I think there's a fundamental problem in the way that we use stats to try to find correlation between guns and crime, in that the predominant way that gun ownership is talked about is as a number of guns per capita. To me you can draw some information of that, but it's not the most useful metric. If you've got two small towns and one has a couple enthusiasts with 100 guns each, and the other has 50 people averaging a couple guns each, you wouldn't expect them to have similar gun usage patterns, but they'd appear as almost identical communities under guns per capita metrics.
This is borne out in US statistics; while the number of guns in the US per capita continues to climb, the number of gun homicides peaked in 1993 (at 7.0 per capita) and has consistently fallen since then to 3.6 per capita. Households with guns correlates much more closely: its recent peak was also in 1993, at 53% of US households having at least 1 gun. That has fallen to 36%, with peaks and valleys that loosely correlate to the homicide rate. There are still too many other factors for this to be really strong correlation across countries, but it seems to be the far more important statistic when looking at gun violence within a single country.
|