An economic analysis of college athlete graduation rates: a closer look at title IX

A.J. Palumbo School of Business Administration

Duquesne University

Pittsburgh, Pennsylvania

AN ECONOMIC ANALYSIS OF COLLEGE ATHLETE GRADUATION RATES: A CLOSER LOOK AT TITLE IX

Mark D. Troyan

Submitted to the Economics Faculty

in partial fulfillment of the requirements for the degree of

Bachelor of Science in Business Administration in Economics

December 2008

Faculty Advisor Signature Page

Pavel Yakovlev, Ph.D. Date

Assistant Professor of Economics

AN ECONOMIC ANALYSIS OF COLLEGE ATHLETE GRADUATION RATES: A CLOSER LOOK AT TITLE IX

Mark D. Troyan, BSBA

1. 
   Introduction ………………………………………………………………………....5
   

2. 
   Literature Review …………………………………………………………………...8
   
3. 
   Empirical Analysis …………………………………………….................................14
   

1. 
   Data …………………………………………………………………………..14
   

1. 
   Empirical Model ……………………………………………………………...19
   

4. 
   Results Analysis ……………………………………………………………………..21
   

5. 
   Economic Implications……………………………………………………………….29
   
6. 
   Suggestions for Future Research ………………………………...............................31
   
7. 
   Conclusion ………………………………………………………................................32
   
8. 
   References ………………………………………………………………...………….33
   

Earning a college degree is of critical importance for obtaining a position of high earning power. A high proportion of students, ranging from 30 to 60 percent do not earn a degree (Eckland 1965). Several bodies of literature claim the graduation rates of students and student-athletes alike are affected by their vastly different college experiences. One factor that makes the college experience different for a student-athlete is their devotion of time to athletics. The time student athletes devote to their sport is physically, as well as mentally demanding, making it different than most of the extracurricular activities of other students. Along with these demands, Division-IA (DI-A) athletes face immense pressure to play at high levels and win at all costs. These entities are amplified depending on the division, tier and reputation of the school, but all seem to carry a negative effect on the graduation of a student. Some studies cited the exact opposite, however, and found student-athletes graduate at a higher rate than the overall student body. For example, over the 2001 – 2007 timeframe the average graduation rate of Division-IA (DI-A) and Division-I-AA (DI-AA) athletes was 62%, whereas the overall student body graduated at an average of 58%. Over the same period of time, Division-II (D-II) athletes showed an average graduation rate of 53%, whereas the overall student body graduated at an average of 45%. Scholars such as Stafford (2004) and Sigelman and Wahlbeck (1999) refer to Title IX compliance as a legitimate cause of the graduation rate discrepancies because female athletes graduate at higher rates than male athletes. “The greatest gender inequality in terms of participation currently occurs in the smaller NCAA divisions who have received much less press coverage.”¹

In 1972, the Title IX legislation was enacted and stated, “No person in the United States shall, on the basis of sex, be excluded from participating in, be denied the benefits of or be subjected to discrimination under any education program or activity receiving Federal financial assistance.” ² Although Title IX includes all aspects of education, the application of Title IX to college athletics has been complicated because athletic programs are sex segregated by sport. In order to meet the standards of legislation many institutions have retracted male programs and/or added female programs. The standards that each institution must comply with are as follows:

“Prong one - Providing athletic opportunities that are substantially proportionate to the student enrollment, OR

An example of compliance is as follows: If a school has 50% female enrollment and 50% male enrollment, then the optimal level of compliance with Title IX signifies that the school has 50% male participation and 50% female participation in sports programs. Even though many schools may adequately comply with the legislation, each institution complies at a different rate. Many NCAA schools show a bias towards male athletics. Figure 1 illustrates the average athletic participation gap over a 25 year period and further evidence of the bias towards male athletics.

