Policing and Racism: Transcript

Raw Transcript

Charles Blue:

Hello, I’m Charles Blue with the Association for Psychological Science. I would like to welcome you to this expert panel on Policing and Racism: Insights from Psychological Science. This is the second in our series of panel discussions about the psychological-science impact on the study of racism and bias.

Before we started, I earlier this morning took a look at the various news stories that came out in the past 24 hours that are actually relevant to this topic. It was a very easy thing to do by simply just going through our search results to find these stories from the past 24 hours.

This demonstrates that the topic of our panel is a daily and ongoing concern. Even today, the Associated Press poll about police violence remaining a high concern in the U.S. was just released a few hours ago. So speaking to these issues, we have our four panelists.

I first will introduce Jennifer Eberhardt who is a professor of psychology and co-director of Social Psychology Answers to Real World Questions, or SPARQ, at Stanford University.

Jennifer is also the President Elect of APS and will be inaugurated APS President next week. Then I have Phillip Atiba Goff who is co-founder and the CEO of the Center for Policy Equity and a professor of African-American studies and psychology at Yale University. Following him will be Keith Payne, professor of psychology and neuroscience at the University of North Carolina at Chapel Hill.

And finally, we will hear from Tom Tyler, the Macklin Fleming professor of law and a professor of psychology at Yale Law School.

Jennifer Eberhardt:

Thank you. I’m Jennifer Eberhardt, I’m from Stanford University, and it’s just a pleasure to be with you today. Many have noted that we are in a period of racial unrest and they point to the killing of George Floyd as the spark that ignited that racial reckoning. Yet, way before the summer of protests that spread across the country, the Pew Research Center released a report which found that six in 10 Americans rated race relations in this country as generally bad and a majority of Americans feared that things were getting worse. They feared we were slipping back in time.


So as a nation, we were already struggling to hang on. And I conduct research to speak to that struggle, I’m here today to talk about bias. Racial bias can get activated in a wide variety of contexts. And today I want to focus on criminal justice. I want to focus on this association between African-Americans and crime, which has deep roots in our culture. With a number of colleagues I’ve conducted studies for the past 20 plus years demonstrating its power, we found that African-Americans are so associated with crime that the mere presence of a black face can cause people to see weapons better.

In a typical study, we invite undergraduates into our lab, we’d have them sit in front of a computer screen with the focus dot at the center, and then they would see flashes of light appear around that focus, that now these flashes were actually the faces of young men that were appearing at such a rapid rate on the computer screen that participants couldn’t consciously detect them. And here it is again in slow motion.

So we use what’s called a subliminal priming procedure to expose some of our study participants to an entire series of black male faces and then well, we had another group of participants where we expose them to an entire series of white male faces. And we found that simple exposure to black male faces enhanced their ability to detect guns from blurry images. Bias can not only control what we see, but where we look, prompting people to think a violent crime can lead them to look toward a black face and away from a white face.

Jennifer Eberhardt:

Prompting police officers to think of shooting or arresting or capturing can lead their eyes to settle on black faces too. A police officer once told me a story that really struck me. He was working undercover and he saw a guy in the distance near a huge office tower with glass walls. And this guy looks similar to the officer in some ways, so he was African-American, he was about the same build, same height, but what was different was that this guy had a scruffy beard.

His hair was uncombed, he looked disheveled. He also looked like he was armed and dangerous. So as they were approaching each other, the officer had an uneasy feeling. And when the officer reaches the office building, he loses sight of the man and then he starts to panic. But then he sees this guy again walking inside the building. Now the officer is still on the outside, but he could see the guy out of the corner of his eye.

And when the officer walked faster, this guy would walk faster too. And when the officer slowed down, he would slow down too. And so the officer decided to stop cold and confront the guy face-to-face. And when the officer looked him in the eye, a chill went through him. He realized that he was staring at himself. He was the person he feared. He was staring at his own reflection through the mirrored wall. That entire time the officer was profiling himself.

Bias can make its way into the courtroom as well, that can affect how we punish. It can even play a role in death sentencing decisions. In a large data set of death eligible defendants, we found that looking more black more than doubled the chances that those defendants would receive a death sentence, at least when their victims were white. And this effect is significant even though we control for factors like aggravators and mitigators, we control for the severity of the crime, we even control for the defendant’s attractiveness.

And no matter what we control for, black defendants appear to be punished in proportion to the blackness of their physical features, the more black, the more death worthy. So what do we do? What do we do when racial bias can be absorbed and acted upon whether we’re sitting in front of a computer screen or walking up a street or deliberating in a courtroom, what do we do? And I want to argue here that although racial bias can touch our lives in so many ways, we are not doomed to be under its grip.

