I've had to move this off the main page, because for some reason I have trouble posting any one html page that's more than about 45 K. I'd bumped up to that limit, and the server was truncating the page by cutting off the ending! So I've had to split off some subheading.

Any attempt to derive useful causal conclusions from comparisons between American states is difficult, simply because of radical differences -- population density, income, culture, and many other factors. Getting the same conclusions from comparisons between different nations is far more difficult.

Canada has a tenth the population of the US -- relevant if raw numbers, rather than rates, are used. To be fair, Canada also has lower rates.

Canada is also more rural than the U.S. (It has one-tenth the population density of the US, although to be fair it has a lot more unoccupied land which "pads" that figure; its population lives largely in the southernmost fraction of the nation. The US has unoccupied land, too: in Arizona, 113,000 square miles, a majority of the population lives in just two cites; same with Nevada, 110,000 square miles.. But while the US has lots of unpopulated land, Canada has... well, lots and lots, what with the ice [outlawed in Arizona]).

Don't cite me figures for "urbanized" populations -- yes, the firgures for US and Canada are in the 70-80% range, but Canada counts as urban an area with 1,000 people, while the US requires 2,500, a considerable difference. Rural areas in the US are relatively crime-free: 85% of U.S. counties reported no (as in zero) youth homicides in 1997; in any given year, about a third of them will report no homicides at all. In large expanses of the non-urban US, homicide is almost unknown. I know of one Arizona county (I represented the sheriff in court), nearly as large as Connecticut, which averaged one homicide and three robberies per year.)

Just to make things even more complicated, Canadian homicide rates don't follow the American pattern, where urban rates are far higher than rural ones. In 2002, Toronto actually had a homicide rate below the national average, as did Calgary. Montreal and Vancouver did have rates about 1/6 above the average -- but that's a tiny disparity, compared to the US one, where it's not uncommon for large cities to have several times the national rate, and for rural areas to have nothing mentionable.

If Moore wanted to find places where you can leave the door unlocked, he didn't need to leave the U.S.. I do that now (in fact the door is open behind me as I type. I did it when I lived in the Washington DC metropolitan area, at a time when it was the homicide capital of the U.S. . I lived in a suburb where crime was close to unknown, and locking the doors when you were home was just an obstacle to going outside. And, no, it wasn't a fancy gated community populated by millionaires. A hispanic family lived across the street, a fellow government worker next to me, a Navy vet diagonally across from me.

Another aspect, suggested in the Canadian Journal of Criminology, is the spread between low and high incomes (not absolute poverty, or median income, but the spread.). The article concludes: "When Canadian provinces and U.S. states are considered together, local levels of income inequality appear to be sufficient to account for the two countries' radically different national homicide rates." The same might well (alright, this is a guess on my part) explain part of the traditional rural-urban crime rate differences. My experience is that small towns have less of an income spread than large ones. Most folks are middle class, with fewer very rich or very poor. Go to an urban area like Washington DC, where there is no industry, and there are a lot more very rich and a lot more comparatively poor. Lots of executives and janitors, as opposed to a population of miners, shopowners, and factory workers.

While we're on international comparisons, How about (again, using raw numbers:

Germany: 202

Canada: 47

Japan: 19

US: 1,586

Those are drug arrests (in thousands) for 2002, taken from interpol.