Do Elite Colleges Discriminate Against Asian Students?

[epiphany]

<p>

Rachel Toor, in “Admissions Confidential”, recounting her experience as an admissions officer at Duke, talked about Asian-American applicants looking a lot alike to admissions people, and that this by inference hurt their applications. She also said that admissions people didn’t talk openly about this, but that they tacitly shared the same viewpoint.

</p>

<p>…not looking nearly as alike as the white students, whose similarity was described with dripping contempt and mockery chapter after chapter – or didn’t you read the book? Because I did. In fact with such contempt and such stereotyping, if such assumptions and judgments had been directed toward any other group than toward whites, those groups would have screamed racism and hired lawyers. (Not having qualified as a “minority” by any definition, an OCR complaint would have been out of the question.)</p>

<p>The big story in Admissions Confidential, for those acquainted with reading analytically, is not buried or subtle. The revelation was that candidates who looked unusual vs. other candidates of the same background and origin, were the ones admitted from that group, once qualifications between candidates were indistinguishable. (“Indistinguishable” is not defined as the absence of score discrepancies.) In the group (BWRK’s) she spends most of the book discussing, she recounts the process of how she focused in on a particular economically advantaged student who had resisted pressures to go elsewhere than Duke, at considerable personal cost within her family. She describes it in maudlin detail. </p>

<p>The author also devoted some ink to discussing how town-and-gown expectations needed to be satisfied for business reasons. (Admitting a critical mass of locals.) There is absolutely nothing new or revolutionary about this. Same holds true for Harvard, Princeton, and more, and always has. Check out detailed viewbooks, which are public.</p>

[fabrizio]

<p>As for the continued “no smoking gun” posts, I see that no one has responded to the [Inside</a> Higher Ed](<a href=“http://www.insidehighered.com/news/2006/10/10/asian]Inside”>http://www.insidehighered.com/news/2006/10/10/asian) article.</p>

<p>

Jon Reider, a counselor at University High School…said that the bias is real – and cited his experience in his previous job as part of the admissions office at Stanford University.

</p>

[fabrizio]

<p>Re 679</p>

<p>Are you talking about magnitude or proportion? You say it “makes all the statistical difference if whites are more represented than Asians…,” which refers to magnitude. However, you then talk about percentages of white and Asian applicants being admitted under early decision, thus referring to proportion. Which is it?</p>

<p>Whites are the majority in our country. It should come as a surprise to no one that Duke has more whites than Asians, in terms of magnitude. However, as the Arcidiacono et al. study shows, Asian applicants are at least as qualified as their white peers, even for the subjective dimensions. Thus, while early decision may produce more white admits than Asian admits, the percentage admitted should be the same. It shouldn’t be “30 percent higher” for whites.</p>

<p>How ironic that an individual harping on others’ “faulty” statistics can’t even differentiate between magnitude and proportion!</p>

[fabrizio]

<p>Re 680</p>

<p>What do you mean he didn’t find a “3:1 disadvantage for Asians”? That’s in the opening article!</p>

<p>

Mitchell Chang, a professor of higher education at UCLA, said Asians have long complained about the “penalty” they face when applying to colleges. But Espenshade’s documentation of a threefold difference for similarly qualified students at elite private universities “is stunning. Really worrisome.”
…
Espenshade found that when comparing applicants with similar grades, scores, athletic qualifications, and family history for seven elite private colleges and </p>

<ul>
<li>Whites were three times as likely to get fat envelopes as Asians.</li>
<li>Hispanics were twice as likely to win admission as whites.</li>
<li>African-Americans were at least five times as likely to be accepted as whites.

</li>
</ul>

<p>Edit</p>

<p>The so-called “underrepresented” minorities’ having better chances than whites can be taken as evidence of “affirmative” action. Asians, however, have worse chances than whites. That is “negative” action. Asians should not have worse chances than whites. If they are equally qualified, their chances should be roughly the same.</p>

[epiphany]

<p>

There is no particular reason why Princeton is less likely favored by Asian cross-admits than by other ethnic groups.

