University of Cincinnati police officer Ray Tensing stands near a car after driver Samuel DuBose was allegedly pulled over and shot during a traffic stop in Cincinnati, Ohio.University of Cincinnati police officer Ray Tensing stands near a car after driver Samuel DuBose was allegedly pulled over and shot during a traffic stop in Cincinnati, Ohio. Reuters

What researchers found after analyzing data gathered from 20 million stops in North Carolina.

While traffic stop interactions with the police may be shrugged off as brief inconveniences for whites, for black Americans, they can lead to humiliation, violence, and even death. This has become clear over the last few years, as videos have surfaced, hashtags have trended, and reports have been released—opening up the black box of negative interactions between the police and drivers of color for the world to see.

A forthcoming book, "Suspect Citizen: What 20 Million Traffic Stops Tells Us About Policing And Race" adds to that conversation, taking an unprecedented, granular look at the traffic stops in one state.

In 1999, North Carolina became one of the first jurisdictions in the country to mandate data collection at traffic stops. The expressed goal was to suss out disparities in policing. The resulting dataset, which includes information about the demographics of the driver, the offense for which they were stopped, where they were stopped, and the outcome of the stop, was made public. But the state never actually released a comprehensive analysis of this information.

That's where Frank Baumgartner and Kelsey Shoub at the University of North Carolina, Chapel Hill, and Derek Epp, now at the University of Texas, stepped in. They took on this unfulfilled mission, analyzing data going all the way back to 2002 when the data-collection mandate expanded to include almost all police stops in the state. "It is pretty much a census of every traffic stop," Baumgartner said.

In the book, he and his colleagues lay out stark disparity in policing at North Carolina's traffic stops, and unpack the reasons behind the trends they observe. CityLab caught up with Baumgartner to discuss these findings:

So who was being stopped?

There's somewhere between a million and 1.6 million traffic stops in North Carolina each year and we [the state] have a population of about 10 million people. That gives us a baseline chance of being pulled over of about 10-15 percent per year. But we found the odds were significantly higher for blacks, than for white and even Hispanic drivers, compared to their respective population shares.

We also looked at a city-by-city comparison of the proportion of whites, blacks, and Hispanics who live in that town to the proportion that they represent in the traffic stop data. Again, we do this with caution, but still it shows that, on average, black drivers are much more disproportionately represented—about 60 or 70 percent more likely to be in that traffic stops data than in the population of that city.

But according to the [U.S.] Department of Transportation, white Americans are more likely to own or have access to a car than black or Hispanic Americans. So we think that the comparison of these populations is not really very accurate—it's actually more likely to lead to an underestimate of any racial bias.

What about searches?

Our main focus in the book is who gets searched after a traffic stop because being searched is sign that the officer views you with suspicion. Hence the title of the book "Suspect Citizens." I'm a white, middle-aged college professor, so the last time I was actually stopped for a traffic violation was 40 years ago, in 1974 ... and I've never had my car searched after a traffic stop in my life. These things are quite rare for people of, for example, my demographic but they're quite common generally.

We controlled for why you get pulled over, what time of day it was, what day of the week was that, what police agency was it, what month of the year—all of those things. We still saw these very, very significant, robust findings that young people, males, and people of color are much more likely to be searched after a traffic stop.

White, middle-class drivers are more likely to get a ticket.

What were the outcome of these types of searches?

Strangely—and this might initially appear surprising—populations who are more likely to be searched are also more likely to be let off with a warning or no action. And white, middle-class drivers are more likely to get a ticket. It would seem counterintuitive to most readers, but if you're objectively breaking the law—you're speeding or you run through a stop sign—you deserve to get a ticket. The officers will only pull you over after they observe you clearly violating the traffic code in an important way. Then they're going to give you a ticket and let you go on your way. And that's typically what happens to white drivers.

Black drivers are more likely to get warnings, overall. That seems like a good outcome, but the question in these cases is why did he or she get pulled over in the first place? Perhaps the reason was just that the officer had a vague suspicion and a desire to do a little bit of an investigation. So they stopped the person based on some kind of pretext, they investigated by starting a conversation with the driver, and [since nothing turned up], they said, "Well, thank you for your time. I pulled you over because you were speeding by five miles an hour. I just wanted to let you know to be more careful next time."

