^No, he didn’t because my DD18 and her friends would have seen him in the admits facebook group and met him at admit weekend.
But turning down celebrities shouldn’t hurt an institution’s reputation. Cambridge/Oxford rejected Tony Blair kids even when he was in office.
Didn’t Jared Kushner get into Harvard with terrible (objectively) grades and SATs?
@imptime18 said:
I am so used to dealing with statistics on a daily basis that I can often tell when something is very unlikely due to random chance without doing the math. But since you were wondering about this, let me do one example to illustrate.
Let’s start with MIT. From the following link we see that MIT has averaged about 5.5% black students over recent years:
http://web.mit.edu/ir/pop/students/diversity.html
Given an incoming class of just over 1100 students, roughly about 61 students are black. About 18% of the black students failed to graduate, so 61 * 0.18 is about 11 students. If black students had the same failure rate as white students (6%), about 4 students would fail to graduate.
So we are only talking about 7 more students that failed to graduate. Can we actually say something useful with such a small number of students? The answer is yes. To see why, let’s use the binomial formula and assume that the true failure rate for whites and blacks is the same, at 6%.
Given a 6% failure rate, what is the probability that 11 out of 61 students fail? Using the following binomial calculator and using p = 0.06, n=61, and k=11, we see that probability of 11 or more students failing is 0.092%, or about 1 in 108. Clearly very unlikely.
Now, some of you might be saying “But he used the worst example–perhaps it could be random for the rest!”. But that is the wrong way of looking at it. The right way of looking at it is to count all the white students across some set of colleges and determine their overall failure rate, and then collect all the black students and determine the overall failure rate, and see how unlikely that is. Without going into the details why, that number will be much, much smaller, likely less than than 1 in a billion.
So no, the number of black students failing is not at all due to chance.
I anticipated a response like this.
If it were these non-academic issues that systematically affected blacks more than whites, then you would expect relatively consistent spread across colleges. But that’s not what you see. Instead you see a spread that generally widens with rigor.
The two colleges that best illustrate this are Columbia and Penn. They are widely recognized as being near-peers in terms of overall quality, usually considered #4 and #5 in the Ivy League. But despite this, Columbia is recognized as being more selective and considerably more rigorous.
Less rigorous Penn has a 3% difference in white/black graduation rates, whereas more rigorous Columbia has a 5% difference. And as explained in my post above, these differences are meaningful.
@OHMomof2 - “I’m not debating mismatch with you… Just challenging your assertion that there ‘isn’t any doubt [that blacks are disproportionately at the bottom of their classes]’ There is.”
Just a few more data points for readers out there, from the world of law schools. Professor Sander’s work has received enormous criticism of course for his conclusion that affirmative action actually hinders black law students’ career prospects (just google his name), but I am not aware of any criticism whatsoever of the facts that he presents. He is a consummate insider, with access to all the data.
The source article is here: http://www.adversity.net/Sander/Systemic_Analysis_FINAL.pdf
Keep in mind that not all black students at law schools are admitted on the basis of preference; some proportion - likely a very small proportion and only at the three or four most elite schools (HYCS) - would be admitted anyway based on their non-preference achievement factors (UGPA, LSAT, rigor, personal characteristics and leadership, etc.). Because of “cascade” effects, any and every well-qualified black applicant will basically be hoovered up by one or more of HYCS, leaving less selective law schools much more reliant on race preference to meet their institutional goals. The shockingly large gaps in credentials at U of Michigan (a “near” elite) hint at this cascade effect (Table 2.2 on p. 405). The even more shocking credentials gap at all schools in aggregate - at greater than 2.3 standard deviations (!!) throughout the selectivity levels - confirms it (see Table 3.2 on p. 416).
For data on the proportion of blacks at the bottom of their respective first year classes, see Table 5.1 (p. 427) for “elite” law schools and Table 5.3 (p. 431) for all the other quintiles of schools. Table 5.4 (p. 435) shows that there is little to no real improvement by graduation, despite survivorship bias that reflects the higher dropout rates for the lowest performing students.
For data on the differential graduation rates between blacks and whites, again disaggregated by quintile of school selectivity, see Table 5.5 (p. 437). Interestingly, just eyeballing you can see that the differences are hugely significant (no reason to run a binomdist) at all school levels, but that the most elite schools are the “easiest.” This is of course well known to anyone who has been in the Ivy League - as kids have been saying for going on 30 or 40 years now, “The hardest thing about this school is getting in!”
