Information on U-M Admissions Lawsuits

Study on effects of diversity reaches wrong conclusions

By Professor Stephen Raudenbush, School of Education,
Senior Research Scientist, Survey Research Center,
University of Michigan

The Supreme Court will soon decide whether the University of Michigan’s admissions procedures are constitutional. In defending its practices, the U-M has cited a stream of research indicating that when students of varied ethnicity interact in well-designed academic settings, substantial benefits flow to the students of all ethnic backgrounds.

Recently, Rothman, Lipset, and Nevitte published research they cite as undermining the University’s argument regarding the educational benefits of enrolling a diverse student body. Specifically, their survey of 140 colleges and universities correlated an institution’s percent African American with certain student outcomes, including students’ satisfaction with their education and students’, faculty members’ and administrators’ perceptions of the quality of education. The authors found a weak negative association, indicating that institutions with large proportions of African American students were perceived to have somewhat lower quality of education than were institutions with a smaller proportion of African American students. The authors reasoned that if more diverse institutions (as indicated by percent African American) display lower quality of education, the U-M’s arguments regarding the benefits of diversity must be false.

Rothman et al.’s conclusions are entirely unwarranted. To see why this is so, let us consider their central finding: “As the proportion of black students enrolled at the institution rose, student satisfaction with their university dropped, as did the assessments of the quality of their education, and the work efforts of their peers.”

Any reader would reasonably conclude from this statement that the authors had followed the history of institutions seeking to achieve higher levels of diversity by recruiting more black students, and found that increasing diversity in this way harmed educational outcomes. In fact, the authors did no such thing. Instead, they compared 140 institutions at a single point in time. Thus, the authors did not study what happens when the percentage of black participation in a preponderantly white institution rises over time.

An accurate summary of their findings would be: “Institutions serving higher proportions of African American students were characterized by lower perceived student satisfaction and perceived quality than institutions that serve lower proportions of black students.”

Do institutions serving large proportions of African American students suffer low quality because of their affirmative action policies? The study provides no basis for answering yes to this question. Affirmative action has been a serious policy option primarily at heavily white, selective, top-tier schools that wish to increase diversity. It has much more rarely been used at institutions that historically attract large numbers of minority students. The schools in the Rothman study varied in their percentages of African American students from 0% to 43%. Schools that enroll more than about 12% to 15% African American students have no relevance to the questions now facing the Supreme Court because those institutions do not employ affirmative action in admission; yet precisely those schools are driving the results of this study. Therefore, the Rothman study provides no basis for any conclusions about the potential benefits of adopting (or discarding) an affirmative action policy.

There are many reasons that institutions serving large numbers of minority students may offer a lower quality experience—reasons entirely unrelated to affirmative action. As part of the legacy of racial discrimination in American society, such schools frequently draw their students from socio-economically depressed communities in which students have little access to academically rigorous secondary and post-secondary educational experiences. These institutions typically have substantially lower operating budgets resulting in larger classes, lesser qualified instructors and relatively limited course offerings.

Indeed, careful scrutiny of the regression results in the Rothman study tells us that students in non-selective institutions, institutions with the largest percentages of black students, and students who commute are the least satisfied with their educations, think students do not work very hard, and think that their schools don’t deliver a high-quality education. To extrapolate from these findings to anything about affirmative action is going far beyond what the data support.

Later in the article, the authors introduce several controls for institutional differences other than "percent black." Among these, they control for private versus public status of the institution, institutional selectivity and commuter status. A common technique in the analysis of survey data is to “hold constant” such characteristics using statistical adjustments. However, controls used in this study are entirely inadequate. The vast majority of all educational institutions are non-selective. Within the non-selective sector of institutions (public and private, commuting and residential), there is massive variation in factors such as social composition, tuition, instructor qualifications, quality and demand of course offerings, and institutional prestige, among others. Simply controlling for selectivity, commuter status and public versus private status can in no way justify the inference that the negative association between “percent black” and quality is attributable to affirmative action.

A more subtle mismatch between the aims of this study and the Michigan case is the study’s focus on “percent black” as the key cause of improved educational outcomes. The University of Michigan does not argue that simply increasing “percent black” is sufficient to produce the student outcomes it seeks to attain. Rather, the University argues that promoting academic interactions among diverse students promotes certain valued outcomes. Having a non-trivial “percent black” (and, more generally, having a non-trivial degree of ethnic diversity) is a necessary condition if those interactions and those outcomes are to unfold.

In summary, U-M administrators are responsible for constructing a wholly qualified, rigorously prepared student body that will enable the university to pursue its educational goals. Rothman et al.’s findings suggesting that schools serving large numbers of black students are inferior to those serving few black students have no bearing on the decisions the University must make and, therefore, no bearing on the issues before the Supreme Court. Moreover, the authors’ suggestion that their findings can be used to predict what would happen if the University of Michigan decreased (or increased) ethnic diversity constitutes a gross distortion of their evidence and a violation of the basic standards of social science.

Top of page