Jackson Jambalaya

The Wall Street Journal published two articles last week on the latest CQ Press crime rankings. The December 4 article reported such rankings can be flawed due to cities different classifications of criminal activity and geographical area. A city may receive an artificial boost because only the city is included while another one is pushed further down the list because it includes its "metro area" or surrounding suburbs (St. Louis was a good example of this discrepancy.).

"When CQ Press's City Crime Rankings were published last week, they prompted headlines nationwide, particularly in cities that ranked near the top or bottom. The rankings showed that Detroit had the country's highest crime rate among the largest cities, while El Paso was the least dangerous.

But critics say such comparisons are meaningless. Cities differ markedly in how tightly their borders are drawn around their inner core, and how much of outlying areas are included within city limits can have a big effect on crime levels. The rankings are based only on reported crime, and reporting rates vary, criminologists say. Also, personal safety varies tremendously depending on lifestyle, income and neighborhood.

"Knowing the city someone lives in tells you next to nothing about their risk for crime," says Richard Rosenfeld, a criminologist at the University of Missouri-St. Louis and a longtime critic of the rankings...."

"The CQ Press ranking takes into account a city's per-capita incidence of six serious crimes: murder, rape, robbery, aggravated assault, burglary and motor-vehicle theft. For each crime, the editors calculate how each city's rate differs from the national average. These six percentage differences are averaged to produce an overall score, and then cities are ranked by that score, which, roughly speaking, represents the difference between a city's rate of serious crime and the national rate.

This method produces some quirks. A city with below-average levels in five of the categories but a very high level of, say, car theft could score poorly. Also, if crime is down nationally, a city whose crime rate falls more slowly could see its crime ranking rise. And there is very little difference near the middle of the rankings, meaning differences of 20 or 30 ranking positions might not be statistically significant..."

The author wrote an additional blog post on this subject. A group of criminologists formed one group, Improving Crime Data that claims it uses a more statistically accurate approach. Its rankings "The rankings reflect (1) the raw rates per 100,000 population and (2) the rates adjusted for differences across the cities in socio-economic factors. The adjusted rates result in important changes in the ranking for several cities".