With regard to licensure, Professor Sander extracts the relevant data: “Out of the 1346 blacks in the LSAC-BPS sample who took the bar, 516 (nearly 40%) failed at least once—nearly five times the white failure rate.” (p. 448)
Mismatch is a very real phenomenon, and I will never understand the reflexive support by some for a system (AA) that effectively identifies the lowest performing students at any institution by the color of their skin.
Those numbers are striking. And I’m more skeptical of AA in professional school admissions than in undergraduate. I still believe it is fine for schools to more or less auto-admit indisputably qualified URM students, but by graduate or professional school, most of the applicants have already spent four years at colleges that have offered them a fine education. Lowering standards to account for differences in background – except perhaps for making allowances for struggles early in one’s college career – is no longer justifiable, IMO, barring a small number of genuinely exceptional cases (for students of any race).
A problem at both the graduate and undergraduate level is that any elite but not super elite school that saw a sharp decline in its URM population would face huge backlash, no matter how transparent they were about their reasons and no matter how justifiable those reasons were. Whereas we should accept that one of the unavoidable consequences of AA, in a system with integrity, would be the existence of a kind of doughnut hole in which the second or third tier of elite colleges simply weren’t getting enough qualified URMs to constitute a significant percentage of students.
One thing I would like to see (but will never happen, for a lot of reasons) is posted SAT/LSAT/etc minimums. Those minimums should be well below the average score for a student at that school to account for other factors, but I think it would be possible to come up with a reasonable algorithm for determining a bottom. I’d suggest something like 100 SAT points below the average score for an unhooked, academic admit. Yes, that would mean Harvard would miss out on a few exceptional students with scores below that threshold, but on balance I think it would be good for the system and the students.
Just a comment on Malia Obama’s and Chelsea Clinton’s school performance/scores, which we will never know.
Both of them had very smart parents that got in to Ivy League schools/Elite Schools before they were famous/hooked (other than maybe the URM bump for Barack and Michelle). (And unlike some other famous people, including Kushner! And I am pretty sure Trump got into Wharton as a transfer from Fordham because he knew someone…)
Barack Obama made Law Review at Harvard so he has to be pretty smart to do that. Not sure about Michelle but I know she was in some kind of magnet school in high school. Malia has a pretty good gene pool and had parents that demonstrated tremendous work ethic. Would this alone get her into Harvard? Of course not, but I think she probably has competitive stats, even for an Ivy. Just my guess. Probably the same for Chelsea.
@apprenticeprof - I agree with you that professional school is a slightly different animal than college. But when we are talking about the academic aptitude of the preference students, there is no reason to think that law school students are not representative of the undergraduate population. In many ways, I’d argue that you are looking at the very “best and brightest” of the race preference undergraduate population pools because law school (1) is traditionally viewed as a good entry point for those with political aspirations, and (2) is not highly dependent on quantitative aptitude. I have no doubt that if the real data on race preference students’ achievement at the undergraduate were ever to escape from the hallowed halls, we would see the exact same pattern of underperformance.
With regard to your suggestion of a floor SAT score - and, by extension, sort of a floor “academic index” of the type that Professor Sander convincingly shows is what is actually being used at the law school admissions level despite the fluff about “holistic” admissions (see pp. 400ff in Sander’s piece) - it would simply be impossible to keep the floor within anything close to 100 points on the SAT. That would be equivalent to about 2.5 points on the ACT, in each case about 0.5 SD. There simply are not enough high scoring black students at the right tail of the distribution. The data are hidden, but occasionally some escape and pop their heads up over the parapet. For the most recent data I know that had escaped, see the 2013 ACT data that were published, but quickly taken down. Web crawler to the rescue, though: https://web.archive.org/web/20141222152409/https://www.act.org/newsroom/data/2013/pdf/profile/AfricanAmerican.pdf
Take a look especially at Table 2.1 (p. 12) which shows actual numbers of black students scoring at given levels. There are fewer than 500 black students (out of about 240,000) who scored 32 or higher on the ACT. 32 is about the 25% for the fifteen most selective schools in the country, which collectively enroll well over 2,500 black students per year. Unfortunately, it is impossible to find comparable SAT data, but the latest fine grained data that I’ve seen (from the 1990s) strongly suggests that there are fewer than 500 black students who score above about a 1450 (using the modern scale). I’ll spare everyone the statistical speculation, but it would be implausible to have a floor - even for the admittedly smaller cohort of otherwise “unhooked” black students (apart from the URM hook) - that is not at least a full standard deviation below the floor for non-black applicants and still maintain approximate current representation in the entering classes. In other words, just as Espenshade found using decades-old data, the “bump” that black students are in fact getting is on the order of 200-250 SAT points and 2.5-3.0 ACT points. We may well see this confirmed as the Harvard Asian discrimination case moves forward under the current Justice Department.
