respondents was 21.29. A majority of the
and distrust - were constructed using a modified version of
sample was white (86.28%) and female (65.83%). A little
30
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009

---

Table 1. Descriptive Statistics for the Entire Sample, Card Holders and Non-card Holders (N = 3,884)
Variable
Frequency
Variable
Frequency
Own a credit card
65.99%
Financial aid
66.15%
Carry a balance
25.00%
Charge school items*
23.29%
Course in finance
28.73%
Independent
30.90%
Have other debt
18.00%
You pay on cards*
87.09%
Knowledge (# Correct)
Source of card*
0
4.97%
1
9.71%
Parents
17.17%
2
16.19%
Direct mail
23.89%
3
24.20%
Campus source
4.81%
4
27.01%
Bank
26.08%
5
16.66%
Store/Retail
14.47%
6
1.26%
Other
7.93%
White
86.28%
Employed
64.01%
Female
65.83%
Business Major
17.82%
Year in school
Parent’s Education
Freshman
19.77%
Sophomore
18.82%
Less than high school
0.70%
Junior
18.87%
High school
9.81%
Senior
20.96%
Some college
21.27%
Graduate?
21.58%
College or more
67.84%
Expected income
Parent’s income
Low
13.90%
Low
19.75%
Medium
59.32%
Medium
34.99%
High
21.81%
High
34.27%
Married
7.60%
Urbanization
Urban
12.62%
Suburban
60.92%
Rural
26.47%
Continuous Variables
Mean
SD
Average monthly spending
$197.26 ($75.00)
375.66
[$257.33]
[397.34]
Amount revolved
$559.62 ($0.00)
1990.88
[$2238.46]
[3479.19]
Number of cards used (max = 8)
0.92
1.01
Age (max = 30)
21.29
2.73
Knowledge (max = 6)
3.14
1.42
Note. * Indicates that N = 2,563 due to the fact that these variables were only applicable to those individuals holding credit
cards (1,321 individuals did not report holding credit cards).
? Graduate student category consists of professional, medical, and law students.
Median scores for the entire sample are presented in parentheses. Means and standard deviations for only those individuals
who reported revolving a balance are presented in brackets.
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009
31

---

over 7% of the sample were married. Slightly less than a
Interestingly, which variables were significant depended
third of respondents (30.90%) were financially independent.
on the stage of the analysis, and the sign of the effect was
Most had come from a suburban area (60.92%). Class rank
not always consistent across stages. Contrary to the initial
was rather evenly distributed across the sample with around
hypothesis, the financial knowledge measure was not
one-fifth in each class from freshman to graduate student.
significantly related to whether or not individuals reported
carrying a revolving balance. This result was largely sup-
More than a quarter of the sample (28.73%) had taken a
portive of findings presented by Jones (2005). However,
personal finance course. Eighteen percent of the sample
among revolvers (respondents who carry a credit card bal-
respondents were business majors. Roughly 14% of the
ance), increased knowledge was associated with carrying a
sample expected to earn less than $30,000 upon gradua-
larger log balance. This finding was contrary to the initial
tion, whereas a quarter of the sample (25%) expected to
hypothesis that more knowledgeable students would have
earn between $30,000 and $39,000 upon graduation. Over
lower log balances.
a third of the sample (34.78%) expected to earn between
$40,000 and $59,999, whereas the remaining respond-
Consistent with prior research, being financially independ-
ents expected to earn $60,000 or more. Roughly a third
ent was positively related to carrying a revolving balance,
had parents with incomes between $49,999 and $99,999
and was associated with higher log balances. Lyons (2004)
(34.99%) or $100,000 or more (34.27%). About two-thirds
noted a strong association between financial independence
of the respondents (67.84%) had parents that had earned a
and being financially at-risk (independent students were
college degree or higher.
more likely to be delinquent, have cards that were maxed-
out, and to not pay their balance in full).
A majority of the respondents received some form of
financial aid (66.15%) and were employed (64.01%).
Contrary to prior research, there were no significant differ-
Respondents obtained their credit cards from a variety of
ences in balance behavior based on gender in either stage
sources, with most having gotten them from either a bank
of the analysis. Previous research by Lyons (2004) sug-
(26.08%) or direct mail (23.89%) source. Commensurate
gested that females had a greater likelihood of being delin-
with previous research, on average, the credit card balance
quent on their cards as compared with males. As compared
carried by respondents was relatively low ($848.05). At
with other races, being white was associated with a lower
$298.93, average monthly spending was also relatively
likelihood of carrying a revolving balance. However, white
low. As suggested by the previous literature, the students
students were not found to be significantly different from
sampled appeared to be generally responsible in their
other races in terms of the amount revolved. This is some-
use of credit cards (Lyons, 2004). Roughly 62% of the
what supportive of the prior literature, which generally
respondents paid their cards in full each month, and 81%
suggested that minority students are more likely to engage
reported balance levels of between $0 and $1000. Roughly
in less responsible or riskier credit card behaviors (Allen &
9% of the sample reported holding a balance of $3,000 or
Jover, 1997; Monro & Hirt, 1998; Lyons, 2004).
more, however, suggesting that there were still many stu-

