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.