^ Sorry for the typo, the ACT “bump” actually being received would need to be on the order of about 4.5-5.0 points. While this does seem a bit high (because of course black candidates with 32 composite ACT scores will often be rejected by Ivies), keep in mind that all we are trying to do is equalize odds of admission for a given metric. In other words, would non-black candidates with a 36 ACT be rejected? Sure, happens all the time. If the scale went to 37, you would also see rejects among the non-black applicants.
On the other hand, given that in 2013 there were exactly 4 black students (out of 240,000 test takers) who achieved an actual 36 on the ACT, and 27 students who received an actual 35, does anyone really think that many of these students were not offered admission to whatever their first choice school was?
Again, the law school world provides a useful analog for thinking about this. I’ve posted this before, but it seems relevant to post again. At Yale Law School in the mid-2000s, the 25% rank of admitted students had LSAT of about 170 and UGPA around 3.8. As Professor Sander showed, law school admission is largely mechanical, except at the edges. In 2002-03, out of 91,000+ applicants to law schools generally, there was precisely one black student who presented an LSAT of 170 or above and UGPA greater than 3.75. and that’s not just applicants for Yale Law School. That was the total for all schools. I bet that student was in very high demand!
To what extent - if at all - should a URM applicant consider any of this score info in building a college app list? Suppose you have a high-SES hispanic with high math scores and grades, ability, etc. Maybe that’s not uncommon, so we slot schools into reach/match/safety as for any other student, according to a combination of academic stats and admission rates and then hope for the best. Is that the correct approach, or is the student uncommon enough that reaches become low reaches, low reaches become high matches, high matches become matches, low matches become safeties?
My recommendation on these forums is that URM students should not generally assume that being URM will change a match to a safety or a reach to a match, since (a) not all schools consider race/ethnicity to begin with, (b) of those that do, they do not reveal how and how much it matters, which means that most forum chatter tends to greatly overestimate its effect. Many forum posters overestimate the effect of race/ethnicity (and legacy), because it is easily observable and comparable to that of other applicants, unlike other aspects of potentially greater importance, like subjectively graded essays, extracurriculars, etc…
You couldn’t find two schools with a greater difference in rigor than Penn and Columbia? LOL. Also, @penn95 is going to be here to contest your characterization of Penn as easy, look out 
That said…
If grad rates are the measure of what group is mismatched and which is not, then men shouldn’t be admitted to elites as often as women.
Women have higher grad rates at almost every college in the country. Should we discuss the special advantages men are getting and the spots they are taking from women who do much better on this metric? Should we talk about how men should step down a rung in selectivity because they clearly can’t hack it at the same rates as women?
6 year grad rates -
Columbia: 96% women, 93% men
Penn: 96% women, 94% men
MIT: 94% women, 92% men
I used those specific ones since you used them above.
But it’s worse elsewhere.
Amherst: 96% women, 91% men
Cornell: 98% women, 92% men
UCLA: 93% women, 88% men
So is still true that “the number of MALE students failing is not at all due to chance”?
It is very clear that women as a group have been outperforming men as a group in both high school and college performance for many years now. Outside of the most elite schools, admission standards for men are clearly more relaxed than that for women, for schools that attempt to maintain a near 50:50 mix.
It’s actually interesting to see how this changes according to university rank. As shown in your stats above, the differences in outcomes are small at the Top-10 level, and grow as university rankings get lower.
After a certain point, the universities stop trying altogether to maintain a 50:50 mix. That’s how BU and Brandeis end up with over 60% women.
Absolutely true.
@OHMomof2 - Of course, the disparate outcomes of men versus women are not due to chance. But I suspect that you have no real interest in the true cause of the disparities, which are rooted in genetic and behavioral characteristics. In short, men are going to show higher variance on intellectual measures, especially at the tails of the distributions (this is the main factor behind the extreme outperformance of men versus women at the very top levels of quantitative accomplishment, even at MIT as we saw from the Putnam results), but men are also going to exhibit lower levels (and greater variance) on some measures that are more desirable in students, like conscientiousness, time preference, impulse control, etc. This is no one’s fault; this is just the way evolution has left us wired. (As an aside, no one seems to doubt that there are genetic factors at work in the huge disparity between men and women - on the order of 10 to 1 - in, say, violent crime, but when it comes to willingness to quit school and start a company or pursue geometric topology somehow we get all bothered by the disparity.)
