ArchPundit continues to ponder on Lott's amateurish surveys. One further point you might like to consider: If the surveyors in Lott's 2002 survey were aware that it was vital for Lott to reproduce his low firing rate result, then that introduces a strong source of interviewer bias. People often answer questions with the answer that they think will most please the interviewer. If Lott's interviewer's gave the impression, either consciously or unconsciously that they did not want to hear the person that they had fired the gun, then the results will be biased downwards. This effect can be reduced by training interviewers so as not to give away what answer is desired, but Lott, of course, used untrained interviewers. He wasted his money. Again.
Tom Spencer still thinks that the idea of 1/2 a person firing a gun is fishy and is down on the whole idea of weighting surveys. Weighting surveys really is legitimate, it's just that doing it an sample size that is far too small gives ridiculous results.
John Quiggin is not persuaded by David Gross that there ever was a survey.