</p>

<p>I take it you have not spent much time on CC. (Given your post numbers, and your conclusions.) Lurk and post for 5 years, then come back and tell me that among Asian families in only this environment (as a non-projectible but important example), Harvard’s ranking is a minor factor in their choices vs. the choices of all other groups. It drives the discussion on thread after thread. Assumptions about public perception of college diploma is more often than not, the deciding factor among Asian cross-admits. Yes, it’s non-scientific, but omnipresent enough to be an unavoidable probability.</p>

<p>Or live in an environment heavily populated by Asian aspirants to Ivies (“heavy” meaning 50%-60% of all those students prepared for competitive college admissions), and then tell us that differences in rankings aren’t first of all the primary determinant of the initial college list and secondly a primary determinant of the enrollment decision – for the overwhelming majority of such aspirants. I’m talking about those with the academic goods to be considered for such U’s, not the entire Asian student population on CC or in Asian-heavy locales. For those of us who work with and among high school juniors & seniors (in such environments), ranking + distance from family are the two most commonly cited reasons for accepting offers, among Asian students. Those two factors trump all others, including differences between/among academic departments. I’ve had arguments with many a student over a “better-fit” school, and have lost every one.</p>

[siserune]

<p>

@fabrizio

Re 679</p>

<p>Are you talking about magnitude or proportion? You say it “makes all the statistical difference if whites are more represented than Asians…,” which refers to magnitude. However, you then talk about percentages of white and Asian applicants being admitted under early decision, thus referring to proportion. Which is it?

</p>

<p>The claim is that if X percent of the white student population of Duke are early decision admits, and Y percent of the Asian student population of Duke are early decision admits, and Early Decision confers an advantage of about 190 SAT points (in the regression models of Christopher Avery and his coauthors) then under such a regression model this adds:</p>

<p>(X - Y) * 190 points to the difference between average Asian SAT and average white SAT at Duke. If the fraction of whites using ED is 30 percent higher this is 57 SAT points. The Duke study observed 47. </p>

<p>Your personal stat gods, Espenshade & Chung, claim that being an athlete at an Ivy League school is worth around 230 SAT points (out of 1600) in their model of admission. Duke has much stronger sports emphasis and is not bound by Ivy League restrictions on athletes’ minimum SAT scores, so it’s pretty safe to assume the figure should be higher for Duke. If, say, 10 percent of whites and 2 percent of Asians at Duke are recruited athletes that’s (0.10 - 0.02)*(230) = 18.4 SAT points added to the white/Asian gap.</p>

<p>This type of calculation can be continued to estimate the additional legacy and donor effects. There is also Duke’s huge quota for North Carolina admits, and geographic and high school effects that one can read out of Espenshade’s studies as being on the order of a factor of 1.5 contribution (each!) to the odds ratio. </p>

<p>50 percent here, 50 percent there, and soon you’re talking about real numbers.</p>

[Rtgrove123]

<p>As a high school senior right now, I have invested a nice bit of time reading over those “2013 SCEA/RD results threads.” Now, I have no real numerical proof to back up this claim, but overall in HYPSM it seems that Asians and Whites are treated about equally. Then, hispanics get cut a bit more slack…but I’ve seen these students with barely avg SATs (2200ish) get denied from these top schools. However, the bar is totally dropped for blacks and native americans. In both classes, they can completely screw up their SATs (<2100) and get in to multiple top ivies. Truly disgraceful. These schools should be reserved for the best and brightest, and if the best and brightest are whites and asians so be it.</p>

[fabrizio]

<p>

The claim is that if X percent of the white student population of Duke are early decision admits, and Y percent of the Asian student population of Duke are early decision admits, and Early Decision confers an advantage of about 190 SAT points (in the regression models of Christopher Avery and his coauthors) then under such a regression model this adds:</p>

<p>(X - Y) * 190 points to the difference between average Asian SAT and average white SAT at Duke. If the fraction of whites using ED is 30 percent higher this is 57 SAT points. The Duke study observed 47.