This is no big deal. It’s only a phony diversity that seeks to prove that we are all the same, rather than truly “diverse”! One answer to the obvious differences in the way males and females respond to instruction might be a move back towards single sex schools 
That would take care of the problem of differential graduation rates and grades, as each system could be tailored to the different intellectual and behavioral strengths and characteristics of its students.
@SatchelSF Re: post 1531. I never assumed URM enrollment rates would stay the same in the system I proposed – although in the case of HYP, I don’t know that they would decline as much as you suppose. That’s consistent with my belief that racial preferences among a pool of students of similar qualifications are OK, but racial preferences that require lowering standards solely on the basis of race are not. Maybe that guts diversity at the top schools, maybe not. Beyond uncertainty over the exact number of URMs that would meet the minimum threshold, a stated minimum would also encourage students who tested just short of that number to retake, and would probably spawn programs (perhaps even supported by schools themselves) to offer retesting vouchers and tutoring for people looking to improve scores to meet a threshhold. But yes, I think at a lot of schools, the numbers of URMs would decline – and while that isn’t ideal, that may be a better outcome (for both the school and students) than admitting substantially less prepared and qualified students.
The higher level of scrutiny I give to racial preferences in professional/graduate admissions is based on the relative opportunity levels facing undergraduate and graduate admit pools. It seems justifiable to me to give some leeway to a seventeen year old (of any race) raised by uneducated parents and attending a failing high school. It seems less justifiable to give equivalent leeway to that same student once he or she has spent four years at a very fine college (elite or not). That isn’t to say that I think the disparity no longer matters, but at a certain point, one has to stand on one’s own feet, or not.
Well said, @apprenticeprof. I agree that URM would not fall as dramatically as supposed (by upwards of 80%) at the very tippy top (HYPS) because they can “skim the top” without yield worries, though perhaps that much in aggregate at, say, the top 15 or 20 most selective schools. I am a bit agnostic about the numbers anyway - that 80%+ drop is basically what Bok and Bowen and other insiders say. But of course they are in the business of defending affirmative action and are insensitive to the message of URM inadequacy that is implicit in their forecasts.
Surely, a “gentle” boost for low-SES talent (of any race) as you propose seems to me eminently fair, and would “add back” a bit of the URM falloff that would result from the removal of the current quotas, a point made by Thernstrom in his critique of Wightman’s position that removal of race preferences in law school admissions would result in a 95%+ drop in black enrollment at the elites (see https://conservancy.umn.edu/bitstream/handle/11299/167638/15_01_Thernstrom.pdf).
And I couldn’t agree more that a more transparent admissions system would act as an incentive to actually improve URM performance, which has been stagnant for decades. That is one of the main points of Nieli that I hadn’t really appreciated enough. That transparent “target” should also mitigate, over time, any near term falloff in URM enrollment.
Don’t hold your breath, though. There is no real interest in changing the system to address URM underperformance and truly doing the hard work of uplifting people, a point which Shelby Steele makes very eloquently in his 1990 essay “White Guilt” (can’t find a link but it is available through JSTor). The large majority of race preference admits are from high SES (that includes both Obamas, btw) and that is just fine with the elites who run the campuses.
[/quote]
Trust me, I’ve noticed and often discussed with friends the fact that it’s men shooting up schools and committing most violent crimes. I am definitely bothered by that.
I’m not bothered by the disparity of men and women’s grad rates either. But if we arguing mismatch, let’s be consistent.
It’s interesting to me that women’s colleges have a smaller difference in % of grad rates for black/white students than co-ed colleges do, generally speaking. No difference at all at Wellesley, for instance.
At Howard women have a 9% higher grad rate than men, Spelman has a 14% higher grad rate than Morehouse. Makes me wonder if a big part of the black/white grad rate difference is due to men.
If getting into one of 20 or so super selective colleges isn’t repping the right tail of male academic 18 year old achievement, who is? I suppose it depends where you cut off the tail, as it were. But we’ve already noted the male/female grad rate disparity is even greater at less selective schools. But it’s fairly significant at elites too.
I don’t doubt that that URMs have relaxed admission standards at many highly selective colleges, but there are too many changing variables for you to assume that is the primary cause of the reduced graduation rate. Did the students in both groups have the same rate of not graduating in 6 years due to financial reasons, including working long hours? The same rate of leaving for family reasons? Tthe same rates of not graduating for sports reasons (redshirt freshman, going pro, etc.)? The same rate of leaving because they felt like they didn’t belong there? Did they get the same degrees? As an example, I’ll do a similar comparison for colleges that are as rigorous as the ones on your list, but do not relax admission standards for URMs (at least to the same extent). There is still a significant difference.
