Sat Jul 21, 2012
The chart in “America is a Violent Country” has been getting a lot of circulation. Time to follow up with some more data. As several commentators at CT noted, the death rate from assault in the U.S. is not uniform within the country. Unfortunately, state-level and county-level mortality data are not easily available for the time period covered by the previous post—though they do exist, going back to the 1940s. What I have to hand is a decade’s worth of US mortality data courtesy of CDC WONDER covering 1999 to 2009. I extracted the assault deaths according to the same criteria the OECD uses (for the time period in question, ICD-10 codes X85-Y09 and Y87.1). The estimates are adjusted to the 2000 U.S. population, which isn’t identical to the standard OECD adjustment. But the basic comparability should be OK, for our purposes.
First, it’s well-known that there are strong regional differences in the assault death rate in the U.S. by state and region. Here’s what the patterns look like by state from 1999 to 2009.
Yes, I know the perceptual qualities of this figure are, shall we say, not ideal. I had a fit of info-graphic-itis and couldn’t not show you it, because it looks so colorful. It does give you a sense of the range of variation and some of the regional banding. The figure excludes the District of Columbia, which has a much higher death rate but is also a city. Also missing are a few states with small populations and low absolute numbers of assault deaths—Wyoming, North Dakota, Vermont—such that the CDC can’t generate reliable age-adjusted estimates for them. If you want a “small-multiple” view with each state’s time-series shown separately from high to low, here you go.
Although it’s clear that geographical region isn’t everything, those tendencies are immediately apparent if you look at either figure above. Let’s examine that more closely, using the official census regions.
As is well known, the South is more violent than the rest of the country, by some distance. Given the earlier post, the natural thing to do is to put these regional trends into the cross-national comparison and see—for the decade we have, anyway—how these large U.S. regions would fare if they were OECD countries. Again, bear in mind that the age-adjustment is not quite comparable.
Despite their large differences, all of the U.S. regions have higher average rates of death from assault than any of the 24 OECD countries we looked at previously. The placid Northeast comes relatively close to the upper end of the most violent countries in our OECD group.
Finally, there’s the question of racial and ethic incidence of these deaths within the United States. Here are the decade’s trends broken out by the race of the victim, rather than by state or region.
The story here is depressing. Blacks die from assault at more than three times the U.S. average, and between ten and twenty times OECD rates. In the 2000s the average rate of death from assault in the U.S. was about 5.7 per 100,000 but for whites it was 3.6 and for blacks it was over 20. Even 3.6 per 100,000 is still well above the OECD-24 average, which–if we exclude the U.S.–was about 1.1 deaths per 100,000 during the 2000s, with a maximum value of 2.9. An average value of 20 is just astronomical. And this is after a long period of decline in the death rate from assault.