Silly Crime Statistics

I am sure that most have seen the new rankings of most violent cities. Detroit and St. Louis top the list once again. As we all know, crime statistics not the best way to make cross-city comparisons. Rick Rosenfeld regularly sends letters to the editor when these crime rankings are released. I posted this on my blog a while back, but it is worth repeating here.

Silly crime rankings versus real numbers

Some figures are a lot more meaningful than others.
Author: Richard Rosenfeld

Americans love to compare themselves to one another. We rank our automobiles, body fat, annual income and our children's standardized test scores. We also rank the crime rates of our communities.Morgan Quitno Press, a statistical analysis firm based in Lawrence, Kan., recently released its annual rankings of city crime. Residents and officials from cities with low ("safe") rankings greeted the news with pleasure; those with high ("dangerous") rankings attacked the rankings as unfair or faulty.

A recent Post-Dispatch editorial cautioned readers against taking such comparisons seriously. But the best response to the crime rankings is not to ignore them but to understand how they are made and what their limitations are. That way, when they appear again -- and they will -- residents will be able to accept or reject them on objective grounds, not simply because their city scored high or low. Here are fundamental questions to consider:*

On what crimes are the rankings based?
Do they mix different types of crimes? Are crimes that differ in seriousness given the same weight? Morgan Quitno ranked cities on a crime index containing offenses as diverse as murder, rape and motor vehicle theft. Although all are serious crimes, most people would rather have their car stolen than be the victim of a violent assault. Yet, Morgan Quitno gave each crime in its index the same weight. Crime comparisons should be based on crimes of equal seriousness.*

How thoroughly are the crimes measured?
Crime rankings are based on offenses reported to local police departments. The FBI then compiles this information in crime statistics for the nation, states, counties and cities. Some offenses, such as homicide, are extremely well reported. Others, such as rape and other assaults, are frequently not reported to the police and never make it into the crime statistics. Police departments also differ in the way they classify and record the crimes reported to them.Rankings, therefore, may be influenced by differences in crime reporting and recording practices that have nothing to do with the amount of crime residents actually experience. Crime rankings should not be based on offenses, such as rape and assault, that are subject to large differences in reporting and recording from city to city. Morgan Quitno's crime index includes both of these offenses.*

Do rankings account for variations among neighborhoods within cities?
All cities consist of a small number of high-crime areas and a much larger number of low-crime areas. Crime rates for different neighborhoods within a single city typically differ more than than differences between cities. Truly useful data would break out crime rates for the various areas of a given city. Such information may be available from local police departments, often on Web sites. Overall city crime rankings say nothing about where in a city crime is high or low.*

How has crime changed over time?
People intending to live or do business in a city need to know something about trends over time, not simply the level of crime at a single moment. Morgan Quitno ranked St. Louis among the nation's most "dangerous" cities in 2004, yet St. Louis' homicide rate has been cut in half over the previous 10 years. By definition, snapshot accounts say nothing about whether crime is going up or down.*

Can local police and city officials be held accountable?

Police and other local officials often complain that they are held accountable for crime, even though they have little control over the economic and social conditions that produce it. For the last several years, my colleagues and I have taken homicide rates for various cities and adjusted them for differences in poverty, unemployment, family disruption and other crime-producing conditions.The adjusted rankings show that homicide rates in some cities, St. Louis among them, are lower than one would expect, based on conditions of economic and social disadvantage. Other cities, meanwhile, have higher-than-expected rates.These figures do not let police or other officials off the hook. On the contrary, they provide a more meaningful comparison of city homicide levels and insight into the effectiveness of criminal justice policies and programs for which local officials absolutely should be held responsible.Rankings of crime rates should be based on well-measured crimes of equal seriousness and identify differences in crime within cities and over time that are produced by factors city officials can control. The Morgan Quitno crime rankings fails on all counts.---

Additional information on reading and using crime statistics is available at http://www.cjgsu.net/initiatives/HomRates-PR-2004-05-14.htm