</p>

<p>So you’re saying that if Asians applied to Duke early decision at the same rate as whites, there’d be no “loss equivalent to 50 SAT points”?</p>

<p>The problem with this particular example as an “explanation” of the Espenshade and Chung result is that it forgets how small the early decision admit pool is compared to the regular decision pool. A pool that small cannot explain away the “loss equivalent to 50 SAT points.”</p>

[siserune]

<p>

@fabrizio

What do you mean he didn’t find a “3:1 disadvantage for Asians”? That’s in the opening article!

</p>

<p>I meant that this central claim of the article by US News writer Kim Clark, quoted in the post #1 of this thread, is wrong. Obviously, totally, embarrassingly, on-its-face wrong in a way that even an innumerate journalist (or College Confidential poster) could have seen a mile away, without any special statistics expertise. </p>

<p>It should be extremely obvious, for all sorts of reasons, that Asians cannot be having their “true” rate of admission deflated by a factor of three, when they account for around 20 percent at Ivy League schools. Berkeley, Caltech, Espenshade & Chung’s article and others all point to 30-40 percent as a maximum rate for Asian admissions under the selection procedures most favorable to Asians, such as just taking the highest SAT scores. The idea that the true rate is around 60 percent is something even a US News journalist should immediately grasp is ludicrous, by comparison with available data.</p>

<p>Note well that in Espenshade & Chung’s statistical model of admissions, undoing the odds-ratio of 1/3 that they find for the predictor variable “Asian”, would literally be the same as jacking up the Asian admission rate by a factor of 3. The top universities accept perfect SAT scorers at a lower rate than the rate of acceptance for Asians that this implies. If you seriously believe that Asians at Harvard, Princeton and Stanford are an academically stronger pool than the 2400 SAT scorers at Harvard, Princeton and Stanford, please tell that to Jian Li and the Office of Civil Rights and let them know that the case can be dropped!</p>

<p>

Espenshade found that when comparing applicants with similar grades, scores, athletic qualifications, and family history for seven elite private colleges and</p>

<ul>
<li>Whites were three times as likely to get fat envelopes as Asians.

</li>
</ul>

<p>That is, once again, a blatantly wrong summary of Espenshade’s regression output, as given in Espenshade’s slide presentation linked in post #1 of this thread. Have you read and understood those regressions, or do you just take what the journalists write as gospel and “spread the Word”?</p>

<p>

The so-called “underrepresented” minorities’ having better chances than whites can be taken as evidence of “affirmative” action. Asians, however, have worse chances than whites. That is “negative” action.

</p>

<p>Espenshade found exactly the same form of “negative action” for other attributes, such as taking more AP exams and having 100 point increases on the SAT. Do you consider that credible? </p>

<p>

Asians should not have worse chances than whites. If they are equally qualified, their chances should be roughly the same.

</p>

<p>I fully agree that students who take four AP exams should not suffer discrimination by admissions officers compared to students who only take one! Let us stamp out Negative Action!</p>

[Bay]

<p>

the bar is totally dropped for blacks and native americans. In both classes, they can completely screw up their SATs (<2100) and get in to multiple top ivies. Truly disgraceful.

</p>

<p>The only thing truly disgraceful about this, is that after all of the time you have invested in reading CC, you still think the only thing that matters for ivy admissions is SAT scores.</p>

[siserune]

<p>

@fabrizio

The problem with this particular example as an “explanation” of the Espenshade and Chung result is that it forgets how small the early decision admit pool is compared to the regular decision pool. A pool that small cannot explain away the “loss equivalent to 50 SAT points.”

</p>

<p>uh, the regular decision pool could be a million applicants and it wouldn’t change the calculation. The percentages in the calculation refer to the <em>enrolled students</em> at Duke. </p>

<p>For instance, if only whites use early decision, and the fraction of enrolled whites who got in through early decision is 10 percent, and early decision is equivalent to a 150 SAT points in its effect on admissions chances, then the white students are reaping a 15 point advantage per student over all other groups. Do you understand it now?</p>

<p>As I mentioned, similar calculations hold for legacy, athletic, engineering school, development and NC resident admissions. In the Espenshade and Chung model, these effects simply add up, regardless of whether they correlate with each other.</p>

[StillGreen]

