Affirmative Action Doesn't Increase Minority Drop-Out Rates. (Also, a Cato Institute report is less than honest – there's a shocker.)

In the comments of an earlier post, Robert Hayes has been arguing that racial preferences in college admission are bad because they harm minority students through what Robert calls “the ratchet effect.” But “the ratchet effect,” as Robert describes it, is dependent on what social scientists have called the “fit hypothesis” or “the mismatch hypothesis.” If mismatch isn’t true, neither is ratchet.

So what is the mismatch hypothesis? Robert describes the mismatch hypothesis perfectly when he writes:

People who get racial preferences turn in dreadful statistics on completion and performance in academia. […] The cost of the policy isn’t paid by the institutions, it’s paid by the students. It’s paid by the really bright black kid who would do great at Cornell but fails out of Yale. It’s paid by the decently bright Hispanic kid who would do great at UT but fails out of Cornell. It’s paid by the adequate fill-in-the-blank kid who would have done fine at Oklahoma State but who can’t cut it at UT.

That’s an exact description of what the mismatch hypothesis claims. Asked to provide evidence for his argument, Robert linked to this Cato Institute report (pdf link). ((The Cato Institute is a right-wing think tank with a pro-market emphasis.)) But Cato’s discussion of drop out rates, written by Marie Gryphon, is shoddy at best. Empirical evidence shows that the mismatch hypothosis is fiction. The truth is, minority students in colleges that practice AA are more likely to graduate than minority students with identical academic “qualifications” (i.e., SAT scores, class rank, etc.) who attend less-highly-ranked colleges.

How can this be? Doesn’t common sense tell us that putting less-prepared students in a harder college is a recipe for creating drop-outs — that “the really bright black kid who would do great at Cornell… fails out of Yale,” as Rob said?

This is a case where “common sense” is mistaken. There are many reasons why that really bright black kid might be more likely to graduate from Yale. One strong possibility is that high-status schools have better support systems for students. Another is that the peer groups at high-status schools create a social evironment in which better academic habits are a norm. Higher-quality teachers and TAs could make a difference. As social scientists Alon and Tienda write, “empirical studies have repeatedly demonstrated the advantages of placing students in higher ability groups with better instruction, less distraction, more time spent on task, more academic role models, and more serious learning climates.”

Having more grants (rather than loans) available to undergraduates also makes a difference. ((Alon, Sigai. 2004, October 28-30. “The Influence of Financial Aid in Leveling Croup Differences in Persistence among Students Attending Selective Private Institutions.” Paper presented at the conference of the Association for Public Policy and Management, Atlanta, GA. Cited in Alon and Tienda, 2005.))

The exact mechanisms will require further research to pin down, but it seems likely that all of the above factors contribute in some way to lower drop-out rates for minorities in higher-status schools, compared to similarly-qualified (i.e., SATs, class rank, etc) minority students in lower-status schools.

So what kind of evidence do right-wingers marshall to deny the facts? Let’s look at the Cato report Robert cited. Cato’s researcher, Marie Gryphon, begins by criticizing Bowen and Bok’s well-known study The Shape of The River, ((William G. Bowen and Derek Bok (1988), The Shape of the River, Princeton: Princeton University Press.)) which examined outcomes for minority students at a group of top-tier colleges. In this article, Bowen summarized some of his and Bok’s findings:

First, there is no systemic evidence that race-sensitive admissions policies tend to “harm the beneficiaries” by putting them in settings in which they are overmatched intellectually or “stigmatized” to the point that they would have been better off attending a less selective institution. On the contrary, extensive analysis of data reported in The Shape of the River shows that minority students at selective schools have, overall, performed well. The more selective the school that they attended, the more likely they were to graduate and earn advanced degrees, the happier they were with their college experience, and the more successful they were in later life.

Second, the available evidence disposes of the argument that the substitution of “race-sensitive” for “race-neutral” admissions policies has led to admission of many minority students who are not well-suited to take advantage of the educational opportunities they are being offered. Examination of the later accomplishments of those students who would have been “retrospectively rejected” under race-neutral policies shows that they did just as well as a hypothetical reference group that might have been admitted if GPAs and test scores had been the primary criteria (which is, itself, a questionable assumption). There are no significant differences in graduation rates, advanced-degree attainment, earnings, civic contributions, or satisfactions with college.

Gryphon criticizes The Shape of the River by claiming “Bowen and Bok argued on the basis of SAT scores alone that equally qualified students are actually more likely to graduate if they attend more selective schools.” But that’s simply not true: Bowen and Bok also analyzed class rank (based on GPA) and SATs together, but still found that minority students are more likely to graduate when they attend high-status colleges. (Gryphon grudgingly acknowledges this in endnotes, where few readers will see it.)

