Colorado’s Affirmative-Action Experiment

One of the key questions that will shape the future of affirmative action programs in higher education is whether universities can create sufficiently diverse student bodies without relying on race per se.  The 2003 U.S. Supreme Court decision in Grutter v. Bollinger affirmed the University of Michigan’s use of race but only because the university said it could not find race-neutral ways of producing racial diversity  – through, for example, admitting economically disadvantaged students, or those who rank in the top of their high school classes.

    The Supreme Court, which has grown more conservative since the Grutter decision, may have a chance to re-examine this question in the near future.  As I mentioned in a previous post, the Fifth Circuit Court of Appeals is currently considering a challenge to the use of race at the University of Texas at Austin.  Plaintiffs contend that Texas had previously achieved sufficient racial diversity through its “Top 10 percent” plan and socioeconomic affirmative action, an argument that may well appeal to the new swing justice on the Supreme Court, Anthony Kennedy.

    Now, a fascinating new experiment at the University of Colorado, Boulder (CU) suggests that some universities may be able to equal or even exceed the racial diversity that have achieved under racial affirmative action programs if they provide a sufficiently large boost to socioeconomically disadvantaged students.  The size of the preference provided is a key variable, the research finds.

    The new research grew out of practical considerations.  In November 2008, Colorado voters considered banning race-based affirmative action.  (They narrowly decided not to.)  In anticipation of a possible ban, CU hired a graduate student in education statistics, Matthew N. Gaertner, to formulate a race-neutral alternative focusing on socioeconomic status.  
    In a paper presented at the American Educational Research Association this past spring, “Evaluating a New Approach to Affirmative Action Policy: Results from a Randomized Controlled Study,” Gaertner explains that while previous research on the effects of class-based affirmative action simulated admissions decisions based on national data, CU conducted an experiment using ten admissions officers reviewing the files of a random sample of 478 actual applicants at CU.  

    These students had applied and been accepted or rejected under the race-based system, which serves as the baseline.  The ten admissions officers then participated in a second review with race stripped from the applications and employing metrics to give a preference to socioeconomically disadvantaged applicants of all races.  In the experiment, CU decided to provide a large boost to socioeconomically disadvantaged applicants – larger than that currently provided to legacies or minorities.  Under the race-based plan, holding grades and standardized test scores constant, underrepresented minorities were 1.4 times as likely as others to be admitted, while under the class-based approach, economically disadvantaged students were 2.2 times as likely to be admitted as other students.

    As expected, Gaertner found that using a sizeable socioeconomic boost, economic diversity increased compared with a system of race-based affirmative action.  But, surprisingly, racial diversity also increased, though the sample size was too small to yield a statistically significant result.  Acceptance rates for economically disadvantaged students increased from 72% to 81%, while acceptance rates for under-represented minorities increased from 56% to 64%.   The “somewhat surprising” increase in minority representation in the class-based approach, compared with the race-based one, “highlights the importance of the size of the boost conferred by identification in class-based affirmative action,” Gaertner writes.

    Significantly, even with the larger boost in admissions, the academic credentials of the two groups – those admitted under class-based and race-based affirmative action – were not much different.  The mean high school GPA was 3.56 for those admitted under the class approach, and 3.58 for those admitted under the race approach.  Likewise, the mean combined SAT score was separated by just 10 points:  1197 under the class approach, and 1207 under the race approach.

    Of course, there are limitations to the findings from any one university.  Boulder is a “large, moderately selective public universities”and the results may not translate fully to institutions that are different.  But the general lesson across contexts, Gaertner says, is that the size of the boost matters.  That’s an important point to remember should this issue come before the Supreme Court again.

Return to Top