Here's another way to conduct this study.
Dependent variable = per-capita firearm murder victims"
Independent variables:
1. Amount of private security
2. Amount of public security (police)
3. Change in police tactics
4. Abortions (Stephen Levitt study)
5. Lead poisoning (
JW Reyes).
6. Total firearms in US
7. etc.
With the inclusion of more independent variables, one can (perhaps) get a firmer grasp on how the dependent variable is influenced.
(there's also the problem of
endogeneity. For example, is the "increase in total firearms" causing a "decrease in firearms murders," or is the "decrease in firearms murders" causing the "increase in total firearms"? In other words, people may still erroneously perceive that firearms murders are increasing, thus will purchase more guns--even though firearms murders are decreasing.