Where is the proof that Harvard rejects so many percent of perfect scorers?

[tokenadult]

I've seen several statements from several CC participants in various threads recently that Harvard rejects 50 percent of all the students who apply to Harvard with perfect SAT scores. Where is there a convenient source to verify that statement? Where is a webpage produced by Harvard that says so? Where is there a journalistic report of exactly such a figure?

One student commenting in a thread just today mentioned that a 50 percent acceptance (among the other 50 percent of perfect-scoring students who presumably were NOT rejected) sure beats a 7.something percent overall base acceptance for all applicants to Harvard, so maybe high scores aren't such a bad idea. Of course, the frequently made statement about half of perfect scorers being rejected implies that high scores are not a guarantee of admission, an idea with which I certainly agree. But I wonder if the statement is even numerically true. I wonder if Harvard even bothers making such a statement officially.

[molliebatmit]

The Revealed Preferences Paper has a graphic of the acceptance rates by SAT percentile at Harvard, MIT, Yale, and Princeton. It has a "100th percentile" data point, which I have always assumed was a perfect SAT score.

If my assumption is correct, then even more perfect scorers are rejected -- that data indicates that 50% of perfect scorers are rejected at MIT, 60% at Yale and Princeton, and 80% at Harvard.

[tokenadult]

I thought that was the source of this statement. That would be an erroneous inference, because the top percentile of scores

http://www.collegeboard.com/prod_dow...tile_ranks.pdf

http://www.collegeboard.com/prod_dow...tile_ranks.pdf

includes scores that are lower than "perfect" scores, and always has.

[cptofthehouse]

Token, I have seen remarks from top schools, not necessarily H, say that top schools have rejected perfect scorers. Also have heard anecdotally from press, and other sources. Not accurate reporting, I'm afraid.

However, I know that many colleges use scores ranging from say 1-5 in assessing areas of the application. When assessing the test scores, the top score of 5 does not distinguish between perfect and the threshhold for that score. Also schools that use an academic index comprised of SAT2s will use all 5 or 6 numbers, again diluting the impact of a perfect SAT1. In other words, the perfect SAT1 scores are not highlighted but are mixed into the other admissions criteria.

[grade inflati0n]

why does it do that?!?!?!? i'm sure correlation=/=causation but why is admissions %s higher around 92%? is that cuz less people apply and if they apply it means they have some qualifications to offset their low scores?

http://img165.imageshack.us/img165/9...ceoffergx6.png

92% is around a score of 2000

[mini]

I'm sure that a higher percentage of perfect scorers are rejected than individuals from families with $10 mil. incomes. (and take out those students and admissions rates are well lower than 7%).

[tokenadult]

I agree that we know anecdotally that some students with truly "perfect" scores (2400 on the three-section "new" SAT or 36 as an ACT composite) are rejected by their favorite colleges, which in many cases includes Harvard. Thus far what I don't see is any evidence of a current statement from Harvard's admission office that perfect scorers are even tracked separately as a noted category. As mentioned above, the revealed preferences working paper category of "100th percentile" includes scores that are not peak scores. Moreover, the revealed preferences working paper information is almost surely not current for Princeton (I don't believe Princeton practices that form of strategic admissions anymore) and may not be current for Harvard.

[interesteddad]

I'll let you guys Google around Harvard's website for a confirming quote, but here is confirmation of the acceptance rates for students scoring a perfect 800 on the SAT Verbal or SAT Math at Swarthmore College.

It's an interview with Admissions Dean Jim Bock in the Swarthmore Bulletin, the alumni magazine, from 2001. I've seen more recent quotes, but this should suffice as authoritative. Of course, Harvard is even more selective than Swarthmore.

One often-cited number, Scholastic Assessment Test (SAT) scores, is no exception. Bock calls them “useful when used appropriately,” but scoffs at the idea that Swarthmore has a particular target or standard. “Sure, our median is high [1,450 for the Class of 2005]; it reflects the overall quality of our applicant pool. But we may accept students who are well below our median when we see that he or she is way above the average student in a school. These scores can mean different things in different schools, so we work hard to understand their context.”

This year, more than half of all applicants with perfect 800 scores on their SAT verbals were not accepted. Just under half of those with 800 math scores were also not accepted. It’s pretty clear that if SATs were the deciding factor, Swarthmore could admit a class with near-perfect scores. “You wouldn’t need us,” says Bock of his staff. “We could just plug in the numbers.”

