it might be the case that the methodology in that study is significantly more reliable than that of other studies on this subject. If you are aware of an argument to that effect, I’d be interested to hear it.
As you pointed out, the study you’re citing is gated, so I can’t really evaluate it. However, based on the portions you quoted, I don’t see any reason to believe that either study is more or less methodologically sound than the other. However, I also don’t see any reason to believe that the two studies actually contradict each other, despite first appearances. The study I cited evaluates national-level crime stats, whereas yours evaluates household-level victimization stats. Both studies’ conclusions could easily be true simultaneously.
As noted in your citation:
None of the studies can prove causation and none have completely eliminated the possibility that the association might be entirely due to reverse causation
My intuition is that reverse causation is at play. That may sound like a cop-out, but keep in mind, I also argued reverse causation for my own citation, despite that being against my political interests, so I’m not just privileging my own position here. In the case of your citation, the intuitive case for reverse causation sounds even more convincing to me: people who live in high-crime neighborhoods are much more likely to decide to buy a firearm, in response to their local crime rates—I can attest to this from personal anecdotal experience.
Since prevalence of violent crime varies neighborhood-by-neighborhood, while gun control laws typically vary only nation-by-nation, there could very well be a positive correlation at the neighborhood (or household) level, while simultaneously having a negative correlation at the national level… especially in light of the fact that your study’s abstract concedes that its results are dominated by a single nation (the US).
The (informal) null hypothesis—that gun control laws have no significant causal effect, either positive or negative, on violent crime rates, would seem to explain both studies’ results equally well, whereas assuming the existence of (non-reverse) causation, either positive or negative, would require either one or the other of the studies’ conclusions to be wrong somehow. Parsimony would therefore seem to recommend this hypothesis.
Thank you for your well-researched response.
As you pointed out, the study you’re citing is gated, so I can’t really evaluate it. However, based on the portions you quoted, I don’t see any reason to believe that either study is more or less methodologically sound than the other. However, I also don’t see any reason to believe that the two studies actually contradict each other, despite first appearances. The study I cited evaluates national-level crime stats, whereas yours evaluates household-level victimization stats. Both studies’ conclusions could easily be true simultaneously.
As noted in your citation:
My intuition is that reverse causation is at play. That may sound like a cop-out, but keep in mind, I also argued reverse causation for my own citation, despite that being against my political interests, so I’m not just privileging my own position here. In the case of your citation, the intuitive case for reverse causation sounds even more convincing to me: people who live in high-crime neighborhoods are much more likely to decide to buy a firearm, in response to their local crime rates—I can attest to this from personal anecdotal experience.
Since prevalence of violent crime varies neighborhood-by-neighborhood, while gun control laws typically vary only nation-by-nation, there could very well be a positive correlation at the neighborhood (or household) level, while simultaneously having a negative correlation at the national level… especially in light of the fact that your study’s abstract concedes that its results are dominated by a single nation (the US).
The (informal) null hypothesis—that gun control laws have no significant causal effect, either positive or negative, on violent crime rates, would seem to explain both studies’ results equally well, whereas assuming the existence of (non-reverse) causation, either positive or negative, would require either one or the other of the studies’ conclusions to be wrong somehow. Parsimony would therefore seem to recommend this hypothesis.