<p>^^
Even though we have little data, that comment may be right.
The least squares regression line of SAT score on GPA leaves a line like this:
.0000579*SAT+3.69= Predicted GPA
R-sq is only .015 meaning only 1.5% of the variation is explained by the least squares regression line. However this will probably increase if there are more responses. Also, the subject tests is not AS feasible considering these scores tend to be inflated because people only take subject tests in areas they are extremely proficient (aka 800 in Chinese means zilch if you were born speaking Chinese) It would only be statistically responsible to have everyone take the same SAT especially since the GPAs can't be normed for.</p>
<p>Math grade (high school junior): 88%
Math SAT: 560</p>
<p>English grade: 93%
English SAT: CR 720 Writing 800</p>
<p>That means my maths score is actually right where my maths grade (as opposed to my overall GPA) predicts it, and right around where my teacher expected. My English scores are actually higher than my English grade predicts- but the 800 is what my grade for the last two years would have predicted.</p>
<p>The least squares regression line of SAT score on GPA leaves a line like this:
.0000455*SAT+3.74= Predicted GPA
R-sq is only .0061 meaning only .61% of the variation is explained by the least squares regression line. Ehh... I'm getting discouraged. I'm assuming when you guys calculate unweighted thats A=4 B=3.. etc in only academic classes.</p>