Sunday, October 30, 2022

Affirmative Action and Racial Prejudice

Suppose you are a white student at an elite college or law school in an America without affirmative action in admissions. You observe that although there are not very many black students in the school, the ones that are there are about as smart as you and your fellow white students, that being why they, like you, got in. You may notice that blacks other than your fellow students are mostly not as smart as you are but that is true of whites as well. So far as your own direct experience is concerned, you have no reason to believe that whites are smarter than blacks.

Suppose you are a white student at an elite college or law school in an America with affirmative action in admissions. A few of your black fellow students are about as smart as you are — the ones who would have gotten in without affirmative action — but most are not. If you are at a school one step down from Harvard or Stanford there will be few if any black students as able as the average white student at your school because students that good were accepted by a better school. On the basis of your experience of the fellow students you interact with you conclude, even if you are careful not to say, that blacks are less intelligent than whites.

Hence one effect of affirmative action is to increase racial prejudice. The point seems obvious but I have not noticed other people making it, so thought it worth doing so.


Anonymous said...

Obvious only if you have some basic understanding of how and why prejudices develop- which unfortunately seems to not be a whole lot of people, despite how zeitgeist this topic has been in recent years

Thomas L. Knapp said...

This post begs the question in that it simply assumes, without evidence, the "intelligence differential" factor that it uses to justify its conclusion.

Andrew Carey said...

Nice and short piece of logical reasoning from Mr Friedman.
I wonder if there's a territory which has ever practiced the opposite of affirmative action which would be affirmative discrimination.
E.g. Asian Americans being denied admission to a top place of learning unless they exceed the admission criteria by far perhaps. What would the majority white students think of the Asian Americans after a year or two of studying alongside these √úbermenschen?

Unknown said...

I would reserve the term "racial prejudice" for cases of animus. What we have here is the generation of statistical discrimination based on not realizing that you are basing your inference from observation of the outcome of biased sampling (in the statistical sense).

dcardno said...

Thomas L Knapp @2:37 am:
This post begs the question...

No, it doesn't. Assume identical intelligence distributions for blacks and whites, but that under affirmative action schools have an incentive to enroll blacks rather than whites. The top couple of percentiles of all applicants will go to Harvard, Stanford, MIT, etc. As we move the next few percentiles down the curve the outcomes change - the 90th percentile black student goes to Harvard / Stanford (because the administration wants to meet their affirmative action goals), while the 90th percentile white student goes to Ohio State or Michigan.
White students at Harvard / Stanford see more black students than you would expect, given their share of the population, and of a lower average intelligence than their white counterparts: this is expected - the black students are drawn from the 90th percentile and up, while the white students are drawn from the 97th percentile and up.
The same thing shows up at Ohio State / Michigan - their white students are drawn from the 90th percentile (to the 96th), but all the 90th to 96th percentile black students went to better schools, so Ohio / Michigan's black students are drawn from the 80th to 90th percentiles - and we get the same relative performance outcome at the 'not-quite-the-top' schools as at the best-of-the-best.
Obviously, this process continues down the line, and at every step here are more blacks enrolled than would be expected based on population share, and they are competing against white who are more capable. The perception of relative intelligence, by both blacks and whites, will be as described in the post - except the blacks will be encouraged to believe that they are the victims of systemic racism. It's true: they are - but it doesn't work quite the way they think.

Thomas L. Knapp said...

Tell me you don't understand the term "begs the question" without telling me you don't understand the term "begs the question."

Andrew Carey said...

@Thomas L Knapp: please explain more about the "intelligence differential" factor that in your view was used to justify Mr Friedman's conclusion.
I think dcardno has advanced a good argument that you could get the same outcome even if there wasn't one.
I'm possibly misunderstanding the term though.

Thomas L. Knapp said...

Well, that's the problem -- how can I explain that? The post simply assumes an "intelligence differential" between affirmative action and non affirmative action students, i.e. it begs the question of whether such an "intelligence differential" actually exists. If it explained/defined the assumed differential and offered evidence for its existence, it wouldn't be begging the question.

dcardno said...

The post simply assumes an "intelligence differential" between...
No, it doesn't - that is a straw man.

If there is an incentive to recruit / enroll students on anything other than their scholastic aptitude (or in addition to their aptitude) then students admitted on that basis will average-down the portion of the student body that shares that attribute. The example I gave explicitly assumes that intelligence is identically distributed between black and white candidates, and I assume David Friedman is making that same assumption. All that is needed is another basis for determining admission that's not tied to academic achievement - and that's exactly what "affirmative action" is.
Affirmative action has to make a difference to the black student at the margin - otherwise it's not doing what it says on the tin. If it's making a difference to the black student who is admitted, then it is also making a difference to the marginal student who isn't admitted.
The same thing happens with 'legacy' candidates or athletes, and no one bats an eye at that observation - especially when applied to legacies. Admission under non-academic criteria averages down (the academic performance of) the group that shares that characteristic, whether legacies or athletes.

Andrew Carey said...

@Thomas L Knapp
Can you explain a bit more what you meant when you said this "The post simply assumes an "intelligence differential" between affirmative action and non affirmative action students"

I read Mr Friedman's post and could only see him talking about colleges that apply affirmative action policies or do not. And Dcardno's post was clearly premised no there being no differential but walking through the consequences. You seem to be suggesting that there is such a thing as a "non-affirmative action" student. What colour skin is this person likely to have in your view?
It's rather confusing trying to establish precisely where you think the desired conclusion has been used to inform the premise.

