A critical piece of data missing is the subset (anywhere from 0 - 2150) of those guns pre-9/11 screening would have found. To do this right, you could take the number of guns found during the period mentioned above, calculate the guns-to-passengers ratio, throw in a fudge factor to account for the number of passengers who didn't show for their flights, and make a predictive apples-to-apples judgment to compare TSA versus pre-9/11.

I suspect the TSA did this type of statistical analysis, but they didn't get the answer they wanted.