JOHN LOTT UPDATE….Glenn Reynolds wrote a short post about John Lott yesterday, and this has prompted Lott to respond on his website to the charges by Ayres and Donohue that his more-guns-less-crime hypothesis is based on coding errors caused by faulty data entry (it’s the 8/20/03 entry).
Unfortunately, as Tim Lambert points out, Lott evades the issue. He complains that Ayres and Donohue erred by saying they had downloaded his data from his website. In fact, he fedexed it to them. He says the corrections only apply to a few hundred data points out of 70,000. He says he’s unconvinced that A&D showed that more guns cause more crime. He says they misstate a claim about a different study. And finally he says that their language is intemperate.
The one thing he doesn’t do is something that he could presumably do with little more than a touch of a button: rerun his very own regressions on the corrected data. Ayres and Donohue claim that if you do that, Lott’s results go away and concealed carry laws have no effect on crime rates. (Actually, A&D say that crime rates go up but the effect is small, so statistically they are saying that concealed carry laws have no effect on crime.)
In their original article, Ayres and Donohue weren’t talking about anyone else’s studies, and they weren’t doing a literature review. They were specifically finding fault with the regressions that Lott ran. If he thinks they’re wrong, all he has to do is rerun the regressions and publish the results.
Why won’t he do that?