Berkeley: 3.2 (29% Black, 9% White)
Caltech: 2.5 (20% Black, 8% white)
It’s well known in the literature that if you match URM, whites and (Northeast) Asians for intelligence, URM will subsequently underperform and Asians will outperform the baseline white performance on subsequent tasks. In other words, intelligence proxies (like SAT scores + HSGPA, for instance) will underpredict Asian performance and overpredict URM performance, again relative to baseline. In this sense, testing is actually biased in favor of low performing groups, and biased against high performing groups, precisely the opposite from what people popularly think.
You can get just a flavor for the issue of predictive validity as regards racial groups here (see “Rumor 2”): https://files.eric.ed.gov/fulltext/ED562751.pdf
As I implied, though, you could read literally for years on this topic - it’s well trodden ground.
The reasons for this prediction bias are not fully clear, and perhaps not relevant as a practical matter. Clearly, there are factors involved in success that are not fully captured by testing or similar proxies, but which are very likely correlated. I have hinted at some possibilities above somewhere (conscientiousness, time preference, impulse control).
The implications for “mismatch” theory and preference systems are uncomfortable, to say the least. True “matching” would imply not adding points to URM test scores and HSGPA, but subtracting them. For Asians, the opposite. To be truly “match predictive” one would ADD points to their test scores and HSGPA.
Similar issues arise with male-female “matching”, but here one needs to determine what one is trying to achieve. That is because of the variance of both male intelligence, especially in the quantitative fields, as well as of male behavioral characteristics (and with regard to behavior it is not at all clear we know the metrics for testing anyway). If it is exactly matching female and male GPA and graduation rates, perhaps diminishing male credentials is warranted. Of course, there will be tradeoffs, like not being competitive in mathematics competitions anymore (extreme right tail talent, where males will outnumber females by at least an order of magnitude) and decreasing the odds that one of your dropouts will start the next Facebook, Microsoft, or Apple (time preference, risk taking and perhaps impulse control). These tradeoffs might well be a price worth paying for some institutions.
One might also consider - egads! - that the institutions in their style of teaching and performance evaluation might just be biased against males, and that the differences in grades and certain other academic metrics observed throughout life confirm that the whole society is hopelessly biased. I don’t really believe that, but note that that is a close analog to what is accepted by many as regards URM underperformance.
The simpler answer is just that males and females are different, and so are racial groups, both in intelligence as well as in behavioral characteristics. So what? No one metric defines - or should define - anyone, and every person is an individual with his own endowment, so who really cares about group averages anyway? Well, I’ll offer my answer here: it is the person who seeks to control the groups, and that is why these are such political hot buttons.
As I said, these ideas are uncomfortable. But if you want a real discussion, imo they must be addressed, keeping in mind Schopenhauer’s sustaining insight: “All truth passes through three stages. First, it is ridiculed. Second, it is violently opposed. Third, it is accepted as being self-evident.”
I’m not sure if this is what you’re suggesting, but I’d be very, very leery of tying performance to inherent biological differences in intelligence, and not just because it is an “uncomfortable” idea.
I’m admittedly no scientist, but valid studies generally require a control. But this is impossible to do when it comes to these kinds of tests, because there is really no way of separating out cultural and environmental from biological influences in a way that would pass scientific muster. Twin studies do a good job proving that intelligence is highly heritable on the individual level, but we have no equivalent mechanism for sorting out differences among groups.
To use an analogy, let’s take the claim by some studies that breast-feeding has a small but measurable difference in adult IQ. I know this claim is not universally supported, but it has been taken seriously enough in recent enough history that I think it will serve, and I’m sure there are other examples of relatively small lifestyle differences correlating with differences in IQ. The decision of whether to breast or bottle feed is a comparatively minor difference but, evidently, it is at least plausible that, on average, even if we are talking about siblings raised in the same homes, the breast-fed children will wind up ever-so-slightly smarter than their bottle-fed siblings.
Now compare that difference to the large, highly significant variations in behavior, culture, and experiences between members of different races. It would be shocking if these differences – some involving choices made by members of the group, some involving conditions faced by members of the group – didn’t manifest themselves in things like differences in IQ. But that’s not a matter of evolutionary destiny or inherent biological capacity, it is a matter of current environmental and social factors.
If people of African descent were biologically inferior in intellectual capacities, it would stand to reason that African-Americans – who, on average, are genetically about 25 % Caucasian – would outperform more recent arrivals from Africa. Instead, discussions of Affirmative Action often note that a disproportionate number of its beneficiaries are immigrants or the children of immigrants, who have far less, if any, Caucasian ancestry.