<p>Who would have guessed? Who else would think of “political entanglement” as “finite transaction costs”? You sure it is not “infinite transaction costs”? As we speak, I heard GMAC is back for the 3rd bailout, just show you what money can buy you.</p>
<p>Yeah, accusing them of not having a definition of merit is an unfortunately turn of phrase. Maybe I should have said too many definitions of merit instead?</p>
<p>I was using a much cruder (no pun intended) analogy- that of a filtration system to explain the admissions game. The elites are using different filters, some fine (Asian) and some coarse (developmental cases) at various (input) sources, and I simply failed to see how the mixing (diversity) improves the quality of the “product”.:D</p>
<p>I think I should release your considerable intellectual back to the topic at hand, before I am getting bulldozed, figuratively speaking of course. Love to see what you can add to some of the other debates ongoing.</p>
<p>
</p>
<p>Did not know that. So Asians can advance their cause with their own strategy, without having to have to resort to ACLU and ADL on one hand, or Al Sharpton on the other. How very interesting.</p>
<p>It is nice of you to continue substantiating my comments, however unwittingly. I do appreciate your labors. Allow me to return the favor with a free statistics tutorial.</p>
<p>
</p>
<p>You miscounted. The East Asian numbers are pumped up by incorrectly adding Canadians to the Winners/Top12 column, after (correctly) excluding them from the starting pool of USAMO qualifiers. The number of Canadian top 12 finishers in recent years was 2 (2005), 2 (2006), 0 (2007-8), 1 (2009), all of them with Chinese names. Correcting your table, it reads as follows:</p>
<p>The percentages clearly decline at the first selection, and the number of MOSP selectees needed to upset that pattern when the 2009 numbers come in, seems unlikely to materialize, though of course the details of the selection process vary year to year and anything is possible. For now, there are a lot of reasons to believe that the numbers only go down at the first selection and have done so for quite a few years. That includes the drop in East Asian representation in the USAMO qualifiers compared to the MathCounts winners (a category solidly owned by Asians for many years), the large drop when you go from IMO team selectees to gold medal winners (at least for the years I counted, 1998-2009), and the continued decline when considering undergraduate and postgraduate academic prizes as I had mentioned earlier. </p>
<p>The implication of this is that in statistical significance tests should drop the probability by a factor of 1/2 (i.e., include a sign test) for each year of known data to account for the decline at the first stage, which in our case means to divide the probabilities by 8 given the 2006-8 data. </p>
<p>If you buy that, the 2006-8 data are significant at 95% (p-values from 2 to 5 percent depending on what tests you run). If you don’t want to assume anything about sign tests, full data is available for 2006-2009 for the USAMOqualifier —> USAMO-Top24 —> IMO selection path. Listing students by highest selection point reached:</p>
<p>Chances of such discrepancies (about 65% better odds for a non-Asian USAMO qualifier to reach the top 24 and the same for the IMO) are tested by chi-square at p=5.1%, Fisher exact test p=4.9%. Either of those with a sign test correction of 1/8 gives p < 1%. </p>
<p>I should note that your tables under-report and mask the decline in percentages, by double, triple and quadruple-tabulating the more successful students in several different columns as they go up the selection ladder. This has the effect of squeezing together the percentages at different stages. Thus, your figures (after the correction above) for 2006-8 are:</p>
<p>In this example the difference is small, but the data is needed in this form for the statistical significance calculations. </p>
<p>
</p>
<p>I answered this already, but the larger point is that anything more than a tiny amount of data will be highly significant here, and your statistical significance estimates are manipulated downward by the usual tricks:
– cutting the sample down to single years with a “denominator of 6”
– using two-sided tests (doubling the probability) where one-sided tests are in order
– considering each year of data in isolation, instead of as a series of experiments that combine into a more significant collective experiment
– excluding additional information that confirms the pattern.</p>
<p>Concerning your other points:</p>
<ul>
