I agree with Glenn Reynolds that the debate over the John Lott coding error question is seemingly interminable.

But I say, the hell with "econometrics," "running regressions," "coding errors," and "clustering adjustments!"

I care only about two simple statistics. Since September 11, 2001:

1. the number of guns in this country has gone up;

2. the number of Americans killed in terrorist attacks has gone down.

Q.E.D.

===========================

Additional note: Even assuming the truth of everything Tim Lambert says, (that the 98% figure is wrong, "Mary Rosh" is an imaginary person, and errors by Lott were demonstrated in the 120 page paper), considerations of statistical data have no bearing on basic constitutional rights. If statistics showed that countries with more censorship have fewer rapes, would that justify restrictions on free speech?

The Second Amendment does not live or die by statistics and counterstatistics about crime rates (much less whether or not there have been verified sightings of the mythical Mary Rosh).