probably the "publish or perish" thing - the poor fellow needed a qualifying paper using an original data set.
Some of my concerns with his approach:
1. he does not establish a hypothetical causal connection between gun ownership at large and homicide rate - stated differently, not every citizen has an equal probability of committing a homicide, so an increase in the gun supply may not translate into increased incidence of homicide; most likely, homicides are committed by a hardcore criminal element so the properly hypothesized causal connection should run between supply of guns in that segmet of society (which I suspect has been constant for some time) and incidence of homicide - for that reasosn alone the paper seems to be a futile exercise in bean counting to fish for significant correletions.
2. He says he used gun ownership reported in GSS to validate the subscription as an indicator of gun ownership. Why the fuck didn't he use the GSS data?
these are my 2 cents, but then I am just a mere sociologist not affiliated with the U of Chicago dept of Econ.
wojtek