<p>No one is reading what anyone else is saying. And try as you might debate til you're blue in the face not a single one of you is going to convince me that this discussion is important.</p>
<p>
[quote]
Yes, a higher SAT score is better just in the same way that a lower golf score is better. But having a higher SAT score from someone else does not make you a "better" person.
[/quote]
That's more or less what I was trying to say. Sorry if that wasn't clear enough. I am just saying that I will <em>never</em> believe that s.o. who score 800 on the SAT math section is in any way more intelligent/mathematically able/whatever than someone who score 750. These 50 points difference convert to two false answers. They may have been caused by carelessness or whatever. Period.</p>
<p>
[quote]
I just knew at least one person wouldn't get the sarcasm there...not surprisingly, it's the same poster who probably "doesn't get" a lot of things in life.
[/quote]
Oh really, I won't comment on this. I feared something like that; why can't some people discuss properly without going off the track?</p>
<p>That said, I still have problems with your MIT-is-smarter-than-me-so-they-are-right-anyway attitude. Don't you see that this is a weak argumentation? What if MIT decides to kill their students? Do you still believe that "MIT has good reasons for doing so since they are so much smarter than you"? Doesn't make sense.
Besides, I like MIT's admission policy. And I agree with Ben Jones when it comes to SAT scores :P</p>
<p>Speaking of MIT omnicience - did anyone in the admissions office ever take responsibility for the "bulk rate postage rejections" fiasco last year?</p>
<p>"Besides, I like MIT's admission policy. And I agree with Ben Jones when it comes to SAT scores "</p>
<p>At the end of the day that's it isn't it. You either prefer the MIT or Caltech model of admissions. Neither is inherently right or wrong. For numerous reasons found throughout a multitude of threads my vote is for MIT's.</p>
<p>
[QUOTE]
That's more or less what I was trying to say. Sorry if that wasn't clear enough. I am just saying that I will <em>never</em> believe that s.o. who score 800 on the SAT math section is in any way more intelligent/mathematically able/whatever than someone who score 750.
[/QUOTE]
</p>
<p>Then write that instead. Although you'd still be wrong. Just not by as much. </p>
<p>
[QUOTE]
That said, I still have problems with your MIT-is-smarter-than-me-so-they-are-right-anyway attitude. Don't you see that this is a weak argumentation? What if MIT decides to kill their students? Do you still believe that "MIT has good reasons for doing so since they are so much smarter than you"? Doesn't make sense.
[/QUOTE]
</p>
<p>You still don't get it. That's okay. Maybe in a couple years.</p>
<p>
[QUOTE]
Besides, I like MIT's admission policy. And I agree with Ben Jones when it comes to SAT scores :P
[/QUOTE]
</p>
<p>That's even better justification for my argument that MIT shouldn't take advice from anyone criticizing their admissions. </p>
<p>
</p>
<p>If this was refered to me, you probably didn't understand what I wrote. That's okay. Maybe in a few years, when you've graduated...</p>
<p>TMiike, the only comment that I directed towards you regarded your rudeness.</p>
<p>However, I'm glad to know you think you're smarter than me. Quick, what was your SAT score? We can even prove it!</p>
<p>
Ugh, agreed.</p>
<p>I haven't been able to get on the site since yesterday morning, and in the interim have somewhat lost interest in this debate. </p>
<p>I will say that in Ben's hypothetical situation on the previous page, I will get $0 because I refuse to choose -- Ben J. says that there's no difference between students who scored 700 in math and students who scored 800 when it comes to MIT freshman math grades. If I wanted to predict which student would get a higher GPA at MIT, I'd want information other than test scores.</p>
<p>I'd also like to highlight this excellent point by CountingDown:
[quote]
My question for chewing on: How much of this debate over differentiation among the top 1-2% of scores has been caused by changes to the testing instrument? Has lowering the ceiling on the test forced colleges to use subjective criteria to distringuish the top 1-2%, since test scores across the applicant pool are so consistently high?
