The Crime-Statistics Con Job

It is a remarkable con job.

Over the last six months, the Police Executive Research Forum, the chief executives of primarily large police departments, has gotten the media concerned that the country is threatened by a sudden upsurge in violent crime and murder.

A New York Times story on March 9th started the current round of hysteria with the headline that "Violent Crime in Cities Shows Sharp Surge."

An earlier front-page story in January in USA Today caused a similar ruckus.

One wonders whether the reporters ever thought of getting a critical comment for their story.

The Police Executive Research Forum report sounded the alarm: "The FBI statistics reflect the largest single-year percent increase in violent crime in 14 years."

It becomes a lot less scary when one realizes that the violent crime rate fell for 13 straight years, a total drop of 39 percent, before increasing in 2005 by less than 1 percent.

The Forum even referred to this minuscule one-year increase as a "trend."

Murder rates did rise from 5.5 per 100,000 people in 2004 to 5.6 in 2005, but they were a little higher a couple of years earlier — 5.7 in 2003 — and 5.6 in 2001 and 2002.

Murder rates have essentially remained unchanged since 2000 after falling from a peak of 9.8 in 1991.

With crime numbers such as these, it is strange that the Forum could get so many people talking about a surge, whether it be nationally or just in cities.

But what the Forum does with the numbers is instantly recognizable for anyone with even a little training.

With the national data, they report the change in the number of crimes, not the change in the crime rate (see their Box 1).

Of course, the total number of crimes may go up in some years, but so has the population.

Further, the Forum very selectively picked what crime categories to report.

For example, they reported murder, robbery and aggravated assault, but not rape. Why? Could it be that the number of rapes, as well as the rape rate, fell?

But the worst is their mangling of the data for city crime. They selectively pick 56 jurisdictions (mainly cities, but some counties) with populations over 70,000. There are 253 cities with over 100,000 people.

The news bite that the media focused on is the claim that murder rates increased by 6.8 percent from 2004 to 2005, 10.2 percent from 2004 to 2006.

Six of the top 20 most populous cities that they just happened to leave out had smaller increases, or even declines, in murders from 2004 to 2006.

There is absolutely no excuse for leaving out these or other cities. After all, local newspapers regularly report how many murders their city had the previous year during the first week or two of each January.

A few hours of Web searches will easily get you virtually all the 50 most populous cities. Many other cities are available as well.

But how could any researcher reasonably pretend that they didn't know the number of murders in New York City? New York City, because of its large population, has the most murders in the country.

If the Forum had also included just that one additional city, it would have brought down their estimated murder growth rate by an entire percentage point.

The crime data for all the cities over 100,000 people are readily available from the FBI for 2004 and 2005.

If you look at them all, not at only the small group the Forum selected, the murder rate rose by only 2 percent, less than a third of what their selective sample implies.

Surely some cities do have serious crime problems, problems that are indeed getting worse. But there are also cities where crime rates have plummeted.

The problems that particular cities face are more local than national in nature. Some cities, such as Philadelphia, have poorly managed police departments and have seen big drops in arrest rates. It is not surprising that violent crime and murder rates have also gone up.

Not surprisingly, last fall the Forum ended its hysterical claims of a gathering crime "storm" with a call for more money for its members' police departments.

There may be good reasons why cities should spend more money on police and prisons, but it should be justified by real numbers.

The one victim of all this ruckus over the crime wave should be the Police Executive Research Forum's credibility.

John Lott is the author of the forthcoming Freedomnomics and the Dean's Visiting Professor at the State University of New York at Binghamton.