Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
In "Point Blank", Kleck analyzed NCVS data and found that while 38% of people who used any means of self-protection against robbers were injured in the encounter, only 17% (the lowest for any means for self-protection) of people who use a gun for self-defence against robbers were injured. Kleck claimed that this showed that guns were the most effective means for avoiding injury.
In their critique of "Point Blank" Alba and Messner point out the flaw in Kleck's reasoning -- the evidence from the NCVS is equally well explained if injury makes victims less likely to use guns. Kleck dismisses this alternative explanation as a "speculation". This is a rather odd argument. The evidence shows a correlation between lack of injury and with-gun defence. Kleck's explanation (gun defence prevents injury) and the alternative (injury prevents gun defence) are equally speculative.
Later versions of the NCVS provide a way to resolve the issue. They ask questions to find out whether the injury occured before or after the self-protective actions. Differences in injury rates before the victim took any self protection action will reflect the effect injury has on which methods of self-protection will be chosen, while differences in injury rates after the self-protection action reflect the effect self-protection methods have on injury.
In "Armed", Kleck analyzes the new NCVS data. Over all means of self-protection against robbers, 34% were injured before they did anything and 7% after. For gun defenders, 13% were injured before they used the gun, and 8% after. The results seem quite clear: gun defence does nothing to reduce your chances of injury, but injury prevents victims from using guns for defence.
Remarkably, in his discussion of these results, while Kleck concedes that defensive gun use "does not appear to be as uniquely safe as data from earlier NCVS data suggested", he does not take notice of the implications of the large difference in the pre-self-protection injury rates. Instead he argues that the data shows that gun defenders face less favourable circumstances because they were more likely to face multiple and gun-armed attackers, ignoring the injury data suggesting that they face more favourable circumstances.