<li>Including women would push the Asian numbers down further, because nearly all (80-90) percent of the high scoring females are Asian but the qualification rate for females (of any group) at each stage is far below the rate for males (of any group). But if you want to include them, by all means update the figures accordingly.</li>
</ul>
<p>-The moving-average (East) Asian share of IMO spots has, as is well known, fluctuated around 30-40 percent in recent years and it is absurd to draw different conclusions by carefully selecting the time window to have a “larger” share by 1 percent. The number on the team in recent years (going backward) has been 3,1,3,3,2,2,2,4,3,3,1,0. Good luck spotting any trends in that string. </p>
<p>
</p>
<p>I missed the 1988 vs 1998 discrepancy in their tables. I don’t think it changes anything to consider 1998-2007, if anything the pattern should be stronger. AMSP has existed only since 2006, as stated in the article that I cited. The implication of the 83 percent figure was clear: representation is higher at semi-selective programs than at the more selective USAMO-plus level of competition; on the other hand, the Asian USAMO numbers are partly a consequence of disproportionate use of such training. I don’t think that those camps are efficient in leading to qualification for the IMO, though they may have an effect on the MOSP numbers.</p>
Again, I’m only going by my own observations of schools, but while Jews don’t “pack” every activity, there is absolutely no activity that people would say that Jews don’t typically do that activity (except maybe for Fellowship of Christian Athletes). My oberservations tell me that Jews have very largely assimilated–there are plenty of Jews who do not want to be doctors or lawyers, and who don’t play the violin–although these were the stereotypes for Jews a few decades ago. I think that as a group, Jews still highly value education, which explains why they are so well represented in elite schools. As I’ve explained before, Asians in the U.S. NOW have a lot in common with Jews in the U.S. a few generations ago.</p>
<p>The 1996 winner is half East Asian. Among the remaining 84 winners and ties since 1925 (start date), 8 surnames are from South Asia, the other 76 are of European origin.</p>
<p>You have a tendency to invent mistakes and attribute them to others. I am just pointing out that exactly the same reasoning leading to the “Asian 3-to-1 disadvantage” also leads to such conclusions as </p>
<p>*** “white lower class applicants have a 91-to-1 disadvantage compared to Asian lower class applicants”. (Discrimination!) How realistic do you think that is and how is it any different getting to that conclusion than the one about the 3-to-1 Asian disadvantage?</p>
<p>*** “taking 3 or more SAT-II exams cuts your admission probability in half, compared to stopping after 1-2 exams”. Do you find this statement believable and if not, how is it different from the 3-to-1 Asian disadvantage also found in the study? </p>
<p>
</p>
<p>Espenshade’s regression coefficients for the above were significant at p < 0.001 and p < 0.01 respectively.</p>
<p>
</p>
<p>The SAT and AP patterns are quite significant. You can’t look at one coefficient in isolation. It is safe to assume that the SAT group of variables and the AP group of variables both are very significant collectively. What is key here is that the patterns within the groups of the SAT and AP coefficients are the same and also match the underlying admissions rate data in Espenshade’s earlier slides. This by itself increases the significance because if the coefficients were zero for all practical purposes, the up-and-down patterns in the coefficients would be random. They aren’t, everything fits together. It isn’t a fluke of randomness for the regression to claim that taking more than 2 APs is bad for your chances, and there really is no logical difference here between believing in the “3-to-1” Asian disadvantage based on this study and believing that taking an extra AP exam hurts you.</p>
<p>Those are much, much better examples than your previous two counterexamples.</p>
<p>Certainly, I find it interesting that the compared to a benchmark of three SAT Subject Tests, the odds ratio is greater than one if only one or two is taken, and that this odds ratio is statistically significant at the 5% level. I’m sure an individual who claims that the USAMO / MOSP / IMO numbers are evidence of certain groups’ encouraging their children to pick “low hanging fruit” won’t mind if I attempt an alternative explanation, yes? The last time I checked, three was the highest number of SAT Subject Tests a student could take in one day. Maybe most of the students who take three take them all in one day and thus get tired by the third exam and so do relatively poorly on that one. Then again, maybe not. In any case, I don’t think that explanation is any worse than the “low hanging fruit” conclusion.</p>
