Abstract
Higher education has been touted as critical to social mobility, even greater longevity; however, these benefits accrue with degree completion, and many students who begin college never earn a degree. Using the most recent Integrated Postsecondary Education Data System (IPEDS), we sought to understand variation in college completion rates and test hypotheses related to a set of institutional factors potentially associated with higher completion rates. Our findings suggest that greater attention to low-resource institutions is paramount if higher education is to fulfill the aspiration of promoting social mobility and equity. Several policy interventions could be used to combat ongoing disparities.
Introduction
Higher education has been touted as critical to social mobility, a great equalizer that can open doors for people leading them to employment, higher lifetime earnings, and even greater longevity. At the same time, concerns of the high cost of higher education abound, particularly in the wake of COVID-19 and the advent of the gig economy as employment markets are shifting, and the value of a college degree has come into question. In this turbulent environment, anomalies persist as the number of applicants to top-rated colleges and universities, particularly the Ivy League, multiply exponentially while less renowned institutions of higher education struggle to fill their seats. With these seismic shifts in the field of higher education, one has to ask: Is this industry still fulfilling the promise of enabling social and economic mobility? As young adults embark on the journey of attaining expensive college degrees, can they presume they will be better off than they would otherwise be?
Examination of college completion rates is key to understanding the potential contribution of institutions of higher education to social and economic mobility. Although the return on investment in college is well-established (Abel and Deitz, 2014; Krueger et al., 2019), these benefits accrue with degree completion. Yet, two decades of research have indicated that many students who begin college never complete college (U.S. Department of Education, 2021; Long, 2018).
Because non-completion has been a persistent problem, researchers and educators have examined risk factors for non-completion. Much of this work has focused on individuals’ attributes. At the individual level, factors that have been found to be associated with the likelihood of completion include a student’s financial resources, gender, race/ethnicity, family’s educational background, and the match between individual goals and capabilities and the institutional offerings (Tinto, 1975; Light and Strayer, 2000; Cabrera, Nora, and Castaneda, 1993; Astin, 1999; Titus, 2004). These landmark studies developed critical evidence about student characteristics and their association with college completion, with less examination of the institutions’ factors and student completion.
Some studies have tackled the question of institutional characteristics that are linked to completion rates; however, these studies have often focused on single institutions or a small sample of institutions (Carbrera, Nora, and Casteneda, 1993) or have been restricted to broad access institutions (BAIs) (Crisp, Doran, and Salis Reyes, 2017). Studies with national samples of institutions (Scott, Bailey, and Kienzi, 2006; Buchman and DiPrete, 2006; Melguizo, 2008) that have explored institutional factors related to completion rates are outdated and bear repeating.
In this study, we sought to use more recent data from a national sample of institutions of higher education to examine the variation in college completion rates and explore institutional factors associated with college completion rates. We hypothesized that several factors would be linked to graduation rates. These factors included 1) institutional financial resources, 2) commitments to faculty and academic programming, and 3) makeup of the student body.
First, we hypothesized that having greater institutional financial resources might be associated with higher completion rates, as financial resources may fund a variety of efforts to retain and support students. Thus, we examined ownership type and expected private compared with public/governmental institutions to have higher completion rates. We also measured endowment per student and expected institutions with higher endowments per student to have higher completion rates. Recognizing financial resource differences, we expected historically Black colleges and universities (HBCUs) to have lower completion rates. Last, also related to institutional financial resources, we expected greater instructional expenditures per students to be associated with higher completion rates.
Second, we hypothesized that institutions with greater commitments to faculty and academic programming might have higher completion rates, as such resources may facilitate students’ progression through the baccalaureate degree. Therefore, we assessed the percent of faculty who are full time and expected it to be associated with higher completion rates. We also expected institutions that confer doctoral degrees, as a signal of faculty and academic programming commitments, in addition to a baccalaureate to have higher college completion rates.
Third, we hypothesized that the makeup of the student body might be associated with completion rates. We measured the percentages of Pell grant recipients, students who have institutional grants for financial aid, and students of color and expected that institutions with higher percentages of these groups may have lower completion rates. We also explored whether the percent of women or size of the student body was associated with completion rates, although the expected direction of the effects was not pre-specified given the diversity of previous findings.
Results of this study may be useful in understanding and then addressing institutional barriers that may limit a college’s completion rates and provide insights into factors that could be tackled to improve these rates nationally, and thus contribute to higher education’s effectiveness in enabling social and economic mobility.
