For the Lott Post below let me fix up some of the stats. Using the Wilson Method (I used both that and the traditional method below), if one takes the number of Defensive Gun Uses (DGUs)–13– and compares that to the number of reported incidents in which the gun is fired–1, the 95 percent confidence interval is 1.4 percent to 33.3 percent. This means, assuming that his survey was adequate in all other ways, in looking at the number of incidents of DGUs alone, the only reasonable conclusion one could draw is that the number of DGUs in which one fires a gun falls somewhere between those numbers.
If one takes the number of individuals who fire given any number of DGUs, the confidence interval would be between 2.6 percent and 51.3 percent (I believe this is what Tim Lambert reports). This is with 7 people reporting a DGU and one person reporting they fired a weapon.
Even if one took the results from his first survey (which he doesn’t have anymore), there would be a total of 38 DGUs, with 3 incidents of the guns being fired. His 95 percent confidience interval in that case is 2.7 percent and 20.8 percent.
In his first survey which was eaten by his dog or some other series of explanations one can examine at Tim Lambert’s page, he claims there were 25 DGUs and people have assumed there were 2 episodes of weapons being fired. In that case, the confidence interval was 2.2 percent and 25.0 percent.
This is all done without any weightings, Lott weights the responses which makes the confidence intervals even larger.
What is truly fascinating is that he claims this is an improvement on previous studies. But as I explained previously, he still has fewer DGUs even in the one-year period than did Kleck and Gertz (1995). They report 56 respondent DGUs in their one-year time frame, 68 household DGUs. Lott’s Survey Methodology is inadequate compared to their survey, Kleck and Gertz have a larger sample size, and more DGUs within the 1 year time frame. Of course, Lott seems unaware that they did a one year sample as well as a five year and he attempted to differentiate between his survey and theirs because they used a five year sample. He is either unaware of their 1 year sample (which would be strange because they report it in their paper) or he hopes others are ignorant of it.