Thomas L. Knapp said...

dcardno: The post did not cite "scholastic aptitude." It simply asserted a differential in intelligence. There's a difference.

Andrew: You seem to want to argue about the question that's being begged. I have no strong opinion on that question. I just noticed that it was being begged.

David Friedman said...


The post did not assert a difference in intelligence between blacks and whites but between students admitted under different criteria. If, to take a simple case, Harvard admits any white student with an SAT of at least 700 and any black student with an SAT of at least 600, the black students on average will have lower SATs. If for BU the cutoffs are 600 (white) and 500 (black) and students would rather go to Harvard than to BU, BU won't get black students with an SAT of 600 or more because they all went to Harvard, but all of their white students will have SATs of at least 600. So students at BU, black and white, will observe that white students always have higher SAT scores than black students.

Isn't that obvious? The reason Harvard, in my hypothetical, has a lower cutoff for black students is that they don't have enough SAT700+ black applicants for the number of black students they want. That could be because blacks are less intelligent on average, it could be because blacks go to worse schools, it could be because blacks are poorer — lots of possible explanations. But affirmative action exists because schools find they can't get the desired number of black admissions without having lower requirements for black students.

dcardno said...

Thomas Knapp:
A distinction without a difference - cite aptitude, intelligence, achievement, or whatever attribute you like; it doesn't matter, and it doesn't change the conclusion.
The analysis does not beg the question: it is quite clear that using multiple attributes for assessment will necessarily average-down the group performance on the original attribute. How could it be otherwise, if the "affirmative action" has any effect? If it has no effect then why, as a university or society, are we doing it?

Thomas L. Knapp said...

David, you write:

"The post did not assert a difference in intelligence between blacks and whites but between students admitted under different criteria."

True. And I didn't argue otherwise. I just noted that it assumed a difference in "smartness" based on those different criteria.

Whether those criteria (scholastic achievement, test scores, etc.) actually reflect "smartness" has always been a hotly argued question. Which you begged.

Side question: Have they started scoring SATs differently since the 1980s? Your hypothetical scores look very low to me. My recollection is that my SAT score was 1410, and it was not the highest score at my high school.

David Friedman said...

I was thinking of one SAT, not of the summed score of two.

I didn't realize that your point was to object to the relevance of admissions criteria such as SAT or LSAT. Whether or not they measure smartness, I think it's pretty clear that they correlate with performance in school, which is one reason schools use them. That's enough for the purposes of my argument.

Anonymous said...

Your premises are not correct; assuming the admission threshold is above both population means, then white students and black students will both tend to have scores close to the threshold. But black students will be much more tightly clustered by the threshold than white students are, because of their lower population mean.

Thus, at a school with no affirmative action in admissions, the black students will still be noticeably dumber than the white students are. But the degree of overlap between the black and white student pools will be much greater.

Thomas L. Knapp said...


I don't "object to the relevance of admissions criteria." I just notice that assuming those criteria reflect "smartness" is begging that particular question.

dcardno said...

@ Anon 11:36 pm
The point is that if anything other than intelligence is used in admissions criteria the group subject to positive affirmative action will have lower group performance on intelligence than the group subjected to negative affirmative action, even in the case where the intelligence distributions are identical between the two groups.
Your argument is that blacks will have a lower mean intelligence across the population, so at any above-mean cutoff they will cluster more tightly at the cutoff point than the whites. It is this argument that I thought Thomas Knapp was objecting to in the second post on this thread. Whether that argument is true or not; it is not necessary to come to David Friedman's conclusion.

Anonymous said...

Thomas Sowell discusses many unintended consequences of affirmative action in his book: Affirmative Action Around the World: An Empirical Study.
I think he may mention this one as well.

Anonymous said...

If there were a group of people that are both interested in keeping racism alive as long as possible, and spread it as wide as possible, and profiting from it by appearing the sole actor which protects the victims of racism from the perpetrators of racism, and without which the racists would take over and hideously oppress the minorities - would they support or oppose the affirmative action programs?

Bruce said...

Professor Richard Samdor makes the same general point in his book "Mismatch". He notes that the disparity in an index based on high school grades and SAT/ACT scores is very high in the second decile of schools (80th - 90th percentile). I don't think he discussed the inference while students might make, but just that the effect is that many black (and to a lesser degree Hispanic) students simply cannot keep up, get low grades, and drop out of more competitive majors or programs.

Two arguments against this.

1. White students might realize affirmative action causes this "cascading" effect and not draw the inference. To test this, one might ask white students questions reflecting negative racial attitudes, questions reflecting an understanding of Bayes' Theorem, and then see if there is a negative correlation between the two. : )

2. Other students including white students are admitted under different preferential standards, including legacy students and athletes. (Of course, some black students have legacy or athletic preferences too.) A white students admitted under none of these programs might observe less capable black students admitted under affirmative action and less capable white students admitted as legacy students or athletic preferences, and not make the inference (or conclude that he or she is smarter than everyone). Obviously, the number of such students in each category would matter.