Taking SAT's after you got into college!

[silverturtle]

<p>

Actually, the 60 point thing is for each section

</p>

<p>Oh, I wasn’t aware that the College Board’s Web site is incorrect.</p>

[silverturtle]

<p>

For data, there is also a whole research section of the Collegeboard website with papers about all sorts of stuff that you might want to look at.

</p>

<p>I already had.</p>

<p>Simple truth: higher scores positively correlate with higher abilities at all ranges.</p>

[shravas]

<p>

To be confident that two scores indicate a true difference in ability, the scores must differ by at least the SED times 1.5. For example, SAT verbal and math scores must differ by 60 points (40 x 1.5) in order to indicate true differences of ability.

</p>

<p>[Statistical</a> Definitions](<a href=“http://professionals.collegeboard.com/data-reports-research/sat/definitions]Statistical”>http://professionals.collegeboard.com/data-reports-research/sat/definitions) should make it more clear that the 60 point thing applies to each section (though I have no idea how this even makes sense applied to composite scores…). I guess then my example wasn’t completely correct, but hopefully it should show you that this isn’t really transitive.</p>

<p>And of course SAT scores correlate with ability, and I do think that the average ability of a person who scores an 800 in some subject is greater than the average ability of a person who scores a 790 in the same subject. The thing is that this difference becomes less and less meaningful when you’re comparing smaller and smaller populations, and doesn’t mean much at all when you’re just comparing two peoples.</p>

[minderbender]

<p>For being such a supposedly intelligent person, Silverturtle appears to know very little about statistical applications. It’s apparent that he thinks someone’s SAT score is a perfect reflection of their intelligence, without regard for confidence intervals, standard error of difference, etc. Of course there is a positive correlation between intelligence and SAT score-- but the correlation coefficient certainly isn’t anywhere close to 1. That’s why the CollegeBoard has determined that the confidence interval is 60. I’d suggest you take a stat class, Silverturtle, before making yourself appear like a fool when you say things like:</p>

<p>

The College Board is mistaken

</p>

[silverturtle]

<p>

It’s apparent that he thinks someone’s SAT score is a perfect reflection of their intelligence, without regard for confidence intervals, standard error of difference, etc.

</p>

<p>I do not believe that, nor have I said that.</p>

<p>

Of course there is a positive correlation between intelligence and SAT score-- but the correlation coefficient certainly isn’t anywhere close to 1.

</p>

<p>Of course not.</p>

<p>

That’s why the CollegeBoard has determined that the confidence interval is 60.

</p>

<p>I do not doubt that confidence interval, and I was not referring to that when, in post #16, I wrote that the College Board is mistaken; I was indicating, rather, that the College Board was incorrect in the wording quoted in post #15:</p>

<p>

There must be a difference of at least 60 points between two students’ scores in order for there to be a true difference in ability.

</p>

<p>Instead, the College Board meant what you said: that the confidence interval is 60. Moreover, what that confidence interval refers to is irrelevant with respect to college admissions. Applicants are not pitted one against the other; admissions officers do not attempt to determine who has the higher demonstrated ability as per the College Board’s statistical definitions of standard error. That is, the functional standard must not be (and evidently is not) one in which admissions officers are sufficiently confident that one student has higher abilities than another before they assume that one student has higher abilities than another. The simple positive correlation that I stated earlier is the basis for the standard: yes, a student with a higher score very possibly does not have higher ability (as the College Board’s data prove) than a student with a slightly lower score, but admissions officers, in the face of limited information, must not assume that and must, indeed, assume that the former student is of higher ability than the latter.</p>

[Jersey13]

<p>

One very good test taker (achieved 2380 on an early sitting) was boasting on one of the HYP fora how he re-took the test to get a 2400 (which he did). People chimed in to assail him for his meaningless venture and immature attitude. Whether or not the college agreed is unknown. However he was rejected at the college in question.

If you’re referring to who I think you’re referring to, he was accepted at both of the other HYP schools.</p>

[silverturtle]

<p>^ Are we talking about christiansoldier? Wasn’t he accepted to only Princeton from among those three? Anyhow, I must defend him: he retook seemingly obsessively only because of his significant personal drive.</p>

[minderbender]

<p>In regards to all the first stuff you said…my bad. It was my interpretation that you were just blatantly disregarding statistical methods in favor of a false assumption that higher scores necessarily = higher intelligence. I apologize.</p>

<p>Now, as far as this statement goes:</p>

<p>

admissions officers…must assume that the former student [higher score] is of higher ability than the latter [slightly lower score].

</p>

<p>I must disagree. If the “slightly lower score” falls within the confidence interval of 60 points from the higher scorer, they can’t assume anything. In other words, they “fail to reject the null hypothesis” that student A has any more ability than student B, and, if they are to assume anything, they must assume the students equal. </p>

<p>From my observation it does appear that admissions officers are pretty good at taking this confidence interval into account, which is why HYPS admissions officers don’t just accept 2400s all the way through, which they easily could if they had the inclination.</p>

[Jersey13]

<p>

^ Are we talking about christiansoldier? Wasn’t he accepted to only Princeton from among those three? Anyhow, I must defend him: he retook seemingly obsessively only because of his significant personal drive.

Yeah my mistake. He was accepted at 2/4 of the HYPS schools.</p>

[minderbender]

<p>You’re right, silverturtle. The CollegeBoard website should say:</p>

<p>“There must be a difference of at least 60 points between two students’ scores in order for us to be SIGNIFICANTLY CONFIDENT that there is a true difference in ability.”</p>

<p>instead of:</p>

<p>“There must be a difference of at least 60 points between two students’ scores in order for there to be a true difference in ability”</p>

<p>Sorry for assailing you without warrant.</p>

[silverturtle]

<p>

Sorry for assailing you without warrant.

