Tuesday, August 28, 2007

Affirmative Action, Richard Sanders and Thomas Sowell

A recent op-ed in the Wall Street Journal discusses research by Richard Sanders suggesting that affirmative action by law schools actually reduces the number of black lawyers. The argument, and the evidence, is elegant and persuasive. If you classify law students by academic credentials and what tier law school they go to, black students and white students have about the same bar passage rate--a black who goes to (say) a second tier law school is about as likely to pass the bar as a white at a similar law school with similar academic credentials. But if you classify students only by academic credentials, blacks have a much lower bar passage rate than whites.

Sanders' explanation is a mismatch between students and schools. Black students do not, on average, end up in the same law schools as white students with the same credentials. Law schools compete to get black students, there aren't enough well qualified ones, so elite schools accept black students with qualifications well below those they require for white students. The result is that many black students end up in schools and classes they are not qualified for, learn little, and fail to pass the bar--students who would have done better in a less elite school designed for students more like them. He estimates that a race blind admission policy would result in fewer black students going to law school but more passing the bar.

After reading the op-ed, I told my wife about it. She pointed out that the argument was not original with Sanders. Almost twenty years ago, in his very interesting Choosing a College: A Guide for Parents and Students, Thomas Sowell made precisely the same point in the context of colleges rather than law schools. Black students at MIT had math scores well above the national average but far below the average for white students at MIT; they would have gotten a better education at a less elite engineering school.

17 comments:

Jonathan said...

Sounds plausible. I'm vaguely reminded of the Peter Principle: these students have in effect been promoted beyond their level of competence -- though for reasons entirely unrelated to the Peter Principle.

Russell Hanneken said...

The explanation is intuitively plausible, but it's difficult to reconcile with standard microeconomic theory.

Affirmative action doesn't force anyone to attend a law school for which he isn't qualified. It just gives some people the option of attending. Some people might overestimate their ability to perform at a given law school, but others will underestimate their ability. Basic microeconomics assumes that, on average, people will have correct expectations.

Scott Beaulier and Bryan Caplan have an interesting paper that addresses this issue. Drawing upon the field of behavioral economics, Beaulier and Caplan argue that there are certain biases people have that might make them inclined to hurt themselves when presented with certain kinds of options. They also say empirical evidence shows that the poor deviate more from neoclassical assumptions than other groups, though what this means in the case of affirmative action isn't clear, since it's targeted at minorities rather than poor people per se.

Jonathan said...

These are young people, right? They're delighted to be accepted by a good school, and I don't suppose it even occurs to most of them that they might be out of their depth until it's too late. They've probably done well relative to their peers up to that point in time.

I doubt myself that the average person in this situation has correct expectations. But some more research could be done to find out.

Mike Huben said...

The title of the Op-Ed is "Affirmative Action Backfires". What Sanders' study seems to show is that affirmative action is not optimal for blacks (what about women and non-black minorities?) What history shows is that affirmative action is much better than the naked free-market discrimination it ended.

Blind admissions might be implementable now, but back when affirmative action was introduced (to much racist and sexist opposition), it would have required extensive auditing to believe the processes were what was claimed. Affirmative action is much simpler to confirm, much simpler to believe, and much harder to evade.

Unknown said...

Affirmative action is much simpler to confirm, much simpler to believe, and much harder to evade.

Yes, but this is irrelevant if it doesn't work.

Anonymous said...

Some people might overestimate their ability to perform at a given law school, but others will underestimate their ability.

Do students select a college based on their ability? In my experience high school grads choose the best school that accepts them.

Daniel Rodri­guez said...

What history shows is that affirmative action is much better than the naked free-market discrimination it ended.

Really? Do you have facts that confirm this? It's against everything I know about affirmative action in nearly every place it has been implemented, from India to the States.

Isegoria said...

If affirmative action came about because of a massive shift in attitudes toward racial minorities and women, can we attribute the progress of the past few decades to the affirmative action policies?

Or should we assume that the massive shift in attitudes played a large part?

Mike Huben said...

Ah, the denialists are out in force.

The same skepticism that leads to doubting affirmative action was responsible for massive changes in wages and employment of minorities should also lead to questioning whether the bar exam is in some way biased, the way so many other exams have been.

Biased bar exams would simply undermine the entire premise of the study.

The whole meritocratic academic test system might well be systematically biased. There's an enormous literature on the subject.

Anonymous said...

'The whole meritocratic academic test system might well be systematically biased. There's an enormous literature on the subject.'

Perhaps there is an enormous literature on the subject - but that in itself is not a convincing argument. There is an enormous literature on astrology.

In the England, Chinese and Indian students outperform White students by a considerable margin. Black students underperform White students. Are we seriously to believe that there is a pro-Chinese, pro-Indian, anti-White, anti-Black bias in the English exam system?

David Friedman said...

Mike writes:

"Biased bar exams would simply undermine the entire premise of the study."

They would also be inconsistent with the evidence offered, which is that if you control for both academic qualifications and level of law school, blacks have the same bar passage rate as whites.

I think "denialist" was your term?

Steve Miller said...

"Really? Do you have facts that confirm this?"

Apparently not. One assertion is backed by another. Is Mike a closet praxeologist?

Anonymous said...

The premise that needs examination is that the proportion of talented is the same regardless of population. See this:

http://www.udel.edu/educ/gottfredson/30years/Rushton-Jensen30years.pdf

Anonymous said...

Link was too long:

Here's the tiny url:

http://tinyurl.com/cxhmw

Anonymous said...

I applied to several grad schools recently and was accepted only to my last choice. I was a little disappointed at first -- it's not a very prestigious school. But as I get into the program I'm realizing that they are geared towards teaching me the skills that I'm lacking that prevented me from getting into the other schools, where those skills would have been presumed from the outset.

It'd be swell if I were good enough to get into MIT, but given that I'm not, it's probably very good for me that I did not somehow get accepted anyway.

Anonymous said...

Mike Huben's first point is important. If you want to enforce antidiscrimination laws, you probably have to either:

a. Ban only explicit "No Irish Need Apply" type discrimination and ignore places that accept applications from blacks but never hire/admit them.

b. Require some kind of measurable performance on hiring/admitting minorities, the simplest of which is "let me see how many black employees/students you have."

I think affirmative action is a bad idea, but I can see why it came about. All the other ways I can imagine to enforce nondiscrimnation seem like they're really hard to get right. The only plausible alternative I can think of would involve regulating the process of hiring (which is also done), but that requires the government or courts to be up to their elbows in every major employer's hiring process, and I suspect that discrimination could be done by a determined person despite most of the procedural safeguards.

Anonymous said...

I wonder if the same effect can be found among legacy admissions.