Documents menu

Driving While Black; A Statistician Proves That Prejudice Still Rules the Road

By John Lamberth [newspaper source not cited]
Sunday, August 16, 1998; Page C01

In 1993, I was contacted by attorneys whose clients had been arrested on the New Jersey Turnpike for possession of drugs. They told me they had come across 25 African American defendants over a three-year period all arrested on the same stretch of turnpike in Gloucester County, but not a single white defendant. I was asked whether, and how much, this pattern reflected unfair treatment of blacks.

They wanted to know what a professional statistician would make of these numbers. What were the probabilities that this pattern could occur naturally, that is, by chance? Since arrests for drug offenses occurred after traffic stops on the highway, was it possible that so many blacks were arrested because the police were disproportionately stopping them?

I decided to try to answer their questions and embarked on one of the most intriguing statistical studies of my career: a census of traffic and traffic violators by race on Interstate 95 in New Jersey. It would require a careful design, teams of researchers with binoculars and a rolling survey.

To relieve your suspense, the answer was that the rate at which blacks were stopped was greatly disproportionate to their numbers on the road and to their propensity to violate traffic laws. Those findings were central to a March 1996 ruling by Judge Robert E. Francis of the Superior Court of New Jersey that the state police were de facto targeting blacks, in violation of their rights under the U.S. and New Jersey constitutions. The judge suppressed the evidence gathered in the stops. New Jersey is now appealing the case.

The New Jersey litigation is part of a broad attack in a number of states, including Maryland, on what has been dubbed the offense of "DWB"--driving while black. While this problem has been familiar anecdotally to African Americans and civil rights advocates for years, there is now evidence that highway patrols are singling out blacks for stops on the illegal and incorrect theory that the practice, known as racial profiling, is the most likely to yield drug arrests. Statistical techniques are proving extremely helpful in proving targeting, just as they have been in proving systemic discrimination in employment.

This was not my first contact with the disparate treatment of blacks in the criminal justice system. My academic research over the past 25 years had led me from an interest in small group decision-making to jury selection, jury composition and the application of the death penalty. I became aware that blacks were disproportionately charged with crimes, particularly serious ones; that they were underrepresented on jury panels and thus on juries, and that they were sentenced to death at a much greater rate than their numbers could justify.

As I began the New Jersey study, I knew from experience that any research that questioned police procedures was sensitive. I knew that what I did must stand the test of a court hearing in which every move I made would be challenged by experts.

First, I had to decide what I needed to know. What was the black "population" of the road--that is, how many of the people traveling on the turnpike over a given period of time were African American? This task is a far cry from determining the population of a town, city or state. There are no Census Bureau figures. The population of a roadway changes all day, every day. By sampling the population of the roadway over a given period, I could make an accurate determination of the average number of blacks on the road.

I designed and implemented two surveys. We stationed observers by the side of the road, with the assignment of counting the number of cars and the race of the occupants in randomly selected three-hour blocks of time over a two-week period. The New Jersey Turnpike has four lanes at its southern end, two in each direction. By the side of the road, we placed an observer for each lane, equipped with binoculars to observe and note the number of cars and the race of occupants, along with a person to write down what the observers said. The team observed for an hour and a half, took a 30-minute break while moving to another observation point and repeated the process.

In total, we conducted more than 21 sessions between 8 a.m. and 8 p.m. from June 11 to June 24, 1993, at four sites between Exits 1 and 3 of the turnpike, among the busiest highway segments in the nation. We counted roughly 43,000 cars, of which 13.5 percent had one or more black occupants. This was consistent with the population figures for the 11 states from which most of the vehicles observed were registered.

For the rolling survey, Fred Last, a public defender, drove at a constant 60 mph (5 mph above the speed limit at the time). He counted all cars that passed him as violators and all cars he passed as nonviolators. Speaking into a tape recorder, he also noted the race of the driver of each car. At the end of each day, he collated his results and faxed them to me.

Last counted 2096 cars. More than 98 percent were speeding and thus subject to being stopped by police. African Americans made up about 15 percent of those drivers on the turnpike violating traffic laws.

Utilizing data from the New Jersey State Police, I determined that about 35 percent of those who were stopped on this part of the turnpike were African Americans.

To summarize: African Americans made up 13.5 percent of the turnpike's population and 15 percent of the speeders. But they represented 35 percent of those pulled over. In stark numbers, blacks were 4.85 times as likely to be stopped as were others.

