Asymmetrical Information - Megan McArdle

06.24.13

Affirmative Action: Who Does it Help, Who Does it Hurt?

Neither the costs nor the benefits are entirely obvious

Like half the other journalists in Washington, I was pinned to Scotusblog this morning, waiting to see what the Court would say about affirmative action.  it turns out the answer was . . . not much.  They remanded the case back to the lower court, saying that affirmative action programs have to pass strict scrutiny, and neither the University of Texas nor the court had really tried applying strict scrutiny.  

I was all set to write a big post talking about What This Means for America, but there's little meaning to be had out of this remand.  So let me talk instead about what affirmative action means for America, and the Americans who live there.  

There's a certain irony in the fact that white students usually bring these affirmative action lawsuits (and that defenses of affirmative action are often framed in terms of white privilege).  The evidence seems to show that if completely race-neutral admissions policies were adopted at colleges and universities, the admissions rates for blacks and hispanic would fall dramatically . . . but the admissions rates for whites wouldn't change much.   The primary beneficiaries would be Asian students, who would fill nearly four out of five of the extra admissions slots.  

One of the oddest facts about college admissions is that everyone seems to be aware that colleges have imposed restrictive admissions quotas to keep Asians underrepresented in their student bodies, akin to the "Jewish quotas" which used to exist at Ivy League schools until the 1950s.  But no one seems particularly bothered about systemic, institutionalized racial discrimination against a large group of Americans.  I'm not even aware of any concerted effort by Asian community groups to shame universities into stopping this.  

That tells us who affirmative action hurts.  But who does it help?  There are effectively two schools of thought on this.  The first is ably outlined by William Bowen and Derek Bok in The Shape of the River, arguing that affirmative action makes a substantial difference in how many minorities are admitted to college, and that those minorities gain substantial opportunities from that admission, not just in terms of academic learning, but from networking, credentials, and general social capital acquisition associated with college attendance.  

The best argument against affirmative action is presented in Mismatch, by Richard Sander and Stuart Taylor.  The subtitle says it all:  How Affirmative Action Hurts Students Its Intended to Help and Why Universities Won't Admit It.   Their argument goes something like this:  the pool of black and hispanic applicants to college is substantially smaller than the representation of those groups in the population, and since they are more likely to come from inadequate high schools, they come into college, on average, with lower test scores and preparation than the average white college applicant.  Racial preferences are used to pull the representation towards (but not up to) the population share of these two groups.  (There are other preferences for pacific islanders and Native Americans, but the small size of these populations makes these preferences fairly uncontroversial, and also, hard to study).  

On this much, basically everyone is agreed.  But where Derek Bok and William Bowen see an advantage for the minority students who receive admissions preferences, Sander and Taylor argue that affirmative action actually places those students at a disadvantage.  They are less prepared for their classwork than white and Asian students, and are thus disproportionately likely to drop out, switch to an easier major (from engineering to English, say), or flunk the professional certifications required to use their degree.  In fact, the research underlying the book began when Sanders (who teaches at UCLA law school) began looking into the performance of his minority students on graduation and bar passage.  Neither was good. Minority students were highly overrepresented in the bottom ten percent of the class, which has only a one in four chance of passing the bar.  

Okay, but would they really have been better off not getting into law school at all?  Well, yes, if they're in that bottom 10%.  Most of those kids are graduating with six figure debt, and a degree that is perfectly useless if you cannot pass the bar exam.  But that's not quite what Sanders and Taylor argue.  In fact, they say that many of these students would have done better in their classes--and had a higher chance of passing the bar--if they'd gone to a school where the average student matched their level of preparation.  When you're among the least-prepared students in the class, you're likely to struggle.  The classes will be taught too fast for you, glossing over concepts that you may not have encountered before.  And it's apt to be discouraging to be behind.  This is not to say that a heroic student couldn't overcome this and go on to shine, but expecting heroics from every affirmative action student is probably not good policy.  

I am personally sympathetic to Sander and Taylor's thesis because I had the experience of struggling when I switched to a harder school.  In sixth grade, my parents moved me from PS 166--which was, at the time, one of the best public primary schools in Manhattan--to a private school.  My fifth grade class at PS 166 had been essentially taken over by two of my classmates, who could not be removed under the school district's rules at the time.  (They now have special schools where they send disruptive students, but at the time, in New York City, the de facto policy was no more than a 1-3 day suspension, no matter what the offense.)  The boys would disrupt the class for a few days, get suspended for a few days, and then the cycle would start over.  My class lost considerable time.  

