To keep this from turning into a marathon debate, I will just make two points, as briefly as I can:
(1) Murray’s numbers, as quoted by Robert, suggest that the SAT II does a moderately better job of predicting college grades than the SAT I. That works as an argument (though it’s not the argument Murray makes) for eliminating the SAT — if we accept Robert’s proposition that “the goal of the admissions process is to figure out which former factors [high-school grades, test scores, etc.], when combined, have the highest possible correlation with the latter [college grades].” I do not accept that proposition, for about half a dozen reasons, though this is an argument for another day.
(2) Demonstrating that SAT I and SAT II are highly similar in predicting college grades does not disprove the fact that they measure different things. It might be true, for example, that the French achievement test and the German achievement test predict college grades equally well, but they test entirely different things.
OK, three points:
(3) I still think Murray’s psychosocial argument is silly.