Gryphon continues:

Economists did subsequently analyze the question of dropout rates in more detail and got very different results than Bowen and Bok. Economists Audrey Light and Wayne Strayer were able to better predict university completion patterns among students of different abilities. They did this by using methods that took into account unmeasured student qualities…. When student differences were held equal, Light and Strayer found that the likelihood of graduating from college depended on how close the “fit” was between a given student and his or her classmates in terms of academic preparedness. They write: “Our estimates reveal that the ‘match’ between student ability and college quality does have a causal effect on college completion.”

Gryphon’s footnotes credit these findings to a 2000 study published in The Journal of Human Resources. ((Audrey Light and Wayne Strayer (2000), “Determinants of College Completion: School Quality or Student Ability?” Journal of Human Resources v35.)) But when I looked the study up, I was surprised to find that far from using a deeper measure of student achievement, Light and Strayer did just what Gryphon falsely accused The Shape Of The River of: they used a single test score (the AFQT) ((AFQT stands for “Armed Forces Qualifying Test”; there are legitimate reasons for social scientists to use the AFQT, but it’s no better than the SATs as a measure of academic ability, and it’s ability to predict future college performance – like the SATs – is mediocre.)) as their sole measure of academic ability.

Furthermore, Light and Strayer’s Journal of Human Resources study isn’t well designed to address the question of affirmative action and “mismatched” minority students. The relevant question isn’t if students as a whole are more likely to graduate from high-status schools if they have high AFQT scores (which is what Light and Strayer’s study primarily addresses), but if students whose college placement might be affected by affirmative action – that is, black and hispanic students – are more likely to graduate from high-status schools than otherwise similarly “qualified” (i.e., identical SATs, class rank, etc) black and hispanic students at slightly lower-status schools. Although this study implies that affirmative action policies “may” lower the chances of graduation for some students, the question can’t really be answered without a more focused study design.

Interestingly, a later study by the same authors and using the same data source specifically addresses the question of affirmative action and college graduation. ((Audrey Light and Wayne Strayer (2002), “From Bakke to Hopwood: Does Race Affect College Attendance and Completion?” Review of Economics and Statistics, p. 34.)) So why doesn’t Gryphon discuss this later, more relevant study (apart from one mention buried in endnotes)? It’s obvious why: In their later study, Light and Strayer conclude that the data is “consistent with the notion that racial preferences in college admissions boost minorities’ chances of attending college and that retention programs directed at minority students subsequently enhance their chances of earning a degree.”

Light and Strayer rightly caution readers not to take their work as conclusive evidence that affirmative action works (there could be unmeasured confounding factors). Nonetheless, Light and Strayer’s results are not consistant with Gryphon’s claim that affirmative action is hurting minorities by pushing them into harder universities. ((Gryphon dances around this in endnotes by falsely claiming that the second, more applicable Light and Strayer study is only about “the most highly selective schools,” but that’s not true; Light and Strayer specifically say their results are for “colleges of all selectivity levels,” and anyhow, both studies use the same database.))

A third, more recent study, by Sigal Alon of Tel Aviv University and Marta Tienda of Princeton, ((Sigal Alon and Marta Tienda (2005), “Assessing the ‘Mismatch’ Hypothesis: Differences in College Graduation Rates by Institutional Selectivity,” Sociology of Education, Vol. 78 (October): 294-315.)) uses a more thorough methodology than either Light and Strayer or The Shape of The River (among many technical improvements, their analysis includes asians and hispanic students as well as black and white students). Alon and Tienda conclude:

…Our results are consistent with claims that minority students thrive at selective postsecondary institutions, despite their disadvantaged starting lines (Bowen and Bok 1998; Massey et al. 2003). Minority students’ likelihood of graduation increases as the selectivity of the institution attended rises. Our findings, based on three data sets and several analytical methods, suggest that the mismatch hypothesis is empirically groundless for black and Hispanic (as well as for white and Asian) students who attended college during the 1980s and early 1990s. On the basis of the robust evidence we presented, we conclude that affirmative action practices both broaden educational opportunities for minority students and enable minority students to realize their full potential.

These findings clearly demonstrate the advantages that are associated with attending a more-selective institution and call for future research to identify the mechanisms that produce such an advantage.

The evidence is clear:There is no “mismatch” problem with affirmative action. Being able to attend better universities increases the odds that black and hispanic students will graduate. Right-wing proposals to eliminate affirmative action, far from helping hispanic and black students, would deprive some minority students of access to the best colleges while lowering their odds of graduating.

[Crossposted at Creative Destruction. If your comments aren’t being approved here, try there.]

UPDATE: Robert responds to me here.