<p>What I said is that early decision together with the other analogous preference categories would predict 50 or higher as the white/Asian SAT point gap. I gave an example of a 30 percent difference in white and Asian ED rates as corresponding to 57 SAT points, but the argument in no way relies on the ED difference being that high. The known regression-based quantifications of admissions advantage in SAT points are 190 for early decision (which Duke uses), 200 points for recruited athletes, 160 points for legacies.</p>
<p>At a typical top 10 school, American whites are around 50-55 percent with URM, Asians and internationals making up the rest. “Hooked” matriculants generally exclude internationals, are about 40 percent of the student body, and include some 15-20 percent (generously) URM. This leaves 20-25 percent of the enrollment as non-URM hooked applicants (athlete, legacy, donor, etc). These are overwhelmingly white. How much of the Asian enrollment do you think is athlete/donor/legacy? </p>
<p>This doesn’t include the North Carolina quota or the separate, higher-SAT admissions process for Duke’s Pratt School of Engineering. </p>
<p>Duke has a 13 percent quota on North Carolina enrollment, according to Rachel Toor’s book. I assume the level of NC preference is something like that for Hispanics (185 SAT points), where the actual numbers would drop by half under pure academic admission, but let’s guess something lower for this calculation, such as 90 SAT points per applicant (alter the guess as you see fit). The Duke study data are from students admitted in 2001 and 2002, when Asians were 1.7 to 1.8 percent of NC population, compared to 75 percent for whites (data from 2000 census + 2005 estimates). There is also a large difference in the within-NC SAT scores, because the Asians getting into Duke from NC are concentrated in the Research Triangle area and are largely children of scientists. You can calculate the North Carolina effect yourself, but it would only raise the expected gap.</p>
<p>Hunt - I thought you said earlier that Asians excel in STEM fields? In this case, music = all majors and violin = STEM majors. I am not aware of data, studies, or stereotypes that show Asians preferring certain STEM majors over others.</p>
<p>Do we know that recruited athletes are essentially all-white? I agree that Asian recruited athletes are few, but what about URMs? Again cf. black basketball players, although this is a fledging hypothesis since I’ve seen no data.</p>
<p>
[QUOTE=siserune]
There is also a large difference in the within-NC SAT scores, because the Asians getting into Duke from NC are concentrated in the Research Triangle area and are largely children of scientists.
[/quote]
You just gave the NC geographic preference boost to only whites, yet then acknowledge that Asians are indeed getting into Duke from NC. Like in CA (where Asian students dominate the UCs despite higher populations of URMs), I suspect the percentage of Asian admittees and matriculants from NC is much higher than the percentage of Asian students in NC.</p>
<p>Wait, it’s wrong because it’s wrong? That is some of the clearest reasoning I’ve read all day!</p>
<p>As for your alleged counterexamples, have you forgotten statistical significance? In Model 5, the 0.74 odds ratio for a SAT score in the 1300-1399 range compared to the 1200-1299 range is NOT statistically significant at any conventional significance level. What’s more, none of the odds ratios for the AP exams in that model is statistically significant. By comparison, the 0.33 odds ratio is statistically significant at the 1% level.</p>
<p>The assumption is that the Duke (non-URM) athlete/legacy/donor matriculants are whites in high majority, not that Asians don’t exist in this pool. With that said, the Duke lacrosse team didn’t look like it had many Asians.</p>
<p>
</p>
<p>URMs were subtracted from the outset for this calculation. </p>
<p>
</p>
<p>You’re confusing two different (and mutually reinforcing) statements.<br>
(1) Asians are outnumbered more than 40-to-1 by whites in NC, so NC preference will overwhelmingly accrue to white individuals. There is no way Asians are 40-to-1 overrepresented in any slice of the Duke applicant pool other than native speakers of Chinese.
