Pim van Meurs writes:
Of course there will always be an uncertainty in the findings that's
why there are statistical error bounds and statistical significance
bounds. However in case of gun ownership at city level Kleck showed
the causal direction of gun ownership increasing the use of guns in
crimes like robbery and assault.
David Veal writes:
Kleck also reports in Point Blank (among other studies) studies by
Murray (state level) and Bordua (county level) which found no causal
relationship between gun ownership (measured directly) and the rate
of gun homicide.
Murray's study looked at gun laws at the state level, not gun
ownership. His measure of gun ownership divided the US into only four
regions, which is not enough for meaningful analysis.
Bordua found a positive significant correlation between gun ownership
and gun murder rates. His interpretation of this was what that gun
murders caused gun ownership, but his reasoning is faulty and the
alternative explanation (that gun ownership causes gun murder) is at
least as plausible.
Pim has a bad tendency to report out of Kleck's works only those
results which support his thesis that gun ownership increases the
lethality of crime.
I've noticed that Kleck tends to subject evidence supportive of the
"pro-control" thesis to a fierce scrutiny, but is much more accepting
of "pro-gun" evidence. See, for example, his discussion of Kennesaw
in "Point Blank".