Missouri Attorney General Chris Koster visited several cities around the state last week, waving the latest study on racial disparities in traffic stops.
The headline: African-American drivers are 70 percent more likely to be pulled over than whites. Ten years ago, the disparity was 30 percent.
That sure looks bad, but what does this report really tell us?
Not much. It's hard to believe police handling of minority-related issues is more than twice as bad as it was 10 years ago, which is the message implied by the numbers.
If this document had been prepared for a statistics class, it would get an F. It rests on a falsehood: The belief that the propensities of various ethnic groups to violate the law is equal to their proportion in the population.
The purpose of the report, required annually by state law, is to see if cops are engaging in improper racial profiling. But the report can't say that. It can only tell us how many people of a given race are stopped, and compare that to the percentages of those races in the population.
What's missing is the rate at which each ethnic group violates the law. Different groups have different tendencies, and they can vary in unpredictable ways. For example, black males commit a disproportionate number of murders, but most serial murders are committed by white males.
Even Koster admitted the report he so eagerly promoted doesn't support clear conclusions. "A high disparity rate is not proof of racial profiling," he said during his Kansas City stop.
Look, I have no doubt it happens. Some cops let their authority go to their head. But these disparity reports do cops a disservice by tagging officers as guilty without offering solid evidence.
Take the case of New Jersey, where state troopers a decade ago were tarred as racial profilers by state officials using poorly constructed disparity studies. The troopers replied that if you don't know the 'violator benchmark" — the rate of lawbreaking by a particular ethnic group — you can't tell if that group is being improperly singled out.
So a study was conducted, in which motorists were photographed while being clocked with radar guns. As Heather Mac Donald of the Manhattan Institute reported, researchers found that black motorists on the New Jersey Turnpike made up 25 percent of the speeders in 65 mph zones, even though they were only 16 percent of the driving population.
Black drivers were twice as likely to speed as whites — yet only 23 percent of the traffic stops involved African-American drivers. By that measure, blacks were stopped less than they should have been.
If a disparity study doesn't provide the "violator benchmark," it's close to worthless. Worse, like the study Koster released last week, it can encourage poorly thought-out ideas aimed solely at moving the numbers.
Gwen Grant, president of the Urban League of Greater Kansas City, suggested a state policy offering monetary incentives to cities that reduce disparities, or penalties for cities that show worse performance.
When I talked to Grant, however, she acknowledged — like Koster — that raw disparity studies were inconclusive. So, what metric should the state use to allocate rewards or penalties?
Even after talking with her, I'm not sure what she had in mind, but she did suggest something that could generate usable results for police officials: If traffic stops by individual officers were logged, the results could be compared to the performance of cops working the same part of the city and the same shift.
Watch commanders would then see whether some officers are disproportionately stopping minorities, and find out why.
Few cities collect such stats, and those that do aren't eager to make the numbers public. In Pittsburgh, Pa., police officials told Time magazine that the data can fluctuate randomly from quarter to quarter, which could mean more negative headlines based on misleading data-snapshots.
Shallow, deceptive studies such as required annually by Missouri law reveal little about the true extent of inappropriate racial profiling.
But these reports aren’t harmless. They needlessly heighten suspicion of the police and worsen relations between cops and the black community, which suffers disproportionately from crime. If these reports can't be structured in a way that makes them statistically valid, they should be discontinued.