This article first appeared in the St. Louis Beacon, Feb. 7, 2010 - Several years ago, I attended a meeting to address crime in American cities that included police officials from cities around the country.
At one point during a discussion of why homicide rates are so much higher in some cities than others, a police official from Baltimore declared, “You can’t compare us to San Diego!” What did he mean?
The Baltimore police official objected to comparing his city’s homicide problem with San Diego’s because he is saddled with crime-producing conditions unlike those his San Diego colleagues have to face. Not only is Baltimore’s homicide rate higher than San Diego’s, so is its poverty rate, unemployment rate, and the incidence of nearly other social condition correlated with crime.
And the police have little control over those conditions; they’re stuck dealing with the crime that results. I suspect the Baltimore police official believes his job would be much easier in San Diego.
Not long after that meeting on city crime problems, I joined with my colleagues Alfred Blumstein of Carnegie Mellon University and Robert Friedmann of Georgia State University to, in effect, put the Baltimore police official’s message into practice.
Since 2002, we have produced yearly rankings of American cities according to their homicide rates, after adjusting for differences across cities in poverty, median income, unemployment and other conditions that are strongly associated with city homicide rates and over which the police exert little control. Basically, the procedure re-ranks the cities based on their expected homicide rates assuming they all have the same level of poverty, unemployment, etc.
We recently published the homicide rankings for 63 large cities based on crime data for 2008. Some cities, like Detroit, rank near the top of the list before adjusting the data for differences in crime-producing factors. After adjustment, Detroit’s rank dropped dramatically. This means that Detroit’s homicide rate in 2008 was quite a bit lower than would be expected based on its dire socio-economic circumstances. Other cities rose in rank after adjusting for socio-economic differences, meaning that their homicide rates were higher than expected based on their relatively benign conditions.
What about St. Louis? In years passed, St. Louis’ homicide rank fell after the statistical adjustment, but not in 2008. St. Louis ranked first among the 63 cities in homicide before adjusting for socioeconomic differences and first after taking into account its comparatively grim circumstances.
It turns out that the city’s homicide rate was higher in 2008 than during the previous several years, and almost twice as great as the rate in 2003, when homicide dropped to a level not seen in decades. Socio-economic conditions do not change this rapidly and therefore cannot explain St. Louis’s recent homicide rise. Even the current economic downturn is an unlikely culprit because the homicide rate actually turned down in 2009 as the crisis deepened.
My colleagues and I used the most recent comparative data available to calculate city homicide rankings for the first six months of 2009. Although no longer at the top of the list, St. Louis placed high in both the unadjusted and adjusted rankings, running neck-and-neck with Baltimore. Meanwhile, San Diego ranked far lower, both before and after accounting for differences in crime-producing conditions.
Meaningful comparisons of city crime problems, especially if they are used to evaluate police performance, should not stop with the raw crime rates. They should also tell us something about how those rates are affected by economic and social conditions for which the police are not responsible but make their job more or less difficult. In other words, compare Baltimore’s crime problems with St. Louis’, not San Diego’s.
Richard Rosenfeld is Curators Professor of Criminology and Criminal Justice at the University of Missouri-St. Louis and President of the American Society of Criminology. The homicide rankings and a description of the statistical adjustment can be found at https://www.cjgsu.net/initiatives/HomRates-PR-2010-01-21.htm .