Methods
Study Design and Sample
We conducted a cross-sectional analysis of 4-year colleges and universities using the most recent data available concerning 6-year graduation rates in the 2018 to 2019 IPEDS. The 6-year graduation rate is the standard measure of college completion, and evidence suggests it captures the vast majority of students who graduate from college after matriculation (Long, 2018). IPEDS is a system of interrelated surveys conducted annually by the U.S. Department of Education’s National Center for Education Statistics (NCES), which gathers information from every college, university, and technical and vocational institution that participates in the federal student financial aid programs, as required by the Higher Education Act of 1965. The unit of analysis was the institution of higher education. Of the 2,847 colleges and universities that registered for the 2018/2019 IPEDS survey, 681 (24%) institutions did not complete key IPEDS data elements used in this study and, therefore, were not included in the analysis. Although many methods exist to impute missing data, in this case, 681 institutions lacked data on completion rate (i.e., the outcome of interest) or key independent variables; thus, we opted not to include these institutions in the analysis. Our analysis sample, therefore, included 2,165 institutions, operating before the onset of the COVID-19 pandemic.
Measures
The outcome was the completion rates measured as a continuous variable indicating the percent of students who completed their baccalaureate degree within 6 years of matriculating. For colleges with associate degree students, completion for the associate degree students was measured as completion within 3 years of matriculation. Independent variables included sociodemographic and economic characteristics of the students and available attributes of the institutions. Sociodemographic and economic characteristics of the students included percent women, percent students of color, percent of students who received Pell grants, and percent of students who received an institution grant for financial aid. Additional institutional attributes included the total number of students enrolled in baccalaureate degree, percent of faculty who were full time, instructional expenditure per student full-time equivalent (FTE), endowment per student FTE (measured categorically as none, $10K-24 K, $25K-50 K, $51K-200 K, $201K-400 K, and more than $400K per student), institution type based on the Carnegie Classification of Institutions of Higher Education, 2021 designation (doctoral, master’s, baccalaureate, baccalaureate and associates, special focus, and tribal), historically Black college and university (HBCU) designation, ownership type (public/governmental, private non-profit, and private for-profit), and geographical location based on the Bureau of Economic Analysis divisions.
Data Analysis
We used standard means and frequency analyses to report the outcome and describe the sample of universities and colleges. In addition, we used bivariate analysis to estimate the unadjusted associations between each independent variable and the outcome, and we calculated and tested the significance of correlation coefficients to examine relationships among the independent variables. In multivariable regression analyses, we included all independent variables that were hypothesized based on the literature and that were statistically significant in unadjusted analysis to fit a linear regression model. We assessed multicollinearity using the variance inflation factors (VIFs), which indicated no concerns of multicollinearity. To ensure adequate fit of the regression models, we conducted the Shapiro–Wilk test for normality; although we found evidence of non-normality, we had no evidence of significant kurtosis or skewness, and results from a log-transformed model did not differ substantially from the non-transformed model. We also plotted the residuals to ensure heteroscedasticity and conducted the Durbin–Watson test for correlated residuals, from which we concluded we had independence of residuals. In the unadjusted analysis and the primary multiple regression, we used p-value <0.05 as the threshold for statistical significance. All analyses were completed using SPSS.
In sensitivity analysis, we used interaction terms to test whether the effects of institutional financial resources and commitment to higher education were modified by characteristics of the student body. Thus, we interacted institutional attributes of ownership type, endowment, HBCU status, instructional expenditures per student, faculty resources, and geographic location with each of the student body characteristics including percent Pell grant and instructional grant recipients, percent students of color, percent women, and student body size; given the multiple exploratory analyses regarding interactions, we used p-value <0.01 to determine statistical significance for the interaction terms.
Results
Description of the Sample
Description of Sample.
On average, institutions’ enrolled students were composed of 47% Pell grant recipients, and 62% of students received institutional grants for financial aid; 43% were students of color, and 55% were women, with an average student body size of 464 undergraduate students (standard deviation 675 students).
The characteristics of the subset of institutions that offered only a baccalaureate degree (N = 476) are also displayed in Table 1. Compared with all institutions of higher education, those that offered only baccalaureate degrees were significantly more likely to be a private non-profit (vs. public/governmental or private for-profit), have higher endowments per student, be HBCUs, and have higher percentages of full-time faculty. They also had lower percentages of Pell grant recipients and higher percentages of students receiving institutional grants and lower percentages of students of color. Overall, the student body size was smaller than institutions that offer baccalaureate and more advanced degrees (all p-values <0.001).
