[tpaine]

AndyTheBear:
So the Donohue study boils down to "you can't be absolutely sure about everything Lott did".


It boils down to a lot more: Lott's results are highly sensitive to minor changes in specification, and they don't hold out of sample. Hence they are highly suspect.
-Tradi-

Such a truism can be asserted about any statistical results, imo. Hence, Donohues results are also 'highy suspect'.
Thus, we have a silly circular argument.
-tpaine-
_________________________________

This is true, but at least Lott had some common sense!
-AtB-

I disagree. Donohue uses Lott's methodology (as well as some variations of it), so if Donohue doesn't have common sense, neither does Lott.
77 -tradi-

Then, as you said earlier, we must use our common sense & principles to make conclusions as to which 'methology' would decrease crime, an armed citizenry, or an unarmed one.
What does your 'sense' tell you? That this is a moot point?

[traditionalist]

Such a truism can be asserted about any statistical results, imo.

No, that is not true. There are many statistical results in the social sciences that are robust and have continued to hold out of sample. The underpreformance of IPOs is one that comes to mind immediately. There are many others.

Hence, Donohues results are also 'highy suspect'.

Yes. Thus, we have a silly circular argument.

Not at all. It meanse that the overall effect on crime of CCW is too small to be discerned by conventional econometric analysis.