<h1>678 to #679, “”^re: 678, it makes all the statistical difference if whites are more represented than Asians in these (ostensibly race-neutral) preference categories.“”</h1>

<p>All you are saying is that (including some of your later messages) some less-qualified (by Duke’s scoring) Caucasians maybe included because of hooks (in-state, athletes, legacy, etc.), which dragged down the scores of Caucasians overall. In-short, you are cherry picking to amplify your <em>desired</em> results ----- that the Caucasians may score higher (without your deleted group). Talk about bias… :-)</p>

[StillGreen]

<h1>662 to #664, ““And a lot (of Asians) are admitted each year to Princeton, are cross-admits to H, and choose to matriculate to H.””</h1>

<h1>685, ““I take it you have not spent much time on CC… Yes, it’s non-scientific, but omnipresent enough to be an unavoidable probability.””</h1>

<p>I shall take it as a compliment. Because, it’s seems that your experience at CC has only taught you to fudge your arguments.</p>

<h1>685, “” Or live in an environment heavily populated by Asian aspirants to Ivies… I’ve had arguments with many a student over a “better-fit” school, and have lost every one.“”</h1>

<p>Telling personal stories won’t make the cut here. The critical issue is on the difference of choices between the ethnics groups, instead of merely Princeton losing some Asian cross-admits to Harvard. You still failed to show that, (a) whether <em>less</em> Asian cross-admits than other ethnic groups, are indeed choosing Princeton over Harvard, and (b) to the extent of lowering the Asian (American) student population at Princeton by more than about 5 %.</p>

[siserune]

<p>

All you are saying is that some less-qualified (by Duke’s scoring) Caucasians maybe included because of hooks (in-state, athletes, legacy, etc.), which dragged down the scores of Caucasians overall.

</p>

<p>No, I am also saying that there are ways of quantifying those effects, if you believe the Espenshade & Chung numbers are credible.</p>

[Keilexandra]

<p>Interesting. I didn’t know that STEM majors required a higher total SAT for admittance; I thought it was merely a higher math SAT. Of course, the purported preference of URMs for non-STEM majors might account for the lower SAT average. Can you point to some studies regarding this?</p>

<p>Rachel Toor is hardly motivated to report that she, as an admissions officer, took part in racial discrimination. She claims that her evaluations accounted for “complaining” recommendations; we have only her word for that, and what of the hundreds of other admissions officers out there? This could be another possible cause of the discrepancy: the discrimination begins in HS.</p>

<p>I do not believe there exists a smoking gun to be fired. Discrimination, if it exists, will be far better hidden than that, deeply under the cover of holistic admissions and subjective judgments.</p>

<p>epiphany - Ancedotally, Asian immigrant parents with whom I am acquainted–including my own parents–see Harvard as the most difficult school in the nation to be accepted to, but would not be surprised by and would not discourage a student turning down Harvard for Yale or Princeton. Anecdotally, of course, but citing CC trends is not much more reliable than anecdote.</p>

<p>A question to throw out there: in, for instance, recruited athletes, aren’t there URMs as well? I’m thinking of black basketball players, et al.</p>

[fabrizio]

<p>Re 689</p>

<p>[On</a> page four](<a href=“http://www.nacacnet.org/EventsTraining/NC10/Baltimore/educational/Documents/C313.pdf]On”>http://www.nacacnet.org/EventsTraining/NC10/Baltimore/educational/Documents/C313.pdf) of the PDF linked in the opening post, we see that there are red figures highlighting a specific portion of the results for “Private - Model 5.” We see that compared to whites, the odds of admission are 5.26 times as large for blacks, 1.99 times as large for Hispanics, and only 0.33 times as large for Asians, with all three variables statistically significant at the 1% level.</p>

<p>Thus, we see that this is yet another case of status quo defenders’ employing anecdotes and rationalizations to try to explain away Espenshade’s findings instead of published research. How very unsurprising.</p>

[fabrizio]

<p>Re 691</p>

<p>And just how many early decision admits are there in the total enrolled student population? It is not large enough to explain away the entire “loss equivalent to 50 SAT points” finding.</p>

