A few weeks ago, the College Board announced that SAT participation is up but scores continue to decline. Intuitively that makes sense; when limited participation by the best students becomes something more like universal, scores should decline. So I was not inclined to take William Bennett seriously when he argued that record low SAT scores reflect a failure of the educational system; I stuck with the consensus that the lower scores reflect the universalization of college education.
As usual, Bennett is full of bluster and bad argument, but it turns out he might be right this time. Comparing the 1996 and 2011 scores shows that the high school performance of test-takers has (in principle) increased, not declined, even as the number of test-takers increased. In 1996, 44% of test-stakers were in the top 20% of their high school class. That year, 35% of test-takers had an high school GPA at or above the A- range, and the mean GPA was 3.20. This year's class looks quite a bit better: 62% are in the top fifth of their class; 44% had A- or better GPAs and the mean GPA was 3.34.
So it seems that students taking the SAT are, according to their high schools, getting better. But an assessment independent of the schools seems to suggest otherwise, at least in this case. I think a more serious look at what the SAT says about the quality of high school graduates is in order.