</p>

<p>It’s alright.</p>

<p>

From my observation it does appear that admissions officers are pretty good at taking this confidence interval into account

</p>

<p>On the contrary, I do not see any evidence of this. Out of every breakdown of acceptance rates over various score intervals, I have seen nothing but percentage increases at each higher interval.</p>

<p>

which is why HYPS admissions officers don’t just accept 2400s all the way through, which they easily could if they had the inclination.

</p>

<p>Do you mean to say that HYPS could accept only 2400ers or all 2400ers? If it’s the former, you have a misconception of the number of 2400ers.</p>

[minderbender]

<p>I meant the latter. HYPS accept roughly half of the 2400 applicants, I believe. But perhaps your right, because if they really considered the confidence interval, they would accept half of 2340 applicants too, which they obviously do not. </p>

<p>Regardless, my most important point still stands:</p>

<p>

If the “slightly lower score” falls within the confidence interval of 60 points from the higher scorer, they can’t assume anything. In other words, they “fail to reject the null hypothesis” that student A has any more ability than student B, and, if they are to assume anything, they must assume the students equal.

</p>

[MSauce]

<p>But what the College Board doesn’t say is how the interval changes after taking multiple tests, which the Colleges do get to take into account.</p>

[silverturtle]

<p>

But perhaps your right, because if they really considered the confidence interval, they would accept half of 2340 applicants too, which they obviously do not.

</p>

<p>I know that I am eating into my own argument with this, but keep in mind that those percentages are not inherently causal.</p>

<p>

Regardless, my most important point still stands

</p>

<p>But there is no evidence that they need to be 95% sure that a student is of demonstrably higher ability before they assume that. If they get two applicants about whom nothing is known save for SAT scores, they have to go with the higher-scoring one every time—no matter the difference. While this situation does not actually occur (both with respect to the one-on-one nature and the SAT’s being the only factor), the mentality should translate to the holistic review process.</p>

[minderbender]

<p>@MSauce- True but then applicants shouldn’t be able to score choice.</p>

<p>If you take the test 4 times and you get to use your best scores, that isn’t really representative of your ability because your highest scores could be way outside your typical range- they’re outliers. Instead of superscoring, applicants should have to average their scores to create a more realistic picture of their ability.</p>

<p>Someone who gets a 2400 on their 7th try (or whatever) because they had a “significant personal drive” doesn’t really reflect their true ability. I’d contend that, at that point, it’s less about ability and more about how well you’ve learned the test, thus diminishing the meaning of the 2400</p>

[silverturtle]

<p>The current system is largely fine, as I doubt that there is even one student in most given years who obtained 2400 after 7+ tries. But your point that some people are better prepared than others (thus reducing the effective validity) is true in some cases, though I cannot think of a good solution. Plus, it is a student’s job to prepare for the test if he or she wants to attend a selective college that considers SAT/ACT scores.</p>

[minderbender]

<p>

there is no evidence that they need to be 95% sure

</p>

<p>How do you know the confidence level is 95%?</p>

<p>

If they get two applicants about whom nothing is known save for SAT scores, they have to go with the higher-scoring one every time

</p>

<p>By your own admission this argument falls. The reality, as you point out, is that they DO get more than SAT scores, and I contend that as long as they are within 60 points of each other, the difference should be considered negligible and other factors should be evaluated. Only when one person scores higher than 60 points above another should his SAT score be given weight in the process. Of course, as you also pointed out, applicants aren’t evaluated on a one vs. one basis. But, for arguments sake…</p>

[silverturtle]

<p>

How do you know the confidence level is 95%?

</p>

<p>Maybe I misrecalled something that I had read awhile ago. Anyhow, the actual number is irrelevant to my point, other than the fact that it was above 50%.</p>

<p>

Only when one person scores higher than 60 points above another should his SAT score be given weight in the process. Of course, as you also pointed out, applicants aren’t evaluated on a one vs. one basis.

</p>

<p>But the 60-point rule breaks down as well when applied to situations that are not one vs. one.</p>

[minderbender]

<p>

But the 60-point rule breaks down as well when applied to situations that are not one vs. one.

</p>

<p>admittedly, yes.
I disagree with this though:</p>

<p>

The current system is largely fine

</p>

<p>When you take into account socioeconomic factors, wealthy students are likely score higher than poor students. Because the SAT (IMO) is as much about how well you know the test as how much you ACTUALLY know, it helps to get resources like SAT tutors, classes, books, etc. For people who don’t have the funds to do this, the test isn’t actually being administered on an equal playing field. Not to mention the cost of taking the test multiple times.</p>

[silverturtle]

<p>

When you take into account socioeconomic factors, wealthy students are likely score higher than poor students.

</p>

<p>No one can prove to what extent that is causal.</p>

<p>

Because the SAT (IMO) is as much about how well you know the test as how much you ACTUALLY know

</p>

<p>Well, the SAT is not a knowledge-based test; it is an aptitude test. Also, familiarizing oneself with the format is, as I indicated, the duty of a prospective applicant to schools that consider the test.</p>

<p>

it helps to get resources like SAT tutors, classes, books, etc.

</p>

<p>The most effective resources are the Official SAT Study Guide and Direct Hits; these are quite feasible for almost everyone. Preparatory classes and tutors are largely useless for top scorers.</p>