Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
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".
