Actually, SPOOFE, there is substantial evidence to suggest that even the “low-end” figure of 700,000 defensive gun uses per year cited by Kleck is a gross overestimate, more than ten times higher than the real rate. This article in the Journal of Criminal Law and Criminology discusses Gary Kleck’s and Marc Gertz’s original 1995 article, “Armed Resistance to Crime: the Prevalence and Nature of Self-Defense with a Gun”, where the 2.5 million estimate is put forward. The article I linked to contains a critique of the earlier article by David Hemenway, a reply by Kleck and Gertz, and a summary evaluation by Tom Smith; so it’s quite a nice survey overall of the different arguments on defensive gun use.
The main objection of Kleck and Gertz’s critics is that their analysis contains substantial overestimation biases that they don’t adequately correct for. First, the “social desirability response” bias—i.e., the tendency of someone who believes in gun use for self-defense (to the extent of purchasing a gun for self-defense) to assume in an uncertain situation that he/she has in fact succeeded in using a gun for self-defense, thus validating his/her own wisdom and good judgement. (This is sort of the same effect that causes the vast majority of people to believe that they’re good drivers, although accident statistics show that many in fact are not.)
The second problem is a statistical effect of the fact that the authors are attempting to estimate the occurrence of an extremely rare event—even if there were three million defensive gun uses per year, in a nation of some 300 million people, that’s still only 1% of individuals. That means that for every hundred individuals reporting their defensive gun use in the past year, 99 run the risk of giving a false positive (by lying, misremembering, being mistaken, whatever) while only one runs any risk of giving a false negative. That’s a huge bias in favor of accidental overestimation, which when extrapolated to the final numbers can give a serious inflation of the actual rate of occurrence. Simply taking the average of your “high-end” and “low-end” estimates as produced by such a flawed methodology doesn’t come close to guaranteeing that you’ll eliminate these biases.
Specific objections to the details of the Kleck/Gertz analysis can be found in the linked article (as well as other sources such as Tim Lambert’s discussions of DGU figures), or I can summarize more of them here if anybody’s interested. All of this, of course, does not mean that privately owned guns can’t be useful for self-defense. It does mean that estimates of actual defensive gun use in the millions or hundreds of thousands per year in the U.S. should be regarded very cautiously.