(2) Asians in NC are concentrated in the Research Triangle area, which has high-SAT schools etc. This will amplify the difference because although there are some whites also admitted from this area (and they too are over-represented), their effect on the white SATs will be diluted by the whites from less educationally advantaged parts of the state.</p>
<p>Did I say the parents were enrolling? Of course not. The students themselves have not asserted “my parents are making me do it,” although in many cases – on & off CC – those students often admit to being heavily influenced by parental persuasion or preferences. I’m telling you what the preponderance of Asian students cross-admitted to H, Y, and P overwhelmingly have decided, in all of the settings I encounter them, especially in person. Of course it’s possible that my many examples are against the grain & are dramatically different exceptions. Did you hear me (read me) mention probability? It is not probable that statistically, the trends to prefer H run hugely counter to that in the broader population.</p>
<p>^ There is a lot less stated preference about that among cross-admitted whites than among cross-admitted Asians. But I have already acknowledged that we’re only dealing with probabilities when discussing this. It would be a lot more helpful (I’ve probably said this 12 times on CC over the last year, minimum?) to have stats on that, than to infer from even consistent experience.</p>
My point is that even if Asians, on average, perform better than whites on STEM majors, that doesn’t mean that the top students in those fields are all Asians. It just means that a higher percentage of the top group will be Asians than their representation in the population at large. Thus (for example) if the math department at an elite college were 45% Asian, that would suggest that Asians outperformed whites on average. But remember, there are white students–plenty of them–with top stats and other achievements.</p>
<p>To put this another way, if Asians are only applying to half of the available slots in the University, you can’t look at the overall admission rate of Asians to determine whether they are being discriminated against. That’s what I don’t think the studies referenced at the beginning of this thread really address. I agree that reasonable minds can disagree on how big a factor this is–but it’s something that could be studied.</p>
<p>siserune - I saw that you subtracted URMs at the onset. I didn’t see you subtract URMs who are also recruited athletes, i.e. the overlap pool.</p>
<p>With regard to statement (2), how does admitting high-stat Asians from the Research Triangle area in NC “amplify” a discrepancy? Does Duke also give explicit geographic preference to underrepresented areas of NC? I would think that the majority of white NC-quota admits are also from the Research Triangle area.</p>
<p>Hunt - As I’ve said previously, I acknowledge the influence of self-selection but doubt that it could account for the entire gap. I agree, it would be very interesting to see ethnic breakdowns of admits by academic interests. I don’t expect colleges to release that data any time soon, though. But my point was that if discrimination is not occurring, within STEM majors Asians should be overrepresented to the same degree as the projected overrepresentation in the overall pool. I don’t know whether this is true or not.</p>
Well, I’m not sure. What if the white applicants to the STEM majors are on average more qualified than the majors in the overall pool?</p>
<p>I looked at the Common Data Set for Caltech from 2008-2009. 40% of the students there are Asian (and it could be more, since the ethnicity of internationals isn’t given). Obviously, plenty of white students are also able to get admitted there.</p>
We are too focused on the undergraduate education here. At Stanford Financial Mathematics Department, out of 40 current graduate students, 3 Koreans, 9 Indians and 14 Chinese – about 70% Asians, if I did not miscount. Most of them received their undergraduate degrees from foreign countries or Stanford.</p>
<p>It seems that Espenshade has actually taken Kidder’s criticism into account. He separates “no affirmative action” from “race-neutral admission.” It appears that there is once more evidence of “negative” action in play.</p>
<p>If “affirmative” action were eliminated, white enrollment would increase by 4.6% while Asian enrollment would increase by only 2.2%. However, if both “affirmative” and “negative” action were eliminated (ie. race-neutral admission), then Asian enrollment increases by 15.1%! Indeed, under Espenshade’s simulation, if admissions were race-neutral, then only Asian enrollment would increase; all other racial classifications’ enrollments would decrease percentage-wise.</p>
<p>Now, Hunt does have a valid point. It would be even better if Espenshade could create a third category: “affirmative” action but no “negative” action. I would expect that black and Hispanic enrollment would remain unchanged, and Asian enrollment would increase by more than 2.2% but less than 15.1%; the change in white enrollment would of course be the negative of the change in Asian enrollment.</p>
<p>Again, I also have to note that Espenshade’s analysis would not reflect the potential factor we’ve been also been discussing–the possibility that Asian applicants are, at least in part, disadvantaged because they aren’t competing for all of the potential slots at selective schools.
The other thing I noticed looking at the numbers this time is the enormous disparity between Asians and whites in terms of whether they are first or second generation Americans. I think this supports the theory that cultural norms may be important here. This is also the point I was trying to make about the similarities between the Asians of today and the Jews of several generations ago.</p>
<p>Re-read my postings (701 and 691) more carefully, and you will see that there is no mistaken double-counting of overlaps.</p>
<p>In estimating the number of whites at 50-55 percent of student body, I subtracted all URM and internationals. No overlap there. </p>
<p>In estimating the number of “hooked” whites and Asians, I started from the accepted figure of 40 percent of enrolled students, and subtracted the 15-20 percent of URM. No overlap there, either. </p>
<p>In reckoning the effect on SAT averages, one must double count the overlaps, for both whites and Asians (post 691: “these effects are additive in the Espenshade & Chung model”). The admissions regression models used to quantify advantages in SAT terms, all assume that everyone is allocated a fixed number of points for each attribute (athlete, Asian, donor, valedictorian, etc) and the total number of points is what drives the admissions probability. This is the core of Espenshade’s simulation calculations of what would happen under different admission policies. If someone is a recruited athlete and a legacy you award them the sum of both effects, because the sizes of the effects were calculated under this additive model. </p>
