Print
Return to online version

May 4, 2008

Clarity on Profiling

Justin Katz

Race's status as an umbrella term for an amorphous category of qualities — from skin color to lifestyle choices — justify a significant degree of skepticism about claims of racial profiling. In the context of a recent University of Rhode Island investigation of state police traffic stops and searches (concerning which, I haven't been able to find more detail than that provided in today's Projo story), for example, there are various factors that could correlate with race and that would be of varying degrees of culpability when it comes to motivation for searches.

Suppose 90% of all people who were stopped and whose cars were searched were — as we used to call them — hoodies. A presumption of increased likelihood of finding something suspicious or illegal in the car is thus premised on visual clues about the character of the driver and the cultural significance of his or her comportment. Perhaps that oughtn't be grounds for searches, but it's certainly less sinister than the specter of racism, and general experience suggests that the criterion would affect races disproportionately. Driving habits and attitudes are other examples of factors that might inherently select for race without indicating racism (unless, as advocates and activists like to do, one treats them as definitional of a people).

Of course, for the analysis to be fair, we'd need more information than is available. Perhaps the presumption that hoodiness correlates with criminality is mistaken, thus making it an unfair reason to treat motorists differently. To answer the question, however, one would have to pull over and search random cars; if more hoodies have contraband, then the profile is reasonable. (In point of fact, we are operating under just such experience, albeit without the mooring of scientifically collected data.)

What numbers the Projo article does supply confuse more than they enlighten:

The authors conclude that if two drivers, one white and the other black, were driving vehicles of the same age, with no passengers, on similar roads in the same area at the same time of day, the black driver would be 1 1/2 times as likely to be pulled over as the white driver by troopers from the same state police barracks. Hispanic drivers would be slightly more likely to be stopped. ...

The study found "substantial evidence of racial and ethnic disparity" in searches where troopers had discretion in whether to search, and said there was little change from the previous studies. Blacks were twice as likely to be searched as whites, and Hispanics 1 1/2 times as likely. After adjusting for a number of factors that could explain some of the difference, the authors said, Hispanic drivers were no more likely to be searched, but blacks were still 1 1/2 times as likely to be searched as whites.

However, despite the more-frequent searches, no more contraband (mostly drugs) was found among nonwhites than among whites.

The fact that troopers searched black drivers more often than whites but found no more contraband, the study says, suggests that there was less legal basis for searching the blacks. In fact, the state police found drugs and other contraband slightly more often in the vehicles driven by whites as in those driven by minorities. (Contraband was found in vehicles driven by: whites, 42.9 percent; blacks, 42.2 percent, and Hispanics, 40.5 percent.)

Obviously, those percentages cannot be of the total number of drivers, because that information is not possible to collect, so they must represent the portion of either stops or searches. If they are percentages of those stopped, then they are noteworthy, because every car stopped but not searched would go into the "clean" category, and fewer minority-driven vehicles were not searched.

More likely, however, they are percentages of cars searched, meaning that more contraband was found in the minority-driven cars, in absolute terms. In that case, impropriety only exists if additional searches of white-driven cars would maintain the police's find rate for that group.

Reporter Bruce Landis writes that the study's "results are consistent with what one would expect from biased law-enforcement tactics," but given the very narrow range of discrepancy in these percentages, they are also consistent with what one would expect if state police officers' instincts are of roughly equal accuracy across races, but are more often triggered by drivers in minority groups.

The prescription for remedying that problem lie beyond the boundaries of law enforcement.

Comments

No one has taken age into consideration.Older people of whatever background are less likely to be searched.
Place is important too.A car driven by a white person in let's say, a public housing project mostly inhabited by non-whites would be much more likely to be stopped and the driver searched than a non-white driver.The reason would be drugs-it's just the truth-most white druggies buy their supply from non-whites and there ain't too much home delivery in that business,although it's not unknown.A non-white driver in the same place would be routine and would have to do something out of the ordinary to initiate the same response.

Posted by: joe bernstein at May 4, 2008 3:58 PM

"What one would expect, Bruce?"

If a reporter uses a phrase that describes findings as "consistent with what one would expect," is the possession of such an expectation the equivalent of bias?

Is it ok for reporters to have "expectations," but not for police officers to have them?

Just askin' . . .

Posted by: brassband at May 4, 2008 4:00 PM

The logic is bewildering. Just because two things may have some "consistencies" with each other, or follow a parallel trend line on a chart, that doesn't in and of itself create a "cause and effect" relationship between them.

Could one then make the case that RICO actions against local organized crime figures are inherently racist, because many of those mafioso happen to be Italian-American? How come no Swedes have been targeted recently for RICO investigations in Rhode Island? I want an investigation!

Generally speaking, demographic groups have some differences, much having to do with cultural expectations and upbringing -- otherwise, why would anyone use demographics to chart anything to begin with? Most of the differences in reality are minor, but to deny that they exist at all, is just denial. Is an 85 year old grandmother likely to have her car radio blasting at 120 db or to be drag-racing on the highway? Is a 20 year old Mormon likely to get arrested for DUI? How come the Amish are underrepresented in car accident reporting data. Hmmm?

Younger people tend to drive more dangerously than older ones, men much more so than females (excepting the application of cosmetics in the mirror). How else would you explain auto insurance rate pools?

Much of the story seems to rely on methodologies that are covered in a book -- that 4 out of 5 people think is great -- called "How to Lie With Statistics."

PS Bruce Landis' reporting is "consistent" with someone with biased left-wing agenda.

Posted by: Will at May 5, 2008 1:13 AM

Isn't it the duty and obligation of a skilled law enforcement official to be discriminating in any and all ways as it applies to their job? This whole issue is pure rubbish and should just go away.

I could be wrong but I don't think that racial profiling is an important issue in most of the other states.

Posted by: Frank at May 5, 2008 8:06 AM

Ah, it wouldn't be Monday morning without a game of "The reporter's a lefty," would it?
Carry on.

Posted by: rhody at May 5, 2008 11:16 AM

Here's a profiling story from when I was on detail with the Providence PD drug unit back in the early 90's.There were still pay phones that took incoming calls back then.There were no cell phones around.But dealers had pagers.We would set up near one of these phones and look for an older model car,usually a bigger sedan,preferably with Mass plates,but RI would do also containing a few scruffy looking white junkie types(male or female,or both)in their thirties or so and see if they made a quick call.If they did and got a call back.If they went to wait in the car,we were on.This was a daylight operation because that's when most heroin was sold.Invariably a car with a dealer or two would show up and we would take everyone down during the deal.Most of the dealers were aliens,both legal and illegal,if you were wondering why I would be working these cases.Racial profiling?NOT.Behavioral profiling based on knowledge of prior events.Steven Brown can go piss up a rope.

Posted by: joe bernstein at May 5, 2008 11:25 AM

rhody-journalism draws more leftwingers than most professions-i don't know why for sure,except that leftists love to stir up s**t.

Posted by: joe bernstein at May 5, 2008 9:20 PM

"After adjusting for a number of factors that could explain some of the difference, the authors said, Hispanic drivers were no more likely to be searched, but blacks were still 1½ times as likely to be searched as whites."

My question is, what was the margin of error of this study? The above "adjustment" does not sound like an m.o.e. for the whole study, only for one aspect of it.

And if the difference in "profiling" falls within that m.o.e., it's not clear that they have made their case.

Posted by: Monique at May 5, 2008 9:51 PM