We’re all vulnerable to bias, but we don’t act on bias all the time. Instead, bias is triggered by the situations we find ourselves in. And as researchers, we know a lot about what those situational triggers are. Responding quickly is a trigger for bias. Using subjective standards to evaluate others, that’s a trigger for bias, lack of accountability, lack of training, the list goes on and on here, even our emotional states matter, cultural norms matter.

So this is where bias happens. And these situational triggers are at work not only within us as individuals, they’re at work at a systemic level as well. So let me give you an example. Here in California, I and a number of my colleagues work with members of the Oakland Police Department to help them to reduce the number of stops they were making. And this is a people who were not committing any serious crimes.

And we did this by simply pushing officers to ask themselves a question before each and every stop they made. And that question was, is this stop intelligence led, yes or no? In other words, the officer had to think, “Do I have prior information, prior intelligence to tie this particular person to a specific crime?” And if they determine that the stop was intel led, they had to list the source of that intelligence.

So this procedure forces officers to pause and think, “Why am I considering pulling this person over?” Now the year before adding that question to the form officers complete during a stop, they made about 32,000 stops across the city. That following year with the addition of this intel led question, that dropped to about 19,000 stops. African-American stops alone fell by over 43%. And this drop happened even as the crime rate continued to fall.

So stopping fewer black people did not make the city any more dangerous, in fact, it made it safer for everybody. So how did this intervention work? Well, bias is triggered when we’re responding quickly so we introduced friction, that is we slowed officers down. We changed the standards of evaluation, pushing them to use their intelligence in place of their intuition. We increased accountability by introducing a metric to track these intel led stops.

Jennifer Eberhardt:

The department defined what an intel led stop was and then trained officers on how to spot it. And then the police leadership encouraged these kinds of stops. They changed the norms for what good policing looked like. Is this stop intelligence led? It sounds like a pretty simple question yet there are a lot of psychological principles baked into it. And when we understand the factors that trigger bias, it opens up a huge array of interventions to mitigate it.

Jennifer Eberhardt:

We’ve also been able to examine what happens when police officers do make contact with people. By best estimates, about 12% of US drivers are stopped by police each year, and that’s millions of drivers. And these interactions are consequential. They are the context through which trust is built or eroded on a daily basis. And until now, we didn’t really have a good way of observing how officers communicate with the public. But with the spread of body-worn cameras, we now have access to how these interactions unfold in real time.

Jennifer Eberhardt:

The footage from these cameras allows us to look for patterns across many interactions. And this lets us test the extent to which there are differences in the respect officers communicate to black and white drivers. So with an interdisciplinary team of researchers at Stanford, we began to look at this. We captured footage from nearly 1000 traffic stops and we used machine learning techniques to comb through the words that officers use during these stops.

Jennifer Eberhardt:

And we found that even when officers were behaving professionally, they spoke to black drivers with less respect than white drivers. They used more formal titles with white drivers. They expressed more concern for the safety of white drivers and they offered more reassurance to white drivers. They let them know it’ll be okay, it’ll be all right. In fact, based on the words officers used alone, we could predict whether that officer was talking to a black person or a white person.

Jennifer Eberhardt:

In more recent studies with a completely different data set, we’re finding that we can use the first 45 words that an officer utters to predict whether that encounter is going to escalate, whether the person stopped is going to be handcuffed, searched or arrested. And we can do this with 83% accuracy. So if we’re looking for ways to change systems, this is low hanging fruit. Most law enforcement agencies, especially the large agencies, they have cameras, right? Yet the vast majority of that footage from these cameras, it’s never examined.

Jennifer Eberhardt:

So how can we expect cameras to serve as an accountability tool when the footage is not analyzed, when that footage is not even treated as data? And this has implications far beyond traffic stops. We can use this footage to examine how no-knock warrants are executed on black versus white suspects. We can look at how witness statements are taken. We can look at interrogation practices. We can look at training.

Jennifer Eberhardt:

In fact, the Oakland Police Department invited us to present the takeaways of our findings in a training on traffic stops. And we’re now looking at footage pre and post-training to see whether that training made a difference for real interactions on the street. So you ask, what can we do? You now know that we have many mitigation tools at our disposal to fight bias. We have tools we can use to fight the mere threat of bias, and we can use these same tools to increase fairness and justice. We can leverage science so that we don’t slip back in time, so that we continue to strive to see one another, so that we continue to strive to see ourselves. Thank you.

Charles Blue:

Thank you, Jennifer. I appreciate that. Was there any follow up from any of the panelists at this point? If not, we can move on to Phillip Atiba Goff who will be presenting his commentary.