The purpose of this paper is to study the variations in graduation rates across divisions of athletics and tiers of academia. This research differs from already publicized articles because I incorporate variables that are not only significant to student-athletes, but to the overall student body as well. This study examines athletic graduation rates by using divisions of athletics and tiers of academia as dummy variables. Other variables of interest include private / public, athletic success, and race broken up into individual categories (Black, White, Asian and Hispanic). Moreover, I construct a different measure of Title IX compliance compared to past studies by using the difference between male and female proportions in the student body and the proportion of male and female student-athletes as a proxy. I expect to find, ceteris paribus, that an increase in the compliance rate at an institution will lead to an increase in the graduation rate overall.

There has been a significant body of work conducted on the graduation rates of students and student-athletes. Many studies look at generally accepted variables such as sex, academic ability, and socioeconomic background to conclude their findings. A different view supports the findings of Sewell and Wegner (1970). This study looks to conclude that the institutional characteristics of a school have an influence on the probability that students continue their education by focusing on a sample of 1,253 male students. This study used a stepwise multiple regression analysis, which enabled multiple regressions to be run simultaneously. These results hold that the variables such as academic ability and socioeconomic background are the most important indicator of graduation rates, but the type of institution attended found significant variance that is not accounted for by student characteristics (Sewell and Wegner 1970).

Maloney and McCormick’s (1987) study is one of the most publicized articles regarding classroom success of athletes in college, accounting for student characteristics. The authors used data from all students enrolled at Clemson University in 1988-1989, which were approximately 12,000 students. Maloney and McCormick used an OLS regression model to determine the fluctuations in the graduation rates of student-athletes. The authors also studied the in-season effect of the student-athlete by using an OLS regression model. They made a dummy variable for the time spent in each sport during the season. They concluded that non-revenue sports did not show any grading differential during the season. In contrast, sports that brought in a lot of revenue for the school showed that the athletes had lower grades during the season than outside of the season. Maloney and McCormick (1987) also conclude that college athletes do not perform as well in the classroom as their non-athletic peers and many student-athletes do worse in college even when accounting for their academic background.

Tucker (1992) found that big time⁵ collegiate football programs may attract a higher quality student to the school, but there are costs that come to the student. The author finds that a university with a football program that is consistently nationally ranked has a lower overall graduation rate than their competitors with unsuccessful football programs. He used the final AP poll of year to get his data. He measured the success of the football program by giving points to an institution for each first place finish during the five-year period 1984 – 1988. He used a GLS regression model to examine the multiple institutions. Tucker confirms the results of McCormick and Tinsley (1987). Big time college football success has significant negative impact on the graduation rate of the entire student body. In contrast, there is evidence that a big-time basketball program does not have an effect on the graduation rate of an institution.

Adams, Bean, and Mangold (2003) used data from editions of the US News Best Colleges in America and the US Department of Education’s Integrated Post Secondary Education System Data to examine 97 out of the 112 Division-IA universities that have both basketball and football programs. They regressed variables that were considered the strongest indicators of graduation rates from other studies. Some of these variables were the student’s class rank in high school, ACT/SAT scores, size of the institution, and sport. Researchers (Blank, DeBuhr, and Martin, 1983; Caldas and Bankston, 1997; and Gilmore, 1990) found that there is a strong correlation between student ability entering the institution as a freshmen and the graduation rate of the students. This study used an OLS regression to examine the effects of student ability variables on the graduation rates. The authors found all of the variables to be significant either by way of positivity or negativity, but the variable characterized as living on campus had the largest positive effect on graduation rates. The positive relationship between graduation rates and size of the school was also found by (Tucker 1992). The findings of these authors prove that social communities and institutional characteristics are a bridge to learning communities and by strengthening these attributes you can promote institutional goals (Adams, Bean, and Mangold 2003).

Gaston-Gayles (2004) used a different approach to conclude the academic outcome of a student. This study examined the utility of academic and athletic motivation as a key variable in predicting academic performance. He measured 211 college athletes at a Division I institution in the Midwest part of the United States where 33% of the samples were female and 67% of the samples were male. He used non-cognitive variables and regressed these variables using a forward stepwise regression. This regression analysis confirmed that ACT scores, academic motivation and ethnicity were significant in the model (Gaston-Gayles 2004).

