<p>taxguy,</p>
<p>Obviously there is not NECESSARILY any difference in GPA performance “between high, mid and low” SAT scores. One can always give anecdotal evidence and compare one student who had a bad day on the SATs (and therefore “overachieved” in GPA) versus someone else who had high SATs but got involved in drugs or just stopped going to class (and therefore had a low GPA). </p>
<p>I am not sure what you are saying about in your experience the SATs having a “55-45” predictive rate. For 1000 students with SATs in the range of 1500 to 1600 compared with 1000 students with SATs in the range of 1200 to 1300, are you saying that there is only a 55 percent chance that the former group will have higher average GPAs (at the same schools)?</p>
<p>Certainly within a small range of scores, SAT will not be very predictive. A student with a 1350 SAT compared to a student with a 1380 is not likely to be statistically different in terms of GPA. But I cannot believe that when comparing a student with an SAT score of 1000 with one who scored 1500 that there is not some VERY significant difference of predicted GPA (at the same school, in the same major, etc.)</p>
<p>“If you check the college board web site statistics, they note that the correlation is about 60%.” “I guess colleges feel that this extra 5-10% over simply guessing justifies the hassle incurred by the students.” What is the reference for this number? (I could not easily locate it on <a href=“http://www.collegeboard.com%5B/url%5D”>www.collegeboard.com</a>) Are you saying that SAT score can account for 60 percent of the scatter in GPA when a linear model is assumed (why would the relationship necessarily be linear, anyway)? Or what does your statement mean?</p>
<p>Thanks.</p>