In other words, in a hypothetical state with no guns and no religion, we’d expect a suicide rate of 17 (per 100,000), give or take a few points. In a state where everyone had guns, and no one practiced religion, we’d expect a suicide rate of 39. If everyone were religious, but no guns: 11. Everyone religious, everyone owns a gun: 21. From a political perspective, there’s really something for both the left and right to like here.

When you add divorces per 1,000 men, it is still significantly related to suicide rates in a state, and population density is even less related.

Of course, to really get this right, we’d probably need a few more variables (although with the few I messed around — income, foreclosure rate, race, male/female ratio — the religion/guns axis remains significant by any commonly accepted measure). We’d also probably want to break down our data among different religions. Perhaps most importantly, to get around the substantial ecological fallacy problem, we would want to do better than the state level (unfortunately, county-level data on firearms possession doesn’t exist, although some rough proxies are available). But as noted above, studies performed at the individual level seem to reach roughly the same conclusions.

We must also remember that all we’re really finding here are correlations. Making the leap to causation is just that: a leap.