Graduation Rates
On average, the graduation rate among the 2,165 colleges and universities was 51% (standard deviation 28 percentage points). The 90th percentile of graduation rate of these institutions was 80%, whereas the bottom 10th percentile was 24%. Within the subset of institutions that only offered baccalaureate degrees, graduation rates were significantly higher (p-value = 0.03) at 55% (with standard deviation 21 percentage points), with one-quarter of these institutions having a graduation rate of 39% or lower and 25% having a graduation rate of 70% or higher.
Factors Associated With Graduation Rates
Unadjusted and Adjusted Associations From Simple Linear and Multiple Regression Models (N = 2,165 Institutions of Higher Education).
Magnitude of Effects and Subsample Findings
The magnitude of effects was notable in some cases (Table 2). For instance, in multivariable analysis, public/governmental institutions had 10 percentage points lower graduation rates than private non-profit institutions. Completion rates in institutions with at least $200,000 endowment per student exceeded completion rates in institutions without endowments by more than 13 percentage points. Completion rates increased by one percentage point for every $10,000 instructional expenditure per student and by two percentage points for every 10 percentage points increase in full-time faculty. The largest effects among the characteristics of the student body were observed for Pell grant recipients: For every 10 percentage points increase in Pell grant recipients, completion rates decreased by about three percentage points.
Unadjusted and Adjusted Associations From Simple Linear and Multiple Regression Models (N = 476 Institutions That Confer Only a Baccalaureate Degree).
Institutions With Multiple Challenges
A total of 530 institutions (about 25%) were below median in several of the factors most strongly associated with lower graduation rates: endowment per student, instructional expenditure per student, and percent of full-time faculty. In these institutions, the student bodies were significantly different from the student bodies at the other institutions. Specifically, these institutions had significantly greater percentages of Pell grant recipients (61% vs. 42%, p-value <0.001) and students of color (55% vs. 40%, p-value <0.001). The average graduation rate from these institutions was 41% (standard deviation: 19 percentage points), with a 25th percentile of 26% graduation rate and a 75th percentile of 51% graduation rate.
Discussion
We found on average that colleges and universities graduate just more than half of the students who matriculate to their institutions. Because completing college, not merely entering college, confers the benefits of higher education—such as improved employment, wealth, and health (Abel and Deitz, 2014; Krueger et al., 2019)—our findings are concerning. Although the industry is enormous, with more than 2,000 institutions across the country, we believe its force as a great equalizer or stepping stone to social mobility must be questioned if on average only half the students achieve the intended college degree.
Additionally, we found evidence to support our three hypotheses. First, in the area of institutional financial resources, we found institutions with fewer financial resources had significantly lower graduation rates. Specifically, public/governmental institutions compared with private institutions, as well as institutions with lower endowments per student and lower instructional expenditures per student, had significantly lower completion rates. Second, we found institutions with more commitment to faculty—measured by the percentage of the faculty who were full time and the offering of doctoral degrees—had significantly higher graduation rates. Third and most concerning, institutions with greater percentages of low-income students and students of color had significantly lower completion rates.
Importantly, low-income students and students of color were significantly more likely to attend colleges with the fewest institutional resources (e.g., endowments per student and full-time faculty) and the lowest graduation rates—reducing the likelihood that these students will reap the social and economic benefits of a college degree. Based on these results, it is difficult to view higher education as a stepping stone to greater employment, wealth, and livelihood. As scholars have argued (Foley and Green, 2016; Bennett and Vedder, 2015), although higher education confers income benefits to some, it also can exacerbate overall inequalities in wealth.
The extreme disparities across institutions of higher education are cause for concern as they point to persistent inequity in a societal structure that was designed to promote democracy, blunt economic inequality, and foster social mobility. If that is reliably true only for those who attend institutions in the 90th percentile by graduation rate, the message is quite different—as institutions of higher education may be re-inscribing systems of inequity that are contrary to their institutional missions. Greater investment in lower resourced institutions is likely needed in order for higher education to play a pivotal role in social advancement.
