I really don't have a dog in the John Lott controversy, but fellow political scientist Mark A. Kleiman has a round-up of the leftist perspective (and Julan Sanchez has some thoughts too). At the very least, the whole "Mary Rosh" business seems at once both silly and undignified for a scholar. As far as the rest goes, I'm much more interested in arguments over the econometrics and evidence supporting the main argument of the book, rather than ruminations over the 98% figure (98% of "brandishments" did not result in use) which is mostly tangential to the main argument (although I will grant that Lott's defenses of the 98% figure are specious at best). I will say that the "predictive power" test of an econometric model is generally not accepted in the social sciences; rather, we seek to maximize explanatory power. Without having analyzed the data myself, though, I can't state to my satisfaction whether the model is appropriate or not.

Michelle Malkin weighs in, and she's unimpressed with Lott's defense.