Phillip Atiba Goff:

Okay. Can anybody see that? Thank you for the thumbs. Thank you Charles. Always nice to be here with good friends, Jennifer, Keith, Tom, Charles, new friends and thank you all for being here and listening. It’s always also wonderful to see Jennifer, usually I prefer to see her going after me so that I don’t have to follow her, manage your expectations. All right. So as Charles said, my name is Phillip Atiba Goff, I am the Carl I. Hovland professor of African-American studies and professor of psychology at Yale and I’m the co-founder and CEO of the Center for Policing Equity.

Phillip Atiba Goff:

And in both of those capacities, I’ll be talking about how we ask the right questions about racism and policing for this APS panel. So I’ve got four goals to talk about in the limited time that I’ve got. Jennifer, how did you manage to go under 15 minutes? You’re the first academic ever to use less time. So again, unable to follow that. My first goal will be to introduce you to the center. Those you don’t know what we do, at the end maybe you’ll know some of the things.

Phillip Atiba Goff:

The second thing will be to specify the right level of analysis, because I think part of the reason why we’re here is because psychologists have tried to analyze this problem at a psychological level, and that is important and it cannot be the only way we talk about it. The third goal is to show some science, some elements, some things it’s just one study I want to show you about that mostly demonstrate that if we don’t bother to look, there’s all kinds of problems we haven’t even begun to explore.

Phillip Atiba Goff:

So as bad as things are, they’re in some ways worse, that means all the solutions that we’ve got, we need to make them even better. And goal four is to give you some examples as Jennifer did, of some promising work, some things we can do. I’m going to subdivide that into the harm reduction work, redesign work, and implementation. All right. Normally if we were in person I’d ask whether or not that made any sense to you.

Phillip Atiba Goff:

I’m usually talking about how I’m not in a black church, but I’m of the black church, I do a little column response. I’m going to pretend that you’re saying all those things back to me so that I can be comfortable. We’ll move right on to goal one. Let me tell you about the Center for Policing Equity. So our mission at the Center for Policing Equity is to make policing less deadly, less racist, and less present where it doesn’t need to be.

Phillip Atiba Goff:

It was founded formally in 2008 after a 2007 conference that some black woman named Jennifer who won a MacArthur Award put together at Stanford University and I was lucky enough to go to as her former graduate student. We leverage social science for better outcomes. So it’s exactly the goal of what I got taught in graduate school to take what we’re understanding in the science and to translate that into better policy that affects people’s lives.

Phillip Atiba Goff:

And importantly, in this day and age we have since 2008 received no funding for or from policing in the process of what we do. We’re funded entirely by federal and philanthropic grants as well as the kindness of individual small dollar donors. So at the center which we constructed to be this interlocutor between law enforcement and science, we have built the largest research and action organization around the police and race in the world. It’s a super humble brag.

Phillip Atiba Goff:

As everybody on this panel can tell you, before a number of years ago this was not an area that was particularly popular. There was not a particular policy emphasis on it. In fact, when we were screaming saying there needed to be something, there weren’t a lot of folks who were listening. At the center we also host the NSF funded National Justice Database which is the largest collection of police behavioral data in the world.

Phillip Atiba Goff:

Also a slightly less humble brag, but as you heard from Jim Comey when he was head of the FBI, it’s an embarrassment the federal government does not collect behavioral data on policing, which is why we know things about crime, but not things that we do in response to it. Where we work as before the summer of 2020, we engaged in harm reduction for vulnerable communities, where we imagined there was power to deliver, we worked with law enforcement and communities to make policing better.

Phillip Atiba Goff:

This intervention model worked with police and communities to produce those better outcomes. But after 2020, we work increasingly directly with electeds and communities to create the conditions so that communities can take care of and keep themselves safe. So I said I was going to talk a little bit about the right level of analysis. Let me talk about that because I think it’s maybe the most important thing I can say to the psychologists who are listening.

Phillip Atiba Goff:

And I’ll start with this statement. Racism is not an audit test. So in the context of policing, here’s kind of what I mean. There are three levels that we want to think about, level of the encounter, the level of community and the level of the city or larger. For those of you who study anything outside of psychology, you’ll recognize this as a micro, a mezzo and a macro level of analysis. And here’s what that looks like in the context of policing.

Phillip Atiba Goff:

In counter level racism is when you have one police officer, by the way, the clip art around this is terrible so I will not be blamed for this, one police officer who treats group of people who look one way or one color differently than another group. So if this one police officer, we’ll call this officer Officer Friendly treats all of the purple people negatively but all of the red people positively, that’s the kind of racism we’re used to, individual level and counter level racism.

