<p>SATs or HS GPAs?</p>
<p>GPA, but it depends on the class rank and the course load. There is a difference between a 3.6 at a public school in the top 35% taking no challenging courses and a 3.4 in the top 10% at a private school taking 4 accelerated courses.</p>
<p>A hard worker who doesn't test well beats out a slacker who can be the straight A student but doesn't want to be, because I know the hard worker will get those A's in college and make the most out of the resources, whereas the slacker is going to skip class and rely on his intellect to pass the tests.</p>
<p>The correlation between SAT 75th percentile and graduation percent at a college is very high (about .9). I am not sure if average HS gpa and graduation percent would be higher, but it couldn't be much higher. The nice thing about SATs is that they are standardized. A 3.5 gpa at one high school can be very different from a 3.5 gpa at another high school, or even within the same high school. There is a lot of "noise" in the high school gpa system. That's why SATs are better predictors, I think.</p>
<p>SATs and HS gpa are both measures of motivation but SAT does a better job of capturing true ability. </p>
<p>Ironically, there is a big difference between predicting college performance for a particular student and predicting AVERAGE college performance for the entire freshman class. The success of individual students is much harder to predict even though the AVERAGE performance is the sum of individual performances.</p>
<p>The best predictor of college performance is COLLEGE gpa in the first semester freshman year but even that isn't very accurate.</p>
<p>What do you mean by college performance? Graduating? Getting a high gpa?</p>
<p>I'd suggest college gpa as well - since that captures motivation better than one test. Consider Thomas Edison, how well did he do on standardized testing...does it matter? I'd say not.</p>
<p>huskem55-
Very interesting article. According to one of my relatives who works in higher ed, the real relationship between SAT and performance in college is very high. The correlation that is sometimes calculated between SAT scores and college gpa is artificially low because students with higher SAT scores attend more competitive colleges, select more demanding majors, and enroll in more difficult courses. High SAT scorers self-select a more challenging course of study, so the gpa is less than it would be if they had majored in an easier subject, taken easier courses, and taken fewer credit hours.</p>
<p>When you control for these things, SATs are an excellent predictor of performance.</p>
<p>Then there is the "unit of analysis" problem. There is a quirk in statistical calculations. Statistical calculations using individual scores are artificially lower than statistical calculations using average scores. There is more "error" with individual scores. This can lead to the rather confusing fact that it is hard to predict the performance of an individual but much easier to predict the performance of a group of individuals. Depending on whether you are focusing on individuals or groups, you can simultaneously conclude that SATs (and everything else for that matter) DO NOT predict performance very well and that SATs DO predict performance very well.</p>
<p>From the point of view of college presidents who are interested in improving the performance of the entire freshman class, SATs are an excellent predictor of performance.</p>