Folks on the left have criticized the new Title IX guidelines because, they say, surveys are a bad way to measure compliance.

Many, like Greg, have answered: "if you have enough of an interest to play a varsity sport, you also should be able to answer and return a survey."

At first blush, that seems reasonable enough... but consider the following story from Brennan's column this morning:

Cary Groth, the athletics director at Nevada-Reno, was another of the 15 Title IX commissioners. She recounted a story from the commission hearings that she said was "staggering." The Illinois high school athletic association said it sent out surveys asking girls if they would be interested in playing volleyball. The surveys came back showing little or no interest in the sport. Lacking confidence in their own abilities, perhaps, and never having played the sport before, the girls by a resounding margin said, no, they didn't have any interest in volleyball.

But the athletic association, seeking more opportunities for female athletes, took it upon itself nonetheless to start volleyball for high school girls in Illinois. And, wouldn't you know, volleyball became one of the state's most popular girls sports, with more than 300 high school teams in the state.

"If they had judged by the survey," Groth said, "they would have thought there was no interest."

This is what people mean when they say that current interest levels reflected in surveys may themselves be a legacy of discrimination. This point isn't unique to Title IX -- it's just a species of the chicken-and-egg problems that we always face in discussions of affirmative action (and Title IX is, after all, an affirmative action program).

It's a debate that raises some complicated theoretical and empirical questions about which reasonable people can disagree. But simply saying "if no one answers the survey, there's no interest" isn't enough.

UPDATE: Greg responds, saying there's a difference between surveying high school girls and surveying college women. He also takes a look at the shifting burden of proof, a change that may be more important than anything about surveys.