So you're saying that in the cases of white motorists, it might be that they are more often given traffic tickets because there's actually a concrete, observable evidence of a traffic offense. But in many cases of motorists of color, perhaps there is a less clear reason for the stop, so they may actually not end up being penalized through tickets.

Yes. There are two reasons for a traffic stop. One is a violation of the traffic law and the other, a desire by the officer to investigate a person because they they think that they might be a suspect. So if you're a person like me, a white middle-class male, there's no reason the officer is going to develop a suspicion. So the only reason why he might pull me over if he observes me violating an important part of the traffic code, like running through a stop. A traffic stop leading to a ticket is really what should occur if the traffic stop is being used to enforce the traffic code, not as a tool for a criminal investigation.

Around the 1960s and 1970s, police agencies around the country changed their style of policing from responding after the fact to crimes, investigating and trying to solve crimes, to being much more proactive and trying to disrupt the activities of people who might be about to commit a crime. And so that marked a very significant shift in the nature of policing the roads. It used to be once upon a time that you know the police will be targeting only those who were driving in a very unsafe manner such as speeding excessively or running through a stop sign or something like that. But today, really for the past generation since the War on Crime, police have used the entire vehicle code as an excuse to use their discretion to pull people over who might appear suspicious to them for whatever reason.

So you mentioned that you don't fit the profile a patrolling police officer is looking for, but in the book, you explain that what they're looking for can be different when the driver is black than when they are Hispanic.

For young, black men, the concern is whether they might fit a stereotype of being involved in criminal activity. And for Hispanics there are two profiles. There is the same profile that we see among men in general, but minority men, in particular—of being involved in crime. But there's also the immigration focus. So if an officer is attempting to inquire into somebody's citizenship status then it really doesn't matter if they're male, female, young, or old—as long as they're Hispanic, they may be subjected to that kind of search.

What did you find about the rate at which contraband was detected at these stops?

In the War on Drugs or the War on Crime, the police have made clear from the very beginning, as one of the state troopers we quote in the book said, that, "You have to kiss a lot of frogs before you find your prince." The idea there is that it's a numbers game, that you have to pull over and investigate maybe hundreds or thousands of drivers before you find anyone with a significant cache of drugs.

Unstated in that is that we're going to deprive potentially very large numbers of people their right to privacy, their right to drive on the highway unimpeded, and their right to be a citizen without being a suspect—all in the hopes of finding a few drug couriers. And the Supreme Court has ruled consistently that that is OK, because it's just, after all, a momentary inconvenience.

We want to simply point out that if you're like Frank Baumgartner and you get stopped once every 30 years and it doesn't result in a search, that is that's probably fine—that calculation makes good sense. It's worth it for all of us. But what we see are these patterns where people like me virtually never get pulled over and young men of color get pulled on a very, very routine basis. And it may not be so trivial—it can be humiliating, it can take a lot of time—and it comes up dry the vast majority of times. [According to the book, in most types of searches resulting from traffic stops, officers are much less likely to find contraband on black drivers compared to white ones.]

Overall, we see contraband hit rates that are on the order of 20 to 30 percent. But then when we look in detail at the contraband that's found, we discover that it's almost always a trivial amount. And so the math of this—using the vehicle code to search for contraband and criminal activity—is really bad. The odds are just not on the side of the police in this case. It's a very inefficient use of police officers' time.

How do policing stops vary by geography?

One of the biggest surprise in our data was a wide degree of variability in search rates or even the number of traffic stops in several ways. One is from year to year. Then, there are incredible amounts of difference from agency to agency. So the state Highway Patrol, on average, is a ticket writing machine. They write a lot of tickets, but they don't search very many cars—just 0.6 percent of the time. Whereas in the city of Charlotte, there have been years when they've had a search rate of 12 percent. And, of course, those searches are targeted by demographic group, and they're also targeted by neighborhood. We know that is true from some other more detailed data we've been able to gather. Policing is very focused on neighborhoods that the police think of as high crime, and it applies to everybody in that neighborhood, whether or not they're involved in crime. And in low-crime neighborhoods, aka the white side of town or the middle-class neighborhoods, the police presence is much lighter, and also the police activity might be much less aggressive. So that means that we're all subject to different forms of policing based on where we live.

If [a traffic stop] is a pretext and it's the third time it has happened to you since you turned 16 years old and you got your driver's license, and you're only 17, you know that the officer is simply suspicious of you almost no matter what you're doing. That's going to make you upset.