dents who could be considered as financially at-risk.
When compared with graduate students, juniors and sen-
iors were found to be more likely to revolve a balance,
The SAS® QLIM (Qualitative and Limited Dependent
though no differences were noted for freshmen or sopho-
Variable Model) procedure was used to conduct the dou-
mores relative to graduate students. Among revolvers,
ble-hurdle analysis. Results of this analysis are presented
graduate students and seniors had the highest debt levels;
in Table 2. In the first stage, factors affecting the decision
all other class ranks had relatively lower log balance lev-
to have a balance are modeled as a probit equation; the de-
els. These findings were consistent with those presented by
pendent variable was set equal to 1 if have a balance, 0 oth-
Nellie Mae (2002). Interestingly, income expectations did
erwise. In the second stage, maximum likelihood analysis
not have a significant influence at either stage of the analy-
was used to evaluate the influence of various factors on the
sis. Specifically, no statistical differences were noted based
level of balance carried among those with a balance. Mar-
on the annual income individuals expected to receive once
ginal effects associated with the variables for the first stage
they had completed college and begun to work full time.
of the double-hurdle analysis are presented in Table 3.
Those who received financial aid were more likely to carry
a revolving balance, though there were no significant dif-
ferences in terms of the log balance revolved based on
32
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009

---

Table 2. Results from the Double–Hurdle Analysis, Credit Card Balance as the Dependent Variable
Stage 1: Probit analysis
Stage 2: Maximum likelihood
(N = 2,368)
analysis (N = 894)
Parameter
Coefficient
St. error
Coefficient
St. error
Intercept
-0.523*
0.262
6.100***
0.519
Knowledge
0.007
0.024
0.119***
0.035
Female
0.058
0.068
-0.030
0.104
White (vs. Other)
-0.389***
0.086
-0.018
0.132
Year in school (Graduate student omitted)
Freshman
-0.138
0.122
-1.005***
0.204
Sophomore
0.153
0.106
-0.963***
0.162
Junior
0.286**
0.099
-0.409**
0.151
Senior
0.442***
0.089
-0.270
0.145
Financially independent
0.316***
0.076
0.298**
0.118
You pay on cards
0.414***
0.115
-0.070
0.237
Expected income (Middle income omitted)
Low expected income
-0.017
0.087
-0.010
0.125
High expected income
-0.089
0.072
0.014
0.109
Origin of cards
Bank
0.212**
0.071
0.376***
0.106
Campus
0.220*
0.112
0.490***
0.152
Parent
-0.144
0.079
0.201
0.130
Direct mail
0.283***
0.073
0.735***
0.114
Retail/Store
0.176**
0.070
0.191*
0.101
Other
0.105
0.094
0.349**
0.136
Financial aid
0.137*
0.065
0.167
0.105
Charge school items
0.236***
0.068
0.247*
0.102
Parent’s education (College or more omitted)
High school or less
0.221*
0.096
0.310*
0.137
Some college
0.251***
0.071
0.211*
0.109
Married (vs. single)
-0.017
0.101
0.451***
0.140
Course in personal finance
-0.040
0.064
-0.005
0.096
Business major
-0.173*
0.081
-0.117
0.127
Parent’s income (Middle income omitted)
Low income
0.027
0.077
-0.089
0.108
High income
0.041
0.068
0.083
0.104
Attitudes
Power
-0.004
0.007
-0.005
0.011
Anxiety
-0.068***
0.013
-0.011
0.022
Second guess
0.071***
0.015
0.044
0.025
Distrust
-0.029***
0.006
-0.036***
0.011
Employed
0.278***
0.066
0.114
0.119
Other debt
0.353***
0.070
0.286**
0.111
Time preference composite
-0.061
0.071
-0.151
0.098
Rho
0.091
0.226
–
–
Sigma
–
–
1.233***
0.033
*** p < .001. ** p < .01. * p < .05.
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009
33

---

Table 3: Marginal Effects for each of the Independent Variables on the Probability of
Revolving a Balance (Probit)
Parameter
Marginal Effect
Knowledge
0.002
Female
0.018
White (vs. other)
-0.123***
Year in school (Graduate student omitted) Freshman
-0.044
Sophomore
0.049
Junior
0.091**
Senior
0.140***
Financially independent
0.099***
You pay on cards
0.131***
Expected income (Middle income omitted)
Low expected income
-0.005
High expected income
-0.028
Origin of cards
Bank
0.067***
Campus
0.071*
Parent
-0.045
Direct mail
0.090***
Retail/Store
0.056**
Other
0.033
Financial aid
0.043**
Charge school items
0.075***
Parent’s education (College or more omitted)
High school or less
0.071*
Some college
0.081***
Married (vs. single)
-0.005
Course in personal finance
-0.013
Business major (vs. others)
-0.055*
Parent’s income (Middle income omitted)Low income
0.009
High income
0.013
Attitudes
Power
-0.001
Anxiety
-0.022***
Second Guess
0.022***
Distrust
-0.009***
Employed
0.088***
Other debt
0.112***
Time preference composite
-0.019
*** p < .001. ** p < .01. * p < .05.
34
Journal of Financial Counseling and Planning Volume 20, Issue 1 2009

---