A few years ago, the National Research Council reviewed the evidence on firearms and crime and concluded:
There is no credible evidence that “right-to-carry” laws, which allow qualified adults to carry concealed handguns, either decrease or increase violent crime.
I do not find Marvel and Moody’s conclusions plausible and they are not supported
by the results of their regressions. The results are all over the
place. Some crimes are up, some are down. So they aggregate using a
cost for each crime. But then they it’s up in more states than it’s
down. So they take a population-weighted average, which even then is
only down if you take a long enough time frame. And even then, if you
exclude Florida, it’s up. I don’t think that their data allows any
conclusions to be drawn.
Ayres and Donohue’s reply has not been published yet, but you see a preprint here:
Armed with the weight of a single new regression for each of seven crime categories,
Carlisle Moody and Thomas Marvell (2008) conclude their remarkable paper, “The
Debate on Shall-Issue Laws,” stating that they are “confident” that “the evidence, such as
it is, seems to support the hypothesis that the shall-issue law is generally beneficial with
respect to its overall long run effect on crime” (292). The paper is remarkable because
the evidence Moody and Marvell present thoroughly undermines (yet again) the
conclusion that RTC laws “generally” have any beneficial effect on crime.
Moody and Marvell essentially make four points, which simultaneously grow in the level of both their ambition and error. … With an appropriate quality-adjustment,
however, most of the “supportive” studies on their list would be deemed to have little or
no current value. … Moody and Marvell turn their gaze to Ayres and Donohue (2003a), which in 119
pages arrayed an enormous amount of information raising doubts about the more guns,
less crime hypothesis. Moody and Marvell ignore virtually all of this discussion and
instead challenge a single table, … their own estimates powerfully undercut their
suggestion that RTC laws are generally beneficial. Moody and Marvell then labor to
refute their own findings by once again unwisely extrapolating linear trends beyond the
period of their data.