Other studies seek to satisfy the notion that student characteristics are the most influential factor of academic completion. Hamagami and McArdle (1994) found that by using a logit model and a sample of more than 3,000 athletes at 68 Division I schools both high school grades and SAT/ACT scores would be good predictors of college graduation of student-athletes. In order to conclude demographic variables, they held all other variables constant with regard to the sought out variable. They found that females obtained higher graduation rates. This modest sized-effect may turn out to be important in further studies on issues of gender inequality (Porter 1990), for example, Title IX compliance.

Purdy, Eitzen, and Hufnagel (1982) studied the academic achievements of athletes at Colorado State University from 1970 to 1980. The study examined how athletes compare to the general student population measuring academic ability and achievement. While most of the literature has measured similar variables this study shows the graduation rate of the institution with respect to the entire student body, not just student-athletes. This study finds that female athletes differ from male athletes in college achievement levels, reiterating the findings of Hamagami and McArdle (1994). In addition, the authors also found significant information regarding race. The study shows black athletes had lower scores on the entire range of educational achievement, which led to lower graduation rates overall (Purdy, Eitzen and Hufnagel 1982).

Two studies from the Journal of Blacks in Higher Education help to prove that blacks have the lowest graduation rates among all races. Neither study, however, uses regression analysis. The authors use data from 1993 and 2002. They studied each Division I university regardless of the tier of academia by looking at the difference among graduation rates between whites and blacks. In 1993, they concluded that at more academic rigorous universities black students’ graduate at much higher rates than lower level institutions. The second study uses the same approach but looks at highly ranked universities with regard to student-athletes. This study confirmed that at 11 of 13 highly rigorous universities in the study showed black athletes’ graduated at lower rates than the rate of the overall black student body. In each of the studies black students regardless of the university they attended had lower graduation rates than white students overall.

Although substantial proportionality has never been explicitly defined, in practice it is achieved when the percentage of female athletes is within five points of the percentage of female undergraduates (Stafford 2004). Title IX essentially requires that post-secondary institutions provide students regardless of gender the access to participate in interscholastic sports. Only a minimal amount of studies, to my knowledge, have examined the determinants of compliance in a regression framework. The first study, Stafford (2004), characterizes compliance as a binary variable and therefore uses two probit models. Each of these regression models included participation, enrollment, scholarship data and different institutional characteristics. One model examined proportionality and the other examined scholarships. This was not the only regression used in this study, however. In order to conclude what factors lead toward the progression of compliance, the author used an OLS regression for the years 1995 – 2001. Comparing the results of the OLS regression and the probit regressions, the study found that large institutions and institutions with lower percentages of female undergraduates were more likely to be in compliance with Title IX legislation (Stafford 2004). The author confirmed that there was approximately a 25 percent increase in compliance among institutions throughout the 1995 – 2001 time periods.

Sigelman and Wahlbeck (1999) analyzed data on the gender composition of 304 Division I athletic programs and student bodies to determine what each school needs to accomplish Title IX compliance. In order to be considered within compliance the institution must be within five percentage points of proportionality, reiterating the claims of (Stafford 2004). The authors used data from published reports of the NCAA’s 1996 survey of member institutions. They proceeded to make tables with the data corresponding to each compliance scenario. They found that many schools were not in compliance. Unlike most literature, Sigelman and Wahlbeck (1999) added the Big Ten’s 60 – 40 intermediate scenario into their study. In 1992, the Big Ten Conference approved a resolution promising in five years that the total number of athletes at each institution would be 60 percent men and 40 percent women. Then by 2002 each school would have complete proportionality. The authors found that at most schools, especially those with football teams, that compliance was nowhere near the appropriate level. They also confirmed that because of the large number of athletes playing football, the Big Ten was unable to achieve neither benchmark of compliance.

