> Total nonsense. Figure 1 does not mean that SAT/ACT scores are predictive of academic achievement! Without seeing the variance you cannot make this statement.

> Total nonsense. Figure 1 does not mean that SAT/ACT scores are predictive of academic achievement! Without seeing the variance you cannot make this statement. At all. The fact that there is a linear relationship means nothing statistically. You are putting words in their mouth from that quote. They didn't use that Figure to claim the test scores are predictive of academic achievement. They specifically say it is evidence of a linear relationship. They argue in the next paragraph that they are and use regression analysis to make their case. > So we're explaining less than 1/5th of the variance. This makes SAT scores basically worthless, this account for almost none of the real world performance of people, never mind how redundant they are with GPAs. Compare that to high school GPA, which explains ~9%. The test optional movement implicitly puts more weight on high school grades by putting less weight on test scores. I'm not saying their analysis is perfect, but your arguments aren't that strong... At the end of the day, the analysis focuses on a subset of people who are going to Dartmouth. This is a highly filtered subsample of the total population. If you took a random sampling of 18 year olds, gave them the SAT, and then enrolled them at Dartmouth, then I hypothesize you would see a significantly stronger relationship between test scores and college grades than what is observed in their data.