If high scores were more exceptional in MIT's applicant pool, you might expect them to be of greater importance, I guess. But the applicant pool already has very high scores, so differentiating between them on the basis of very small score differences seems odd to me.</p>
<p>Incidentally, I think one of my bigger problems with the SAT and other standardized tests is that a lot of kids with high scores seem to view those scores as reflective of more than the tests themselves claim to be. It's like a lot of 19th-century science -- let's find some scale that puts rich white men at the top, because obviously we know rich white men are at the top. And so smart kids who score well on the SAT put too much value on the test, because it confirms their views about the world (ie, that they're amazing). </p>
<p>And, hey, I've been there -- I took the ACT in eighth grade and got a great score, and read The Bell Curve and felt smug and superior. But over the course of the past nine years, I've come to realize that the skills necessary for success on standardized tests are just not the same skills that are required for success in college/graduate school/science, and that makes them lose a great deal of value for me. I mean, those of you who are at MIT/Caltech, how many tests have you taken in college that were at all like the SAT? Single-answer multiple choice? Please.</p>
<p>Mollie -- well said, about how the SAT differs from my topology final. But I'd point out that no apparent similarity between the evaluation procedures is needed for performance on them to correlate. That article you posted some days ago (which I did read in its entirety, pace pebbles) has a good survey of the various evidence that most metrics of cognitive performance measure the same thing. Surprising as it may seem, ability on the trivial cognitive tasks given to you on the SAT correlates well with ability at vastly more complicated tasks. (Just look at you.)</p>
<p>As for your cited claim of Ben J. about 700s and 800s having the same freshman MIT grades, I have two answers. One is that I'd like to see the data -- not because I don't believe Ben, which I do, but because it would be fascinating to see some hard numbers on this much-debated issue. (And were the same grades achieved in the same difficulty courses? I don't have to tell you that an A in 18.01 is not the same as an A in 18.014.) The other answer is that this is not so surprising if the committee looked at applications as a whole and required "compensation" for below-average test scores. To put it simply, this statistic, taken at face value, says nothing about how SAT scores correlate with ability in the general population, or even the applicant population (the one of interest here), but how they correlate in the very nonrandomly selected group MIT admits. That's an interesting but entirely separate question.</p>
<p>Oh, and if you say you wouldn't choose, you're either not telling the truth or haven't been a grad student long enough to learn to like money. If you choose neither, you get $0 with probability 1. If you choose either one, the probability that you get $100,000 is definitely bigger than 0. And you can't have any more information, greedy animal.</p>
<p>The College Board is well aware that the SAT test taken multiple times (w/o special time accomodations") by random students having completed 10 years of schooling - will roughly correlate with overall IQ</p>
<p>This simple reality conflicted with a central tenet of the politically correct world of academia which required all to believe that all individuals are in effect equally intelligent, and that any differences must be a result of some unfairness, oppression, victimization, racism, or poor basic educational opportunities - invariably connected to some lack of governmental funding - then it is very clear why the College Board was required by the early 1990's to then pretend the word "aptitude" (used for 50 years) really had nothing to do with the term "SAT" and claim that the abbreviation SAT actually (and conveniently) stands for nothing</p>
<p>My principles are worth more to me than dollars. </p>
<p>(Not in real life. Just with hypothetical money and hypothetical principle. ;))</p>
<p>A big problem I have is that, while it may be true that SAT scores correlate with performance on complex tasks, it's also true that SAT scores correlate highly with socioeconomic status. It's been many years since I read The Bell Curve, but one thing I do recall is that basically the best predictor of IQ scores is parental socioeconomic status -- you could predict who's going to be "successful" using wealth instead of SAT scores if you wanted, and you'd be right just as often.</p>
<p>That suggests, at least to me, that the test is not measuring what we'd like it to measure with the purity that we'd like it to measure with.</p>
<p>But anyway, I think we are in more-or-less agreement here:
[quote]
To put it simply, this statistic, taken at face value, says nothing about how SAT scores correlate with ability in the general population, or even the applicant population (the one of interest here), but how they correlate in the very nonrandomly selected group MIT admits.
