LA leads the nation in number of murders
Again from
Nick Denton, this
story on the absolute number of murders per year shows that Los Angeles has claimed that dubious national honor. But note that Detroit leaps off the page with 34 plus murders per 100000 per year. And that brings an important consideration to mind: the more relevant statistic, in my opinion at least, is the
per capita murder rate, a category that Washington DC usually leads in.
If one were enterprising enough, one could use the
Statistical Abstract of the United States along with information on the geography of jurisdictions to create a "murder rate per capita" vs. "% black" or "%Hispanic"
scatter plot. [1] We'd keep the murder rate on the y-axis and the %b/H on the x-axis, as befits their status as dependent and independent variables respectively. With blacks in particular, I'd bet the correlation between "%black" and "murder rate" is substantial, and probably a better predictor than most politically correct indicators of criminality (like degree of urbanization or a spatial variant of the
Gini coefficient).
The point is that demography is statistical destiny. One might have two jurisdictions with similar black/Hispanic population percentages and somewhat different crime rates, but a) they're probably within a fairly small band and b) it's almost a sure bet that they both have a higher crime rate than a locale with a small or nonexistent black/Hispanic population. This might be the explanation that Nick is searching for:
The city's murder rate is, in relation to population, less than half that of Los Angeles or Chicago. One explanation: the New York metropolitan area includes more peaceful suburbs, which fall outside the borders of metropolitan Los Angeles. But still.
Mass Hispanic immigration has changed the demographics of Los Angeles even more than it's changed New York. I'm too busy to look up the demographics now, but I'll do so later today (or one of you readers can do so for me). Who wants to bet that demography = statistical destiny?
Note:
I believe that intensity of policing is probably the second most important variable after demography in determining the crime rate. In a bivariate regression we could theoretically assess the influence of both variables on criminality simultaneously. But policing intensity is far more difficult to quantify than a simple percentage, as the number of police officers per capita is (in my opinion) a poor surrogate for the dedication of said officers. So for the purposes of this analysis, one can consider geographic variation in policing intensity to be the primary source of vertical displacement off the scatter plot's
trend line.
[1] Note that I don't use brown to be synonymous with Hispanic as in "driving while black or brown", for what are (i hope) obvious reasons...