Here's the link to the full article:

The Challenge Is to Choose - Swarthmore College Bulletin

[tokenadult]

I'll note here that a student recorded as having a perfect score on the SAT math section (800) may have any score at all on the critical reading section (from 200 to 800), and similarly for a student who has a perfect score on the critical reading section. It's considerably rarer for a student to ace two sections, or all three of the new SAT, than it is for a student to ace one.

http://www.collegeboard.com/prod_dow...al_reading.pdf

http://www.collegeboard.com/prod_dow...athematics.pdf

http://www.collegeboard.com/prod_dow...up_writing.pdf

http://www.collegeboard.com/prod_dow...tile_ranks.pdf

http://www.collegeboard.com/prod_dow...tile_ranks.pdf

Brown University publishes figures for base acceptance rate (and, incidentally, for yield) for students with peak scores on one section or the other, but those figures are subject to the possible interpretation that the peak math scorers who are rejected are those with low critical reading scores, and likewise the other way around.

[cptofthehouse]

Token, you can ask Harvard directly. Don't know if they separate out apps with perfect scores, however. Some colleges truly do not. They may separate out scores but not per student. The admissions directors at some highly selective schools I know are not impressed with the perfect scores at all, and told me their process does not highlight them, and decisions are made without that knowledge. The scores are packaged within thresh holds and SAT2 scores.

[tokenadult]

I was shadowing the Stanford admission officer rather than the Harvard admission officer at last week's Exploring College Options program in my town. The Stanford officer repeatedly said, "It's not a numbers-driven process" when asked various questions of this nature about admission to Stanford. Stanford's website is certainly consistent with that. It's my belief, from having visited Harvard's website, that "perfect scorer" is not even a tracked category in the Harvard admissions process. But I could be proved wrong if someone can find a journalistic quotation from one of the leaders of the Harvard admission officer specifically referring to admission results for perfect scorers in some recent year.

[2boysima]

Harvard tracks perfect scores. Possibly not 2400 total, but definitely individual Math and Verbal scores.

None of the links below give an answer to your specific question on %,....but allows for some "intelligent" inference.

Also...if you trust the self-reported scores on CC admit threads...you could do a non-statistical, but somewhat inferential look at admits and rejects with perfect scores.


2007 - Harvard rejected 1,100 with perfect math SAT (Article by Sam Dillon, NY Times, April 4 2007 http://www.nytimes.com/2007/04/04/ed...yt&oref=slogin


Data somewhat old...from 2001: see table 8 at the end of the study (When Harvard still had early action: 37% probability of getting in regular decision with 1600 SAT, 62% probability of getting in early decision with 1600 SAT) http://www.econ.yale.edu/seminars/st...ery-011107.pdf


And...some inference...Harvard accepted 2058 for the class of 2011. “Nearly 2500 applicants had perfect verbal scores”, “nearly 3,200 applicants had perfect math scores”. A record pool leads to record results

We can be fairly certain that all 2058 admitted did not have either perfect math or verbal scores...just need an educated guess on how many did!

[cptofthehouse]

2boys, I know a number of kids with perfect math scores but unimpressive verbal scores. Now with a third part to the SAT1, it is difficult to come up with a number of kids with a 2400 especially when you know that they can get it at multiple settings. Also, it was my understanding that H and a number of other highly selective schools add in the SAT2 scores to get their relevant number for judging the kid.

[2boysima]

# of kids with perfect math and possibly unimpressive verbal appears to be validated by the applicant #'s at Harvard (for class of 2011)

OP didn't indicate if he was looking for stats for combined perfect score (2400 + SAT II's) or individual perfect scores....I just noted what I had found.

Also...Harvard states very directly that they superscore, so they could have quite a few applicants with a 2400 on the SATI. Harvard College Admissions Office: frequently asked questions

[coolweather]

I notice the parabolic distribution in all categories from the stats posted by tokenadult (post #9).
This is not the pyramid distribution we see in other kinds of competition. If we apply the minimum value of calculus, it looks like students with 780 Math, 790 CR, and 780 WR are more preferred.

Math 800 - 9857
Math 790 - 4447
Math 780 - 2928
Math 770 - 5936
Math 760 - 5703
Math 750 - 6265

CR 800 - 8567
CR 790 - 627
CR 780 - 3167
CR 770 - 4289
CR 760 - 3675
CR 750 - 8041

WR 800 - 4443
WR 790 - 2477
WR 780 - 2000
WR 770 - 2860
WR 760 - 3094
WR 750 - 4940

CR+Math 1600 - 1206
CR+Math 1590 - 534
CR+Math 1580 - 583
CR+Math 1570 - 1028
CR+Math 1560 - 1001
CR+Math 1550 - 1380