Obliterated
“Diversity” doesn’t improve education. Not even a little bita.

By NR’s John J. Miller & Ramesh Ponnuru
April 4, 2001 5:25 p.m.

he recent split decision on the University of Michigan's race-based admissions — one judge allowed them for undergraduates, another barred them for the law school — is another example of racial preferences standing on their last leg. And today the National Association of Scholars gave it a good, hard kick to the shin.

There appears to be a single argument for racial preferences in admissions that the courts will now consider: The argument that a diverse student body improves education for all students because it is diverse. This is the claim the Michigan lawyers advanced, and Judge Patrick J. Duggan bought it last December in the undergraduate case. "The University Defendants have presented this Court with solid evidence regarding the educational benefits that flow from a racially and ethnically diverse student body," he wrote in his opinion.

Except that the evidence isn't at all solid. The school relied on a study by one of its own, Patricia Gurin, head of the psychology department, to "prove" that diversity aids learning. Gurin predictably found exactly what she wanted to find.

But if she really had been looking, she would have seen nothing at all. The authors of the NAS report, California Association of Scholars director Thomas E. Wood and SUNY-Albany statistician Malcolm J. Sherman, use the same data as Gurin. They reveal how she selectively ignored information devastating to her cause. Wood and Sherman get right to the point in describing what Gurin did: "It is quite possible that the University of Michigan has been deliberately misrepresenting the data. The alternative hypothesis, of course, is that the University's own researchers are misrepresenting the data out of simple methodological confusion and error. Neither hypothesis does credit to the University."

Wood and Sherman conclude that there is no correlation between racial diversity and educational outcomes.

For more on the NAS report, in the full splendor of its regression coefficients and t-statistics, go here.