[/quote]
To me, the accepted student population is a population screened by test score and selected by other facets of the application. This results in a group of students with varying test scores, but statistically equivalent outcomes in a later trial. Sure, they're nonrandomly selected on the whole, but with regard to test scores, they can be modeled as being selected randomly from among the 700+ math scorers.</p>
<p>I guess I think the admitted student population is more interesting simply because we have more data on them and their outcomes.</p>
<p>(Um, if this makes no sense, I'm sorry? It's finals week and my neurons aren't all firing together, as it were.)</p>
<p>
[quote]
Sure, they're nonrandomly selected on the whole, but with regard to test scores, they can be modeled as being selected randomly from among the 700+ math scorers.
[/quote]
</p>
<p>That's such offensive begging-the-question ;-) By *assuming<a href="or%20modeling%20them%20as%20if">/i</a> they are random sample of the 700+ applicants, you have already presupposed that whatever criteria are used to distinguish students above 700+ don't correlate with SAT. But that doesn't follow, of course, from the fact that SAT's aren't explicitly used in the 700+ range. But the argument isn't just invalid -- its conclusion also happens to be false. There's a pretty plot in that much-cited Avery et. al. paper "A Revealed Preference Ranking of U.S. Colleges and Universities" showing that probability of admission rises monotonically at MIT between 1400 and 1600 (old style).</p>
<p>I agree that the admitted population is easier to study.</p>
<p>On the broader and more interesting question of wealth and intelligence, I agree that it seems hard to disentangle wealth and IQ. That can be either because smarts (as measured by IQ) get you wealth or wealth gets you smarts (as measured by IQ). The Bell Curve certainly failed miserably to give a convincing argument that the causation went mostly in their preferred direction ("smarts" --> wealth) but it seems to me that it's still an open question. There is some evidence that high-IQ children of poor parents do vastly better economically than low-IQ children of poor parents, so IQ does often seem a reasonably good predictor of which way your social mobility is likely to go.</p>
<p>The socio-economic argument conveniently (and understandably) overlooks the well documented fact that when upper middle class black students even when put into the best public school systems or the best private schools - still end up with very large gaps in SAT scores - and this trend remains the same even when the immersion is for the student's entire schooling period for example from grades 1 to 12</p>
<p>The socio-economic argument (favored by the College Board in their literature) ends up largely as a device to avoid certain very very unpolitically correct debates - although the (newly devised) "low expectations" line of reasoning is (in part) now being use to head off that debate - should it ever occur</p>
<p>Can we please get a citation for that?</p>
<p>Yeah, the model is imperfect. (I had a really good one once, but I cannot find the spreadsheet, so I must have deleted it.) It is based on the actual admissions data for last year, though, which I would assume is a better data set than what Avery et al. had in front of them. The admissions data is pretty broad (separates people in 50-point intervals), so it's not as good an instrument as I'd like, but it works pretty well.</p>
<p>Of course there are also those schools that have a dip in their aaccptance rate at "great but not perfect" scores. </p>
<p>E.g., a 2300 might have more trouble getting in than does a 2200 but less trouble than a 2400.</p>
<p>Let's see. "The 2300 has all the ego problems of a 2400 but not the potential..." This is a tough one.</p>
<p>this from a minority based blog</p>
<p>However the evidence of the black/vs median SAT/(despite better schooling) GAP is everywhere -and it shows up in the supporting briefs related to the Grutter and related Sup Ct cases </p>
<p>Note the usual "low expectations" explanation</p>
<p>alluded to by CNN as far back as yr 2000 - same issue</p>
<p>Mollie: single answer multiple choice? That comment was so cruel. I'd completely forgotten about the existence of such tests. =P</p>
<p>"The 2300 has all the ego problems of a 2400 but not the potential..." </p>
<p>Or vice-versa?</p>