We did not obtain data on the race of drivers and passengers searched after being stopped or on the rate at which vehicles were searched. But we know from police records that 73.2 percent of those arrested along the turnpike over a 3 1/2-year period by troopers from the area's Moorestown barracks were black--making them 16.5 times more likely to be arrested than others.

Attorneys for the 25 African Americans who had been arrested on the turnpike and charged with possessing drug or guns filed motions to suppress evidence seized when they were stopped arguing that police stopped them because of their race. Their motions were consolidated and heard by Judge Francis between November 1994 and May 1995.

My statistical study, bolstered by an analysis of its validity by Joseph B. Kadane, professor of statistics at Carnegie Mellon University, was the primary exhibit in support of the motions.

But Francis also heard testimony from two former New Jersey troopers who said they had been coached to make race-based "profile" stops to increase their criminal arrests. And the judge reviewed police in-service training aids such as videos that disproportionately portrayed minorities as perpetrators.

The statistical disparities, Francis wrote, are "indeed stark . . . . Defendants have proven at least a de facto policy on the part of the State Police . . . of targeting blacks for investigation and arrest." The judge ordered that the state's evidence suppressed.

My own work in this field continues. In 1992, Robert L. Wilkins was riding in a rented car with family members when Maryland State Police stopped them, ordered them out, and conducted a search for drugs, which were not found. Wilkins happened to be a Harvard Law School trained public defender in Washington. With the support of the Maryland ACLU, he sued the state police, who settled the case with, among other things, an agreement to provide highway-stop data to the organization.

I was asked by the ACLU to evaluate the Maryland data in 1996 and again in 1997. I conducted a rolling survey in Maryland similar to the one I had done before and found a similar result. While 17.5 percent of the traffic violators on I-95 north of Baltimore were African American, 28.8 percent of those stopped and 71.3 percent of those searched by the Maryland State Police were African American. U.S. District Judge Catherine Blake ultimately ruled in 1997 that the ACLU made a "reasonable showing" that Maryland troopers on I-95 were continuing to engage in a "pattern and practice" of racial discrimination.

Other legal actions have been filed in Pennsylvania, Florida, Indiana and North Carolina. Police officials everywhere deny racial profiling.

Why, then, are so many more African American motorists stopped than would be expected by their frequency on the road and their violation of the law? It seems clear to me that drugs are the issue.

The notion that African Americans and other minorities are more likely than whites to be carrying drugs--a notion that is perpetuated by some police training films--seems to be especially prevalent among the police. They believe that if they are to interdict drugs, then it makes sense to stop minorities, especially young men. State police are rewarded and promoted at least partially on the basis of their "criminal programs," which means the number of arrests they make. Testimony in the New Jersey case pointed out that troopers would be considered deficient if they did not make enough arrests. Since, as Judge Francis found, training points to minorities as likely drug dealers, it makes a certain sort of distorted sense to stop minorities more than whites.

But there is no untainted evidence that minorities are more likely to possess or sell drugs. There is evidence to the contrary. Indirect evidence in statistics from the National Institute of Drug Abuse indicates that 12 percent to 14 percent of those who abuse drugs are African American, a percentage that is proportionate to their numbers in the general population.

More telling are the numbers of those people who are stopped and searched by the Maryland State Police who have drugs. This data, which has been unobtainable from other states, indicates that of those drivers and passengers searched in Maryland, about 28 percent have contraband, whether they are black or white. The same percentage of contraband is found no matter the race.

The Maryland data may shed some light on the tendency of some troopers to believe that blacks are somehow more likely to possess contraband. This data shows that for every 1,000 searches by the Maryland State Police, 200 blacks and only 80 non-blacks are arrested. This could lead one to believe that more blacks are breaking the law--until you know that the sample is deeply skewed. Of those searched, 713 were black and only 287 were non-black.

We do not have comparable figures on contraband possession or arrests from New Jersey. But if the traffic along I-95 there is at all similar to I-95 in Maryland--and there is a strong numerical basis to believe it is--it is possible to speculate that that blacks travelers in New Jersey also were no more likely than non-blacks to be carrying contraband.

The fact that a black was 16.5 times more likely than a non-black to be arrested on the New Jersey Turnpike now takes on added meaning. Making only the assumption that was shown accurate in Maryland, it is possible to say even more conclusively that racial profiling is prevalent there and that there is no benefit to police in singling out blacks.