That’s a pretty bold claim. So you’re saying the only difference in SAT score is due to SAT prep? I would doubt this would even be the case, because if all things are held equal except SAT score, that means income is held equal as well. I didn’t know that the only way to get an SAT score was to buy one. I thought the score I earned was because I answered questions correctly, but that’s obviously crazy.</p>
<p>
A much smaller correlation I would bet. The student with the higher SAT score has demonstrated that they perform better under pressure compared to the lower scoring student. That’s a pretty important ability.</p>
<p>I’d point out that the reality is that there is no such thing as “all things being equal” – every application is going to have some sort of variation. Different LOR’s, different EC’s, different essays, different interests. So the college ad com is going to make its decision based on the whole package.</p>
<p>“All things being equal, why not accept the student with the lower score?”</p>
<p>If you’re an admissions officer, and you have to choose exactly one of two people, and for some reason all you know are the SAT scores, who do you pick? Person A with a 1900, or Person B with a 2100? If all you have to go on is an SAT score, and you have to assume one is smarter than the other, why wouldn’t you assume the person with the higher SAT score is smarter? It’s not infallible, the person with the 1900 could be smarter/better student than the person with the 2100, but you would expect that it’s more likely that the person with the 2100 is smarter than the person with the 1900.</p>
<p>Anyway, I didn’t read the linked paper because it’s scanned and hard to read, but I’m assuming the data says something like there are several times as many students in the top 50-point bracket, but the average is still the same?</p>
<p>That means standard deviation went up… To make it the way it was before, replace some of the easy questions with even easier questions (I don’t know how this is possible on the Math section… Maybe instead replace some of the medium questions with easy ones?) and replace some of the hard questions with harder questions. </p>
<p>The SAT people probably doesn’t want to do that. The more reasonable it is to get a 2300+ score, the more people will take the SAT 5 times to try to get that score. If you make it pretty much impossible, then people won’t bother. Maybe they lose 10K people who get that 2300+ on the first try, but they gain 50K+ people who take the SAT several times to try to get that. </p>
<p>Since all of that was pretty much common sense, and I’m sure it was mentioned in the last 19 pages, if anyone wants to do me a favor and direct me to the post # or quote some posts which were responses to those claims, I would appreciate it. Thanks.</p>
I am saying that since such prep is allowed and is common, the college has absolutely no way of knowing one way or another. That’s why the test scores are of such little value – they generally provide more reliable information about the size of the parents’ bank account than about student “ability”.</p>
<p>This seems like a good time to point out that the person with the higher score may not have actually answered more questions correctly, even at the top of the spectrum. For example, if somebody got one question wrong on the entire October 2009 test (presuming 12 essay also) and it happened to be in the Math section, their score would have been 2370. If another person on the same test got 3 questions wrong on the reading and everything else correct, they would have received a 2380. In fact, they could have received a perfect score while still getting twice as many questions wrong.</p>
It’s not really that hard to read, but yes, you have the gist of it.</p>
<p>
You’re right. It’s common sense, that’s probably why nobody mentioned it in detail, although I did mention making questions “trickier” at the harder end of the test. And there was sigificant discussion of the old test verses the new tests in here somewhere.</p>
<p>Because we don’t work for the college board we discuss the test that exists. For elite college admissions I certainly believe it is less than ideal, but it may well serve admissions across the spectrum, I haven’t thought about that in detail.</p>
<p>My whole original thought was really pretty pedestrian, I’m surprised at what it spawned. Let me explain it with a slightly exagerrated example.</p>
<ol>
<li>Assume there are 10 schools that everyone wants to attend (I am assuming this.Please no debate on this, it’s a simplification).</li>
<li>These schools have a total enrollment of 16,000.</li>
<li>You have a test with a top score of 2400.</li>
<li>If you miss one question your score drops to 2350.</li>
<li>300 kids get 2400.</li>
<li>40,000 kids get 2350 or above.</li>
</ol>
<p>How precise is that test for determining who gets into those 10 schools? I know that somewhere on an old thread somebody actually did this analysis for some SAT or other and it isn’t quite as exagerrated as I put it here. But I was just commenting it seemed a little less bunched at the top in the old report. To tell the truth, purely statistically that may not even be true. But I’m tired of thinking about it.</p>
<p>^^^
Maybe I should have called this thread “SAT - A Blunter tool for top-end admissions now or in 1973?”</p>
<p>At first blush, that’s how it seemed to me. But I fully admit that I didn’t go through and do an analysis grouping number of examinees per question missed. And I fully admit I don’t intend to.</p>
<p>Bovertine, where did you get 40,000 get 2350+? That’s not anywhere near close enough to look at any sort of simplified example. The real number was 1,996 in 2009. </p>
<p>Actually looking at my post, 2300 was way off too… Replace 2300 with 2200.</p>
<p>On height… actually, there was a time when Princeton in particular did favor taller candidates. I’ve forgotten which history of the Ivies I found that nugget in, but it was fairly well documented.
I think the ‘if two candidates were identical except for a small difference in SATs’ argument of little weight for two reasons. First - ad coms don’t work by comparing pairs. Second - there aren’t really any ‘identical’ candidates. Even if the numbers - test scores, GPA,rank are all the same, there will be very telling differences in terms of actual courses, ECs, personality, essay, background, finances, gender, etc.</p>
Actually I just made that number up. For my original post, I used this, and was really only talking about the growth in the 750 and above math from 1973 to 2008.</p>
<p>But you are right. I didn’t realize there was so little overlap between top scores in the different tests. That cuts it down significantly, and makes it a better tool. Probably because of the addition of the writing section.</p>
<p>My original post about the growth still stands though.</p>
<p>I took the PSAT once as a junior and the SAT as a junior. I took a practice test before each to familiarize myself with test, and used sparknotes to learn some grammar rules. I didn’t do amazing, but I’m still within range of the Ivy League universities/others.</p>
<p>I might retake, but it doesn’t look like there will be an opportunity again for me until like next December.</p>
<p>I am not going to accept it blindly as the number of 2009 college applicants in these ranges. Does this take into account score choice? If not, the numbers in the upper ranges would certainly increase under such a scenario. What exactly does this table mean? Highest single sitting score? Scores from the latest exam a student took? And please provide a link, not just an opinion.</p>
<p>I admit that my numbers are very high. What is the exact numbers of college applicants with multiple sitting score choice in each range? I have yet to see a definitive number supported by an explanation of what the data shows.</p>
<p>And once again, my original statement about the 73 and 2008 reports stands, which was all I really wrote about in the first place.</p>
<p>As far as I know, this is how most colleges consider an applicant’s SAT scores-
The applicant takes the test as many times as they like. This may be 5 or more times. The college then considers their top score in each section, and adds them together. Let me call this “method 1”, since apparently score choice doesn’t apply.</p>
<p>Is this true?</p>
<p>The next question is, now do they compile this table? </p>
<p>If they just take one sitting of an exam, or the latest sitting of an exam for each student, and compile these numbers to put in this table, and since the numbers from method 1 represent the very best possible composite scores for each student over multiple administrations, the numbers from method 1 will certainly be higher than the numbers from method 2.</p>
<p>Yes, I see now that you meant to say “superscoring.”</p>
<p>As for the College Board data, my best guess is that they tabulated the highest single-sitting score for each test-taker – either that or the most recent single-sitting score for each test-taker.</p>