I'm not sure how much that contributed to the difficulties I had at my new school.  Even if my class hadn't lost half a year to obstreperous 10 year olds, Riverdale had smaller classes and a lot more resources, and disruptions weren't tolerated.  I was ahead in reading, but very far behind in everything else.  

I didn't realize at the time why I was struggling.  It was only looking back over a few decades that I realized that all my classmates had much more preparation to handle the coursework than I did--including preparation on how to work hard.  I was used to sailing through my classes with absolutely no effort.  I'd say it took me until tenth grade to fully catch up.  

Based on my own experience, it seems plausible to me that affirmative action might make things hard for people admitted with large preferences, even if they themselves don't see affirmative as a source of problems.  I can even see why the differentials might persist into graduate school.

But what about Bok and Bowen's work?  They suggest that preferences increase graduation rates, particularly at very elite schools?  How do we reconcile this with Taylor and Sander?  

Actually, this might not be quite as hard as you might think.  Sander and Taylor say that this is exactly what mismatch theory would predict, because preferences cascade.  

Ummm, what? I hear you cry. 

Let me explain.  

Recall that at the top of the piece we pointed out that thanks to poverty, the percentage of African-american and Hispanic applicants in the application pool is smaller than their representation in the population.  Again thanks to poverty, segregation, and related issues, their average test scores are lower than those of the average white or Asian applicant.  

What this means in practice is that there is a small number of people in those pools with the very top test scores.  Those people are heavily recruited by the very top schools--the Harvards, Stanfords, and so forth.  They take most of the kids in the pool who have, say, higher than a 1300 SAT--not only because they have an elite name behind them, but also because they have bigger endowments to offer more generous financial aid packages.  Obviously, the Ivy league does not get every single top minority applicant; there are the historically black colleges and universities, and many students who choose a parent's alma mater or a school that is close to home.  In fact, William Bowen has done work suggesting that poor students often undermatch--attending schools ranked lower than their test scores and grades indicate they could handle.

Nonetheless, Sander and Taylor argue fairly persuasively that applicants from underrepresented minorities with good prep scores, indicating a high level of preparation, very disproportionately end up at the most elite schools; undermatching may be a function of where you apply, not where you matriculate.  

With elite schools sweeping up most of the minority applicants with scores about 1200 or 1300, this means that the schools with an average SAT of, say, 1300, need to dip farther into a small applicant pool in order to get a class that has a substantial number of underrepresented minorities.  In other words, they need to use a bigger preference than the very elite schools.  And then the schools ranked below them are dipping even farther into the pool, using an even bigger preference.  The unexpected result is that the less selective the school, the bigger the racial preferences, and the larger the mismatch.  Minority students at Harvard or Stanford, or an elite state school like UVA or Michigan, are very close to their fellow students in preparation.  Minority students at a second or third tier school are dealing with much bigger gaps.  According to the Sander and Taylor thesis, then, it is not surprising that Bok and Bowen found that minority students in elite schools had higher graduation rates than those at less selective schools.

If Sander and Taylor are right, affirmative action may be a policy that hurts Asians and helps no one.  But this is an uncomfortable thing to say.  For one thing, we don't know that they're right--there is hot debate over their thesis.  But even if they are right, the remedy is bound to be very divisive.

Ideally, if they're right, the remedy would be to end affirmative action at state institutions, and work hard at beefing up schools and early childhood education in minority areas.  But those two things aren't paired.  So what we're talking about is ending affirmative action, which feels--correctly or not--like taking something valuable away from two groups that have historically gotten a raw deal from America.  

Moreover, if we end it, we won't do so because it turns out that affirmative action doesn't work; we'll end it because the Supreme Court will have ruled that you can't give blacks and other minorities special help.  We'll have an officially race-neutral policy laid on top of an unofficial system that views unskilled men with black and hispanic names as equivalent to a white man with a prison record.  

It's tiresome to end an article with a call for more research.  But in this case, that's really what's needed.  It's worth noting that it should be relatively easy to tell whether Sander and Taylor are right, except that it's very hard to get the data.  And it's hard to get the data because institutions fight like hell to keep it from being released.  It's no surprise that we can't all agree on a remedy for historical racism.  But it should be easy to agree to study the problem.