<p>I can’t find your “white lower class…compared to Asian lower class” statistic. Is it on page four?</p>
<p>Regarding your final paragraph, as usual, you’re making a huge assumption and then proceeding to act as if the assumption is in fact true when you do not know that it is true. Moreover, you say that “if the coefficients were zero…, the up-and-down patterns in the coefficients would be random.” Do you actually think the null hypothesis is that the odds ratio is zero? I don’t think so, but I admit that Espenshade does not detail what the null hypothesis was. If you ask me, the more likely null hypothesis was that the odds ratio is one. That is, he’s testing to see whether taking more AP exams affects your chances. We see clearly on page five that for model five, none of the odds ratios is statistically significant, indicating that there is insufficient evidence to conclude that the true odds ratio is not one.</p>
<p>The key is simple: the odds ratios of your SAT and AP counterexamples are not statistically significant! That’s why you can’t claim that “taking an extra AP exam hurts you.” You don’t have enough evidence to assert that the odds ratios were statistically different from their null hypothesized values.</p>
<p>This is very misleading, as it completely ignores the dramatic change in the demographics of the competitors.</p>
<p>Between 1925 and 1999, only three champions were of South Asian origin: Balu Natarajan in 1985, Rageshree Ramachandran in 1988, and Nupur Lala in 1999. There hardly seems to be any evidence of South Asian dominance.</p>
<p>But wait! Since 2000, five of the ten Scripps Spelling Bee Champions have been of South Asian descent. That’s pretty dominant in my book.</p>
It’s interesting that so many South Asians are excelling in this competition. Why would that be? Perhaps these are native speakers of English (with parents who are also native speakers of English)? It seems to me that the main qualities needed to be a champion speller are (a) a good memory and (b) the willingness to do a lot of studying to prepare. I think the second of these may well be something that is connected to cultural norms.</p>
<p>Siserune’s table appears quite misleading by only subtracting the Canadian USAMO winners but not Canadian USAMO qualifiers and MOSP. In scientific world, fudging number will be a career terminating event if get caught.</p>
<p>I think the strong performance of kids with South Asian heritage at the Scripps Spelling Bee is due to their “willingness to do a lot of studying to prepare.” Their parents don’t have to be native speakers of English; they only need to push the kids the right way.</p>
<p>In #761, Hunt inquired about the representation of Asians among top scorers on the National Latin Exams. coolweather responded by noting that Asians are top scorers in top national spelling bees every year. You responded by attempting to “refine [coolweather’s] broad statement.”</p>
<p>And, that’s my point: your “refining” is misleading. coolweather is right; Asians are present among top scorers in the Scripps Spelling Bee. It is implied that he is referring to recent years, not the Bees of many decades past. Even a casual observer can see the preponderance of young brown faces on ESPN when it broadcasts the Bee every year.</p>
<p>In the five cumulative post-2000 years that students of South Asian descent did not win, I am sure that they were still among the “top scorers.”</p>
<p>NCL specified that his/her count of USAMO qualifiers excludes Canadians. This means, of course, that Canadians must be excluded from all later numbers as well. But then NCL “added” Canadians when making the list of USAMO winners. Please tell us whether you think that is, to use your words, “career-ending scientific fraud” and “fudging the numbers”.</p>
<p>NCL was right to assume that Canadians should be excluded, because (a) they are, by, definition, not counted in the MOSP and IMO columns of the table, and we are discussing the progression through the different selection stages; and (b) the smaller and less competitive pool for making the IMO from Canada increases the chances of the top one or two Canadians for making the USAMO Top 12. People who in the USA would be gold medalists on the IMO team, have almost a one-year head start in Canada in reaching the level where they are basically guaranteed to be in the Canadian top 6, and thus have advantages in their preparation for the IMO, Putnam (and in consequence, USAMO) competition.</p>
