:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/56055559/GettyImages_545477316.0.0.jpg)
Is racism in law enforcement the problem of a few bad apples, or is the system as a whole rotten?
A new working paper looking at police officer discretion in speeding tickets in Florida tries to answer this question — and it finds that the answer is somewhere in between. In total, the number of police officers who show racial bias in the study is around 25 percent — not all cops, but still a fairly high number.
The researchers caution that the findings only apply to the specific police department (the Florida Highway Patrol) and type of policing (what kind of speeding ticket someone gets if they are being ticketed) that they studied. If you’re looking at different agencies or police work, it’s entirely possible that the rate of officers who are biased won’t be 25 percent; it could be higher or lower.
But the findings have big implications for policy. If racial bias is a systemic problem that affects every single officer, then the solutions would need to target entire police departments — by, say, changing the training that all cops go through. But if it’s just some cops, then more targeted interventions may be better.
Given the rise of the activist movement against police brutality and racism over the past few years, answering these questions is now crucial. The study brings us a little closer to those answers.
What the study found
The study, by Princeton researchers Felipe Goncalves and Steven Mello, looked at how Florida Highway Patrol officers doled out speeding tickets from 2005 to 2015.
As in other states, the penalty for speeding in Florida increases at certain intervals, based on how far over the speed limit a driver has gone. Officers, aware of this, often give speeding tickets below the jump to avoid giving a driver a harsher fine.
So you get funny situations like one in Florida in which there are a lot of traffic tickets for going 9 mph over the speed limit — just below the 10-mph threshold that triggers a steeper fine.
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8964809/fines_and_tickets.png)
Goncalves and Mello looked at how patrol officers in Florida use this discretion based on a person’s race. They found a statistically significant, but small, difference.
Without controls, black and Hispanic drivers are, respectively, 3.8 and 14.9 percentage points less likely than white drivers to be cited at 9 mph. But once researchers added controls (such as gender, age, income, speed, previous tickets, and vehicle type) the difference shrunk to a small, but still statistically significant, disparity: Black drivers were at least 2 percentage points less likely to be ticketed for driving 9 mph above the speed limit, while Hispanic drivers were about 1.4 percentage point less likely.
Goncalves and Mello then looked at whether it was all police officers or just a few driving this bias — by analyzing some of the individual characteristics of the officers. Ultimately, they concluded that the racial bias was driven by roughly 25 percent of officers. They also find that older officers are more likely to be racially biased, while women and college-educated cops are less biased on average.
The researchers argue that firing biased officers and hiring more women could help reduce the gap, but not substantially, based on their statistical models. Instead, the most impactful change could be to shift the more lenient officers — those who are less likely to give out a lot of harsh traffic tickets — to places with larger minority populations to counteract the effects of racial bias. (Right now, the opposite tends to be the case: The researchers found that the most lenient cops are more commonly put in Florida communities with smaller minority populations — perhaps a sign of systemic racial bias on its own.)
The paper comes with a few caveats. It’s not peer reviewed, so it’s possible that some of the statistical tools used — and the results they led to — could change once they go through more rigorous scrutiny. It also only looks at levels of bias after someone is stopped and ticketed; it’s entirely possible — and even likely, based on previous research — that there’s bias based on who the officer decides to stop and ticket in the first place. And the findings only speak to Florida state police; it’s likely that different agencies would see different results.
The researchers readily acknowledge the caveats. “We study a specific decision (fine discounting) and a single department,” they told me by email. “Several studies have found no racial bias in other policing activities (e.g. vehicle searches) and other police departments, and we would not suggest using our study [to] say that 25% of officers are biased in these contexts as well.”
Jennifer Doleac, a researcher at the University of Virginia who was not involved in the study, called it “a really nice paper.” But she agreed that there are some caveats to it.
Doleac gave a few examples of how the 25 percent estimate could be too low in terms of measuring overall police bias. It’s possible that officers are more likely to stop and ticket black drivers; in that case, the number is likely an underestimate — because it doesn’t pick up the larger number of white drivers who are let off with a warning or never stopped at all.
But it’s also conceivable, Doleac explained, that the estimate could be too high: “Police are sensitive to drivers’ ability to pay the ticket, and black drivers on average are less wealthy than white drivers. They don’t discriminate when deciding whom to pull over, but are more likely to let black drivers off with a warning because they worry a ticket would be a greater burden for those drivers. The bigger mass of white drivers at 9 mph over the speed limit would then reflect white drivers’ being less likely to get a warning (and thus more likely to show up in the data).”
More research will be needed to tease out which scenario is right.
Still, the findings help provide a bit of insight into what, exactly, is going on at police departments. With so much attention going to racial bias in law enforcement, this study suggests that the problem may be by and large caused by a significant minority of cops.
This isn’t the first study to find racial bias in policing
Of course, this is hardly the only paper to produce evidence of racial bias in policing.
Doleac pointed to another paper by UC Santa Cruz researcher Jeremy West that looked at racial bias in how police officers punish people during car crash investigations. This is fairly close to a truly randomized trial, because officers don’t get a say in whether a car crash happens. Yet West found that officers were more likely to give citations for both moving and non-moving violations, although not felony violations, to drivers of a different race.
There are less clever studies that produce evidence of racial bias, particularly in use of force. A 2015 study by researcher Cody Ross found, “There is no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates.” That suggests something else — such as, potentially, racial prejudice — is going on.
Other studies show that officers are quicker to shoot black suspects in video game simulations. Josh Correll, a University of Colorado Boulder psychology professor who conducted the research, said it’s possible the bias could lead to even more skewed outcomes in the field. “In the very situation in which [officers] most need their training,” he previously told me, “we have some reason to believe that their training will be most likely to fail them.”
It’s not just police; there’s evidence that the public as a whole is biased — which could, of course, trickle down to police departments. A recent series of studies published by the American Psychological Association, for example, found that people are more likely to see black men as larger and more threatening than white men, even if the black men are not actually larger. Another study published in the Journal of Personality and Social Psychology in 2014 found that people are more likely to view black children 10 years and older as “significantly less innocent” than their white counterparts.
There is research — depending on how it’s done, what types of policing it looks at, and which police departments it analyzes — that doesn’t find significant evidence of racial bias.
One study by Harvard economist Roland Fryer, for example, found that cops in 10 jurisdictions in three states weren’t more likely to use lethal force against black suspects than white ones. But that study has huge caveats — most notably, it only looked at whether an officer shot once he had stopped someone; that wouldn’t pick up whether there was a bias in who the officer stopped, which could still lead to racial disparities in police use of force if the officer had a similar chance of shooting people of any race but was more likely to stop, say, black people.
Overall, however, the research indicates that there’s racial bias in at least some types of policing and at least some police departments. (Even Fryer’s study found evidence of racial bias in non-lethal use of force.) The study by Goncalves and Mello is just the latest evidence on that pile.
For more on the problems facing American policing, read Vox’s explainer.