As the value of higher education—particularly amidst the pandemic and in the wake of rising tuition costs—comes under increasing scrutiny, it is critical to recognize the enormous diversity of the field and consider the context of the education. Experiences in colleges and universities with greater endowments, instruction resources, and full-time faculty have vastly different outcomes from those with limited resources, particularly in terms of successful degree completion. Sustaining a minority of highly selective institutions without needed investment in public and less selective institutions—which are largely attended by students who are low income and of color—seems at odds with the espoused goals of the field and with what would be best for a society seeking to promote the equalizing value of higher education.
Education researchers, policymakers, and educators who wish to foster greater equity through higher education would do well to explore ways to level out the playing field for institutions themselves, including potential partnerships that may benefit under-resourced institutions. Creative solutions could help under-resourced institutions make strides in improving outcomes for their students. These solutions might include finding ways to increase or shift funding to these institutions and their students in sustainable ways that promote education as a public good that requires diverse sources of public and private investment.
Connecting with the emergent literature on wealth and asset-building policy provides a valuable entry point for thinking through possible solutions (see, e.g., Shapiro, 2004; 2017; McKernan and Sherraden, 2008). Similar to the outcomes among well- and under-resourced institutions, this literature helps shed light on the relationship between individual or family wealth (much of which is tied to the home and assets one owns) and educational outcomes (what schools your children attend). Over the last two decades, this field has made various recommendations to combat asset deprivation. These interventions range from children’s savings accounts funded by private employers or charitable organizations to advocacy for lowering the estate tax exemption, which is currently $5.3 million for individuals ($10.6 million for couples) and redirecting tax revenue to asset-building programs for Americans with lower assets.
Within higher education, similar and practical solutions could be helpful in reducing economic disparities that shape the results documented by this study. For instance, advocates for educational equity such as The Education Trust (Del Pilar, 2022) have argued that increasing the minimum and maximum amount of financial support for Pell-eligible students might allow low- and middle-income students to focus on their college education, rather than also balance outside work commitments, thus increasing timely graduation. While Congress and the Biden administration increased the minimum (6.2% increase) and maximum (8.2% increase) Pell awards in Spring 2022, the increases fell short of the desired doubling of Pell awards hoped for by advocates (Del Pilar, 2022). Another approach might include using the higher education endowment tax revenue, generated by The Tax and Jobs Act of 2017, to support lower resourced institutional efforts to increase student graduation rates. Last, partnerships between differently situated institutions to help foster collaborations that benefit both private and public schools around the recruitment and retention of underrepresented groups and curricular offerings could be beneficial. Such partnerships might examine “positive deviant” (Bradley et al., 2009) institutions that have higher than expected graduation rates (given limited resources) and identify effective practices that could be replicated in peer institutions to enhance completion rates.
Our findings should be interpreted in light of some limitations. First, this is a study of institutions and not individual students. The organizational perspective is important, but we were unable to explore individual-level characteristics using multi-level modeling due to the lack of individual-level data in this national sample of institutions of higher education. Second, the data are cross-sectional and descriptive, and as a result, we cannot infer causality; nevertheless, given the consistency, plausibility, and strength of significant effects, the observed patterns warrant attention. Additional experimental studies are needed to establish whether these patterns arise from causal relationships. Third, because data were collected just before the COVID-19 pandemic, we cannot estimate how COVID-19 may have affected our findings, both in terms of overall graduation rates and factors associated with higher or lower completion rates. Last, the IPEDS data include a limited set of quantitative measures to characterize institutions. Future research would benefit from in-depth, qualitative data to more fully understand the variation in graduation rates across institutions and, importantly, to identify key features that may distinguish those institutions that have graduation rates that are better than predicted; such a study was undertaken nearly 20 years ago with a selected sample of colleges and universities (Gansemer-Topf et al., 2004) and bears repeating.
In conclusion, the field of higher education in the United States is at a crossroads. About half of students who matriculate to baccalaureate degrees do not complete the degree requirements within 6 years. This startling figure points to a potential challenge for many colleges and universities, especially as researchers and policymakers continue to debate the value and need for renewed public investments in higher education. Furthermore, institutional graduation rates are predictable and highly dependent on available endowments per student, instructional expenditures per student, and the percentage of faculty who are full time. Furthermore, our data indicate that the institutions with the fewest resources have on average larger percentages of low-income students and students of color—highlighting persistent inequity, which enrolling in college or university is unlikely to overcome. The field would benefit from directing more attention to these poorly resourced institutions of higher education in order to foster higher graduation rates and fulfill the promise of college to promote social and economic mobility in the country.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
