<p>Columbia 2002, A contributing factor is also the many different type of intelligences. Some students may have great mathematical intelligence while others may have articulate intelligence, and others analytical intelligence, other linguistic intelligence. I feel that this is why the SAT should not be weighted as heavy as ECs, Recommendations, essay, etc. etc. </p>
<p>But indeed NYC seems to attract many loser who as you say "throw a hail mary" aiming for Columbia even though it's a long shot. I find that NYC is a contributing factor to this, as many people who chose NYC for the college experience will apply for Columbia regardless and once rejected attend NYU. However, I feel like Columbia students are more the type of student who are willing to speak their mind and have a passion to change, a real sense of leadership. It seems as HYP students are molded to be business oriented, while Columbia students are more concerned with various issues around the world. Perhaps this is why Columbia owns when it comes to The Nobel Prize.</p>
<p>
[QUOTE]
But then, you certainly have an interaction effect between income and the SAT, as well as ethnicity and the SAT-- though then you have issues with collinearity, assuming you do an ordinal variable for ethnicity...
[/QUOTE]
</p>
<p>This is very true. Take the LSAT for example, many practicing lawyers offer personal tutoring service that guarantee dramatic increases (for an extremely expensive rate.) Students who can not afford this do not see increases in practice tests as dramatic as those students'.</p>
<p>Since Columbia College's selectivity last year was 8.9% (when SEAS is subtracted) I think it's pretty clear Columbia could build a class with higher SAT scores if its adcom wanted to.</p>
<p>
[quote]
For one, we have to be willing to assume that the SAT is an accurate measure of intelligence. That's unlikely at best. Then, taking into account the confidence interval of 80, and you're really in trouble saying that the 1500 group is somehow smarter.
[/quote]
SAT scores correlate with the results of IQ tests equally as well as results of IQ tests correlate with each other. I read that somewhere, so a little googling could probably turn up a cite. In that sense, it may be as good a proxy for intelligence as any other.</p>
<p>Where I draw the line is about a 1500. If you can get 1500, you are smarter than the test - and the differences between that and a 1600 are pretty much just careless errors or the result of distractions. Just like IQ tests are pretty well meaningless outside of 3 standard deviations, there is very little information contained in someone getting a 1520 vs a 1580. Below a 1500, though, it's always seemed to me as though there is a real difference in cognitive reasoning the farther you get from 1500.</p>
<p>If you want to distinguish between really smart people, though, for my money the LSAT is about as good as it gets. That test is HARD (and thus a lot of fun, for me at least), and if you can get a 170 or above it really says something powerful about your reasoning abilities.</p>
<p>The problem with ALL of these tests, however is the preparation industry creating a lot of noise in the system. </p>
<p>The only reason that I'm more likely to accept an IQ test over the SAT is the fact that few, if any, actually prepare for IQ tests when they're actually supposed to be taken (childhood.) How do we control for all the prep and actually find a way to use the SAT in some sort of intelligence modeling?</p>
<p>I do see the correlation papers thrown around, but I find most of them to be based on pretty poor modeling.</p>
<p>I don't believe in the idea of SAT being a measurement of IQ. The IQ should be on a constant level and shouldn't change dramatically over a, especially short, period of time.</p>
<p>However I had been able to improve my critical reading score from 500's into 630 and I didn't score more because the last sections left me too tired. In case of SAT, the practice makes perfect. In essence, I should be able to go and take the test without any preparation and get the same results as after an extensive one; it doesn't work that way. Kaplan and Princeton Review would have gone bankrupt if this was true.</p>
<p>Furthermore, the intelligence part fails when it comes down to the international students who don't have English as a first language. The chances of getting a great score are again limited by the vocabulary or knowledge of idioms; I would have got 80 on the MC part of writing if I didn't make 4 errors in idioms. The foreigners pass the intelligence-measuring aspect of the test by being able to take the test in a foreign language, showing they are far smarter than their 1200's, 1300's or 1600's.</p>
<p>For all of that, the SAT should not be a decisive factor in admission.</p>
<p>
[quote]
The problem with ALL of these tests, however is the preparation industry creating a lot of noise in the system. </p>
<p>The only reason that I'm more likely to accept an IQ test over the SAT is the fact that few, if any, actually prepare for IQ tests when they're actually supposed to be taken (childhood.) How do we control for all the prep and actually find a way to use the SAT in some sort of intelligence modeling?
[/quote]
this makes sense and I accept your argument. Along similar lines, I generally looked down upon those who had extensive preparation for the SAT in high school and in college as not really being deserving of their scores. On the third hand, I'm not quite sure whether to consider "taking a bunch of practice tests yourself" as quite getting the same prep as having Kaplan rigorously force you to think within their framework. The only SAT prep I ever did was a few practice analogies sections (where have you gone, SAT analogies?), and I think all it did was calm my nerves on them a little bit.</p>
<p>I maintain that the SAT has plenty of use as a general tool for admissions, however, and that absent noise (which you point out is an increasingly confounding factor), it tells a lot about the cognitive abilities of the test-taker. Or at least, it does so better than any other standardized test you could devise.</p>