<p>People with 760's get accepted at the same rate that people with 800's get rejected. At this level, numbers never drive decisions. Trust me. I'm very in touch with my subconscious. :D</p>
<p>Arrrg! The downside is that we've created a generation of grade-grubbing self-conscious over-stressed perfectionist freaks!</p>
<p>Very helpful with the blog and benjones, not just for this score, but it puts into perspective (from someone who knows) that highly selective colleges look at more than just scores</p>
<p>People with 760's get accepted at the same rate that people with 800's get rejected. At this level, numbers never drive decisions. Trust me. I'm very in touch with my subconscious. </p>
<p>Does not it mean that if the people with 760 get accepted with the rate of 1%, people with 800's will get rejected with 1% rate and accepted with 99%? :)
I can clearly see now that you ARE very much in touch with your subconscious.</p>
<p>Sorry, benjones, could not resist.</p>
<p>Well, the only plausible solution is 50% admission/rejection rate for both categories.</p>
<p>I am pretty sure that Ben meant "people with 760's get accepted at the same rate that people with 800's get accepted."</p>
<p>Yes, I know, but it is not what he actually wrote :)</p>
<p>Remember, I did apologize.</p>
<p>No, I meant that for every kid with a 760 who gets in, a different kid with an 800 doesn't get in. Hundreds of kids with perfect scores across the board don't get into MIT. Hundreds of kids with 760's do get in. The point was simply that numbers don't drive decisions. :-)</p>
<p>Doesn't your clarification imply that it DOES make a difference whether one got a 760 or an 800? That 760s and 800s are NOT accepted at the same rates. You just feel it's not an important difference and one shouldn't obsess over the small improvement an 800 makes?</p>
<p>The Avery, et al (2004) paper on Revealed Preference Ranking
<a href="http://papers.ssrn.com/sol3/papers.cfm?abstract_id=601105%5B/url%5D">http://papers.ssrn.com/sol3/papers.cfm?abstract_id=601105</a></p>
<p>seemed to have pretty clear evidence that probability of admission to MIT increased steadily and monotonically with SAT I scores. I can't say much about the SAT II.</p>
<p>But the Avery paper only gave percentiles -- and anything above a 1480 (old) is 99th percentile (Collegeboard</a> percentile table). So that isn't really a useful way to discriminate between MIT applicants.</p>
<p>For what it's worth, yesterday I saw the SAT score range vs. percent admitted for the class of 2010 (which will be up on MyMIT when the new site comes out at the beginning of September). I believe students with a 700-750 math were admitted at a rate of ~18%, while students with a 750-800 math were admitted at a rate of ~22%. That's probably not a statistically significant difference.</p>
<p>Hey Ben, is a 690 on the writing portion of the SAT1 worth retaking? My son is more than fine with all his other scores. :-)</p>
<p>Well molliebatmit, if the acceptance rate for 760s were almost always lower than for 800s for the last decade or two, my guess is that a test <strong>would</strong> show stat significance even if the percentages were very close. But the relevant question is not stat sig; it's substantive significance and that's the point that benjones (presumably) was making.</p>
<p>However, his clarification was so pointed [number of 760s accepted ~ number of 800s rejected] that I began to wonder what the real numbers were.</p>
<p>Moreover, a stats test won't determine the "true" significance of the gap even if it were really small. That depends on the relative value of time and test-effort assigned by the applicant vs. perceived gain at the <em>margin.</em> Consider that if someone applies to 6 schools all of which show an 18 vs 22 percent difference in acceptance rates, it might be worth it if retakes are "easy."</p>
<p>Anyway, I don't disagree with benjones (much), but I am curious.</p>
<p>Haha, I'm a biologist -- we only think in terms of statistical significance, and we can't do any of the harder math. :)</p>
<p>Nice paper, Not quite old, thanks for sharing it with us.
There is one point though: what do they mean by 100%?
As far as I know, there is 99% and 99+%, do they extrapolate to 100?
Do they consider 99+ = 100?
Well, it might be somewhere in the paper, I have not read it through yet.
Poor kids applying to Princeton though, P(99) = P(95) < P (92-94)!
So much for trying to raise your SAT.
At least it paints a good meritocratic picture for MIT, there is some math biologists can relate to!</p>
<p>Those are percentiles. You can't be in the 100th percentile, because that would mean that you are better than 100% of all the test takers, which is impossible.</p>
<p>That is exactly what I meant. There are data points on the graphs for 100 percentiles. Yes, I know I should have spelled out "percentiles".</p>