<p>Since the pools of USAMO qualification, MOSP and USAMO winners are purely based on score, not citizenship, and the canadian did not get any break, I don’t see why canadian should not be included in the analysis. It would be better if all asian decents be included in the analysis. I believe that south asian kids performed just as good. It makes no sense to just analyze the Chinese and Korean decent resutls, which is less than 2% US population.</p>
<p>I wonder how this whole conversation–and the underlying data–would look if we disaggregated South Asians from East Asians. It seems to me–and again, this is a perception based on observation–that South Asians may be a bit more diverse in terms of academic interests and ECs than East Asians (specifically, Chinese kids). I’ve seen a number of South Asians interested in writing, for example. Are we really talking about Asians here, or only some subsets of Asians?</p>
<h1>781, ““Who would have guessed? Who else would think of “political entanglement” as “finite transaction costs”? You sure it is not “infinite transaction costs”? As we speak, I heard GMAC is back for the 3rd bailout, just show you what money can buy you… The elites are using different filters, some fine (Asian) and some coarse (developmental cases) at various (input) sources, and I simply failed to see how the mixing (diversity) improves the quality of the “product”.””</h1>
<p>If transaction costs were infinite, we are doomed. But are we already doomed, only awaiting for the end game to gradually unfold? Actually I don’t object to your hypothesis of the elites’ involvement with the ruling class and I appreciate your insight of “divide and conquer” (more Asian applicants are displaced by URMs than by Caucasians), but elite admissions is likely far down the list of probable causes and solutions to the recent problems in our society. I suppose you know very well that diversity and political correctness can also be effective management schemes for the-power-to-be.</p>
<p>Talking about money, have you considered that probably the Jews have become valued customers of the elites, through steady donations to the university endowments over the decades? Incidentally MIT and Stanford, among the elites most populated with Asian (American) undergraduates, have also received substantial contributions from Asian donors. You should have seen how courteous was the Chancellor of Berkeley (although a public elite), when he flew over Hong Kong to express gratitude for a 9-figure donation (of course, in US$). Economic Might can go a long way in pushing racial discrimination aside. :-)</p>
<p>The problem is that “perception[s] based on observation[s]” can vary wildly from region to region. Of the thirteen editors of Georgia Tech’s student newspaper, at least six are Asian: one Thai-Filipino, one Korean, two Chinese (three if you count Taiwanese as Chinese), and one Indian. Among the Georgia Tech “English” faculty, there is one professor of Pakistani origin and one professor of Chinese origin. Thus, my observations do not lead me to the same conclusions as your observations did!</p>
<p>Related to the oft-cited possibility of Asians’ not having sufficiently diverse interests is the equally common “Asians do not apply to a broad enough range of schools” statement, which emerged most recently in post #757.</p>
<p>If you go by USNWR’s national universities rankings, among the top twenty universities, only Vanderbilt and Notre Dame have freshman classes that are less than 10% Asian (8 and 7, respectively). Despite being the lowest in that group, both figures already indicate “overrepresentation”! Moreover, quite a few of the Top 20 schools had freshman classes that were at least 20% Asian.</p>
<p>I don’t think the evidence supports the notion that Asians are not applying to a broad range of schools, when you can find an “overrepresentation” of Asians at every school in the National Universities Top 20 UNSWR ranking. As what I hope is a nail in the coffin, even West Point has an “overrepresentation” of Asians; their most recent freshman class was 6% Asian.</p>
As we speak, I heard GMAC is back for the 3rd bailout, just show you what money can buy you.</p>
Actually I don’t object to your hypothesis of the elites’ involvement with the ruling class and I appreciate your insight of “divide and conquer” (more Asian applicants are displaced by URMs than by Caucasians), but elite admissions is likely far down the list of probable causes and solutions to the recent problems in our society. I suppose you know very well that diversity and political correctness can also be effective management schemes for the-power-to-be.</p>