Wednesday, July 1, 2009

Ricci vs. DeStafano

The Supreme Court handed down its decision on the controversial Ricci vs. DeStafano case a few days ago, which reminded me of an article that provides textbook case of bad statistical reasoning.

The passage in question is:

"Independent experts, too, warned against making too much of the three-judge panel's decision to rely on the lower court ruling. That's what appeals courts often do.

The 2nd District Court of Appeals ruled on 2,859 cases last year. Of those, 381 decisions — 15 percent — came with a signed opinion, the clerk's office said."

The problem is that no one was making a fuss just because Sotomayor and her colleagues didn't give a signed opinion, they were arguing it was odd for a case this controversial, one taken up by the Supreme Court and expected to have a 5-4 decision.

The real statistic people are interested in is, conditional on a case going on to the Supreme Court, having a 5-4 decision (and, as we now know, generating 4 different opinions) how often does a lower court feel there's no need to give a signed opinon? I suspect it's higher than 15% though I have no idea how much.

No comments:

Post a Comment