<p>
</p>
<p>By making it larger. There’s one effect from preferring NC residents; they have a lower overall SAT score than those admitted from other states. There’s a second effect from the difference in SAT distributions between quota-sampled NC and the other nonquota-sampled states. The quota sampling makes Asians relatively higher scoring (compared to whites) within NC than they would have been in most other states, by diluting the pool.</p>
<p>In the most IQ-taxing fields, East Asian representation actually declines as one goes to higher levels of selection, while the white, Jewish and Indian (Asian) representation progressively increases. Apparently the latter groups use more sophisticated educational strategies leading to better long term results, even if they don’t feast quite as fully on the low hanging fruit in (and before) high school. The disparity becomes especially pronounced as you move further up the food chain looking at student research prizes, faculty appointments, professional awards and other top-level measures. </p>
<p>For example, there is something odd about the Asian percentages at the upper reaches of US high school mathematics. The percentage of Asian USAMO qualifiers is higher than the percentage that qualify for the US selection program for the International Math Olympiad, which is higher than the Asian percentage of IMO team selectees. According to the logic of Asian posters here, this disproportion should be a signal of “discrimination”. Good luck selling the idea that the math tests are culturally biased against Asians!</p>
<p>Here is an interesting exercise that might surprise you. Wikipedia has a list of computer science prizes and their winners. What fraction are East Asians, and what fraction are East Asian-Americans (i.e., were domestic first-degree applicants to US colleges, not foreign import PhD students)? Both numbers are very low compared to East Asian representation in those fields whether as students, professionals or academics. Indians, Jews and whites show the opposite pattern; their representation stays the same or grows in tandem with selectivity.</p>
<p>With these two comments, you turned from being a hero to a villain. I hope you know that. I agree with you, 100%.</p>
<p>
</p>
<p>This simply meant that Asians with higher SAT and more AP exams are less likely to be admitted than URM with lower SAT and fewer AP. </p>
<p>On another thread, I learned that at Penn, Jewish students made up 38% of the enrollees. Considering that their population is only about one half of that of Asians, this is truly amazing. Anybody has data showing why this may be so? This would go a long way in explaining the Asian phenomenon we are discussing here.</p>
You have to wait for that, maybe for another 20 years. What you feel about overpopulated East Asians – basically Chinese – are the first generation born in the late 1980s, when their parents started to come here as graduate students from mainland China during 1980s. By early 1990s, there were only about 40,000 Chinese graduate students in the States. I believe that class of 2011 was the first year those kids, whose parents often have 6 college degrees, started to apply colleges. In the past two years, the valedictorians of all the schools around me were those kids and everyone of them went to HYPS.</p>
<p>With regards to asian students, the new Espenshade’s analysis based on the rehashed 1997 data really does not reveal much that has not been discussed in his 2004 and 2005 papers. The fact that Espenshade is still mining these decade old data suggests either (1) he is very lazy, or (2) he has no access to newer admission data.</p>
<p>Without looking at newer and more detailed data, the OP’s question cannot be definitively answered and neither can the suspicion of discrimination be erased. The intended majors of applicants can certainly be a factor affecting admit rates, but it cannot be confirmed without the data. </p>
<p>From the available admission data of Princeton over the past few years, however, one has to conclude that Li Jian’s complaint and OCR investigation have had a significant impact on Princeton’s admission practice (even if we assume Li Jian’s complaint was completely baseless).</p>
<p>I’m calling bad statistics, again. There is nothing odd about “East Asian representation…decli[ning] as one goes to higher levels of selection, while [others’] representation progressively increas[ing].” When you change the denominator from 470 to 6, is it weird that some groups will be “underrepresented” and others “overrepresented”?</p>
<p>Even a cursory glance at [the</a> 2009 USAMO qualifier list](<a href=“http://www.unl.edu/amc/e-exams/e8-usamo/e8-1-usamoarchive/2009-ua/09-Qual_list.pdf]the”>http://www.unl.edu/amc/e-exams/e8-usamo/e8-1-usamoarchive/2009-ua/09-Qual_list.pdf) should convince anyone that the percentage of East Asian qualifiers is very high, perhaps even high enough to form a majority. I did not take the time to analyze all 470 Americans by ethnicity, but I did do it for the second and third pages. By my count, of the 114 qualifiers listed on those pages, 78 were of East Asian descent. I acknowledge that my figure may be an underestimate, as I did not include students whose mothers (but not fathers) were of East Asian descent, and I did not include students whose last names were anglicized.</p>
<p>Since siserune frequently fudges numbers to give examples, I’m sure he won’t mind if I do the same. 78/114 is about 68%, and .68*470 will round to 320. So, let’s assume 320 East Asians and 150 non-East Asians.</p>
<p>From the [IMO</a> page](<a href=“International Mathematical Olympiad”>International Mathematical Olympiad), we see that half were East Asian (Chinese) and half were white. Only Berman seems to be Jewish based on the last name. Larson probably has Scandinavian heritage, and O’Dorney probably has Irish heritage.</p>
<p>I’m not surprised if the percentages of USAMO qualifiers of Irish descent or Scandinavian descent were less than 16.66 bar %. If the percentages are, in fact, less than 16.66 bar %, boom, there’s your “progressively increasing” representation.</p>
<p>You think inferring discrimination from Espenshade’s results has race-neutral alternatives. When it comes to concluding that certain groups like to pick “low hanging fruit” from USAMO / IMO numbers, please enjoy your apples. They’re on me.</p>
I agree with you, 100%.</p>