Another study by Anderson, Cheslock, and Ehrenberg (2006) it was found that there are distinct factors leading to the Title IX compliance of an institution. They used data from the Equity in Athletics Disclosure Act, IPEDS, and NACUBO for almost 700 institutions in Division-IA (DI-A), Division-II (D-II), and Division-III (D-III). By using a cross-sectional regression model, they were able to determine significant variables for the years of 1996 – 2002. Their regression controlled for division, conference and whether the institution had a football team because these variables may be endogenous (Anderson, Cheslock, and Ehrenberg 2006). They found that the greatest compliance and the most improvement in compliance were seen within DI-A institutions that are able to compete in the Bowl Championship Series (BCS). The BCS are reserved for teams that are able to play for the national championship and considered to be big-time D-IA football programs. The authors also confirmed that private institutions have significantly larger proportionality gaps among them then do public schools. This could be due to universities using athletics to increase their male enrollment levels and their tuition dollars.

The study I am conducting implements division, tier, and Title IX variables. Also, the study controls for variables mentioned in the literature review. The following sections of the paper will describe the data and the set up of the empirical model.

3. Empirical Analysis

3.1 Data

Each year the NCAA website issues graduation rates of student-athletes with regard to gender, race, and an assortment of individual sports. I collected a majority of my data from this website, but also collected data from the US Department of Education’s Integrated Post Secondary Education System Data. Data for this model is panel data on 63 institutions for seven years from 2001 - 2007. There is a sample size of eight institutions for each represented division throughout each tier of academia. The divisions of athletics are characterized as DI-A, DI-AA, and D-II. The difference between these divisions is the amount of time a student-athlete must obligate to the sport in which they play and their overall level of competition. For example, DI-A athletic programs may have practice and other athletic related functions for 25 hours a week, whereas D-II may only confer for 15 hours a week. Also, DI-A schools are noted for having the most elite athletes in the country, whereas D-II and to a lesser extent DI-AA athletes can be categorized as lower level athletes or athletes with less potential, based on size, strength, and ability. The only difference between DI-A and DI-AA athletics is within the football programs. DI-A programs brings in much more revenue for the school, has more media coverage, and is able to compete in the Bowl Championship Series (BCS). The BCS is considered all the major collegiate football games throughout DI-A, for example, The National Championship and The Rose Bowl.

The ranking or tier of a school is important in the attainment of higher quality students. For example, Duquesne University recently obtained a Tier 1 ranking: Coincidently the admission standards for Duquesne University have continually been on the rise. The tier of academia an institution achieves is computed by using seven variables. These variables and there weighted percentages are peer assessment (25%), retention (20%), faculty resources (20%), student selectivity (15%), financial resources (10%), graduation rate performance (5%), and alumni giving rate (5%). This ranking is computed each year by the US News and World Report. As you can see, by increasing the graduation rate of student-athletes at a particular school, ceteris paribus, the ranking of the school increases simultaneously.

The list of schools may seem incomplete because it does not contain any Ivy League or D-III institutions. School’s such as Harvard, Princeton, Yale and other D-III athletic institutions were exempt from the study because they do not give athletic financial aid. Only schools that give some form of athletic scholarships were used in the study because students not receiving athletic financial aid could show a self-selection bias. A self-selection bias could occur because higher performing students are choosing to go to school without athletic scholarships in order to excel in the classroom. Therefore, the graduation rates at these institutions should be much higher than institutions that give out athletic aid. Having a free or partially free education with an athletic scholarship could, in fact, deter the student-athlete from studying or from reaching their potential in the classroom. Because of this, I am only trying to conclude what factors go into the graduation rates of student-athletes who are not motivated solely by their own invested financial obligations. Students that are paying their own money and not receiving athletic scholarships could value their education more because they have more invested into their education; this is known as a self-selection bias. The table below lists the institutions that are used in my study. I chose the institutions in the study at random, with no prior data on any of the schools.