<p>Your explanation can easily be verified / disproved if we know the percentage of whites and Asians applying to Duke via early decision. Maybe we should ask Duke and see if they’ll give us this information?</p>

[fabrizio]

<p>Disclaimer: I am not trying to “shut” anyone up.</p>

<p>Now, having said that, I just have one question to those who insist that Espenshade’s findings are invalid / baseless / crap / [insert negative adjective here].</p>

<p>If it’s so bad, why hasn’t anyone published a paper pointing out how bad it is?</p>

<p>As I’ve repeatedly mentioned, Kidder’s study didn’t criticize any of Espenshade’s methodology or his actual results. As one who believes that “negative” action is a solvable problem, Kidder even agreed that the “loss equivalent to 50 SAT points” finding was evidence of negative action. Kidder’s sole criticism was that Espenshade had conflated affirmative action with negative action.</p>

<p>Some users’ posts here seem to suggest that it’s so obvious how invalid / baseless / crappy / [insert negative adjective here] Espenshade’s research is, you’d have to be an idiot to accept it. Well, then, gosh darn it all, a lot of social scientists must be idiots in our country if none of them have critiqued Espenshade’s research after four years!</p>

<p>Edit</p>

<p>Even though it was wrong, it was actually better when people used to respond to a discussion of Espenshade and Chung’s research by invoking the Kidder paper and claiming that E&C’s findings had been refuted. That’s how these discussions are supposed to be like; one side brings up a paper, and the other side brings up another paper that slams the original paper.</p>

<p>This doesn’t happen anymore because as I’ve said, Kidder’s paper did not actually refute Espenshade and Chung’s, and there are no other papers out there that cast doubt on their findings despite some users’ claims that the findings are “obviously” bunk.</p>

[siserune]

<p>

@fabrizio

On page four of the PDF linked in the opening post, we see that there are red figures highlighting a specific portion of the results for “Private - Model 5.” We see that compared to whites, the odds of admission are 5.26 times as large for blacks, 1.99 times as large for Hispanics, and only 0.33 times as large for Asians, with all three variables statistically significant at the 1% level.

</p>

<p>I’ll repeat again, and you can get the full statistics tutorial from your professors.</p>

<p>It is clear that when Kim Clark’s US News article says that “whites are three times as likely to get a fat envelope”, she means the 0.33 odds ratio from one of Espenshade’s logistic regressions.</p>

<p>The problem is that interpreting the odds ratio as “being Asian reduces your admissions chances by a factor of three” is wrong. If you believe that interpretation, then (keep reading further down in models 5 and 6 to see the SAT and AP odds-ratios) you also believe that “[in some ranges] increasing your SAT by 100 points reduces your admissions chances” and “taking more than a couple of AP exams drops your chances of admission”. Let us know how you feel about those propositions.</p>

<p>It also remains true that a 3-to-1 Asian disadvantage would (e.g., in Espenshade & Chung’s admissions model) imply nonsense such as a “true” 60 percent Asian share of admissions at Harvard, and that Espenshade did not create pools of similar white and Asian applicants and compare their results (as falsely reported in the US News article). US News should have noticed both of these things, as they don’t require any special statistical knowledge to understand. What the article shows more than anything else is that journalists and apparently quite a few others need adult supervision when using statistics.</p>

[Hunt]

<p>

If Asian students are rated higher on both quantitative and qualitative factors than white students, and white students enroll in a greater diversity of majors, and Asians are not being discriminated against, then it follows that Asians should represent almost all of the students enrolled in the “popular Asian majors” because they are overall best-qualified for that major. Is this true? (I haven’t looked.) If the admissions department is, as I suggested earlier, also trying to racially balance specific majors, then that still constitutes discrimination.

I don’t think that follows. Asians may be better qualified on average overall than white students, but that doesn’t tell you how qualified the white students are who are also seeking admittance to those science 'n math majors.
To use an analogy that I understand better, it may be that Asian musicians are, on average, more accomplished than white musicians. But if the Asians play only the violin, they are competing not with all white musicians, but only with those who *also *play the violin. It may well be (and in fact, I think this is true) that the average white violin player is a better musician than the average wind or percussion player.</p>