The downside of all this is that they are consistently given the signal that they're a suspect.

The biggest predictor of low disparity is having black representation on the city council.

What did you observe about the places with very low rates of racial disproportionality in traffic stops?

We looked systematically across all the municipalities of North Carolina and we found that the biggest predictor of low disparity is having black representation on the city council. That is correlated with having a large black share in the population and having a large share of black voting in the most recent election.

Whereas when you look at a more typical community in North Carolina, that's got very low levels of black political power, those are the towns where you see the highest rates of disparity. It indicates that the agencies of government do respond to politics.

Whenever a case comes up of a traffic stop gone awry—and there have been many showing shootings or excessive force or even deaths—there's often a defense that the individual officers are "bad apples." Do your findings support that hypothesis?

We define a "bad apple" as an officer with a very high rate of disparity across the races, in terms of their search. We can identify many, many hundreds of them across the state. But when we isolate the bad apple officers, we still see that over and above that, there remains a systemic pattern of racial disparity. So that means that while we do identify some bad apples, that by itself can't explain the disparities that we observe. These are broad systematic, institutionalized, culturally normative practices and they're just part of almost every police department almost.

We've also looked beyond North Carolina. We find that North Carolina is not an outlier—it's quite typical of these patterns, nationally. I feel like the professor who drives a bulldozer through a door that's already open—to try and demonstrate that yes, there are racial disparities in traffic stops, everybody! No, that shouldn't be a surprise to anybody, but is it a 3 percent disparity or a 100 percent disparity? We show that black drivers are about twice as likely to be pulled over. And then once they're pulled over they're about twice as likely to be searched. So that's a 400 percent disparity; that's pretty shocking. It's not just a small thing, it's a really, really big difference in lived experiences for whites and minorities.

So do we have to rethink traffic stops? What are the solutions here?

Go back to the original purpose of a traffic stop, which is to keep the community safe by maintaining safe-driving habits. Focus traffic stops on people who burn through stop signs, and don't focus the traffic stops on things like expired registration tags, cracked taillights—things that might be technical violations of the law, but really amount to "poverty crimes," which poor people are more likely to be caught up in than middle class and wealthy people.

I do feel like the police are given an almost impossible mission sometimes: to be clairvoyant, to know ahead of time who is carrying drugs, who's involved in a crime, and who is preparing to commit a crime. They're given a lot of authority to use their judgment, and legally speaking, they're not required to pay much attention at all to the frustrations that they generate among those who they stop and investigate. If we could instill in the culture of policing a recognition that each time there's a fruitless investigation, that they slightly alienate that individual, then that would be very useful.

So a broader reckoning with the costs of predictive policing...

Yes. By deciding that they need to investigate somebody just because of the way they appear, or what neighborhood they might be in, rather than an observed violation of the law, they're sending a signal to that individual that they're suspects. That's costly to all of us. These people won't call 911 when there's a crime because they don't trust the police. People who feel that they're not treated fairly by the police are more likely to develop an aversion to police and that translates into an aversion and a hostility towards government, because the police of course are such a visible representation of the government. And so voting rates may go down.

The War on Crime and using the traffic code and the vehicle code as a way to investigate people more broadly has not given very great benefits in terms of crime reduction, but it's had an unintended and unnoticed consequence: It has made people feel like they don't have full citizenship. They walk down the street and they're suspects.

Tanvi Misra

Tanvi Misra is a staff writer for CityLab covering immigrant communities, housing, economic inequality, and culture. She also authors Navigator, a weekly newsletter for urban explorers (subscribe here). Her work also appears in The Atlantic, NPR, and BBC.

  1. A cow stands in front of a barn painted with a U.S. flag

    Not all rural areas are declining: Some are thriving, while others are undergoing significant transitions.

  2. University of Cincinnati police officer Ray Tensing stands near a car after driver Samuel DuBose was allegedly pulled over and shot during a traffic stop in Cincinnati, Ohio.

    What researchers found after analyzing data gathered from 20 million stops in North Carolina.

  3. Fog is seen clearing around the Shard skyscraper in Londo.n.

    It's not always big leaders with big plans.

  4. Life

    The city's latest move to limit vacation rentals could come in handy for other cities trying out their own regulations.

  5. A photo illustration shows a homeless encampment in Seattle.

    From mapping apps to the blockchain, new tools are intended to give cities the information they need to address this growing challenge.