Phillip Atiba Goff:

If the purple people were black and the red people were white, we understand this all the time as if the officer’s level decision is the problem. That’s not the only level on which racism functions. So now Officer Friendly treats everyone within the community where he goes exactly the same. And sometimes he’s assigned to a black community or a community where majority of the folks are black, we have white people, there’re Asian folks who live there, Latinx folks who live there, native folks who live there, right? But everybody in this majority black neighborhood gets treated the same.

Phillip Atiba Goff:

And then a different Officer Friendly treats everybody who lives in a majority white neighborhood the same. There are black people who live there, native folks, Latinx folks. But if the majority black neighborhood is treated negatively, the majority white neighborhood gets treated positively, we still on the aggregate have racial disparity. And because policing is so localized, if we don’t engage this mezzo level factor, we’re missing out on a lot.

Phillip Atiba Goff:

But if you’ve studied policing, then you understand there are some researchers who do this, not mostly in psychology, but there are researchers in sociology, demography, economics who do this. Oftentimes we missed the macro. So imagine you have a group of officers that police everybody in the city the exact same way. Let’s take the greatest city in the history of the world, Philadelphia. I happen to be there, but that’s just a coincidence.

Phillip Atiba Goff:

We say everybody in the city of Philadelphia is treated the same way and everybody in the city of let’s say Bridgeport nearby where I live right now gets treated the same way. Well, because we know what the demographics of Philadelphia and the demographics of Bridgeport are, Bridgeport, Connecticut, if everybody in Philadelphia gets treated worse than everybody in Bridgeport, we still have racial disparities. And we oftentimes don’t talk about those different levels where that’s happening.

Phillip Atiba Goff:

So the psychologist’s job within this context is to specify which level of analysis to include those other levels of analysis in our thinking and talk about the psychological factors that influenced behavior at each level and the psychological consequences at each level. Now for social psychologists, our focus is likely to be on those first two levels, right? Our job then is to identify the common situations in both of those levels as exactly as Jennifer did just before, to study those situations, to compare, to prejudice as a predictor, and then to find interventions.

Phillip Atiba Goff:

And the interventions can just be that manipulation that you have in the lab that you take out into the world. So with that conceptual framework, let me move on to talking just a little bit about the science that we’ve been doing at the Center for Policing Equity. I’m just going to give you one example to show you what happens when we start looking in places where previously we thought there was no reason to look. We’re looking at the chronic situation of exposure to members of a vulnerable category.

Phillip Atiba Goff:

So this is work that I’ve done in collaboration with CPE senior vice president of science, Tracey Lloyd. I hear we’re extending on work that we did on who gets to be a child. So we had done this work previously as had Sandra Graham, Brian Larry, [Anita Rutton 00:26:52], with somebody on this call as well, you guys see the heavy Eberhardt influence on all of my work. And what we found was that black children are seen as older, less innocent and more responsible for their actions, right?

Phillip Atiba Goff:

This is in the context of policing as a situational factor where the more it is likely that a child has engaged in something illegal, the older they got treated to be. But what does that mean out there in the actual world? I’ve got one study in that 2014 paper that looks at it out in the actual world, but we wanted to go broader. So we examined nine cities for use of force rates and we compared by race and age. Here we’re collapsing across gender and some other categories, we’re looking at race and age.

Phillip Atiba Goff:

We’re looking at black and white, adults and children, children just defined as under 18. And here we’re able in these nine cities to benchmark against race, age, arrests. So all of these rates are benchmarked against the arrest rates for that race and age category for white adults, white children, black adults, black children. And what we expected in this study was simply that childhood would be protective, but more for white children than for black children. Again, nine cities looking at use of force benchmarked against the race and age arrest rate.

Phillip Atiba Goff:

And we do that to get a sense of all right, well, how much are they likely to come into contact with law enforcement for something that maybe they did? We understand that that’s not a perfect benchmark, but you’ll see it doesn’t prevent us from seeing some pretty distressing things. So normally I would show you some complex hierarchical regressions because that’s my jam, but sometimes descriptive science is the best way to go. So all I’m going to show you is the child to adult force ratios. That’s the first one. I’m going to show you three slides. So children on top, adults on the bottom of this fraction, right?

Phillip Atiba Goff:

And we want the number to be small because that means adults are more likely to face force in these nine cities than are children. I’m going to show you white in the dark blue and black in the light blue. And what you’ll see is that child to adult force ratio, it’s about 0.41 for white folks and more than twice that for black folks. And again, this is out in the world, we’re using actual use of force data from the nine police departments.

Phillip Atiba Goff:

What about the more traditional, black to white force ratios? Nationwide, this is about a four to one force ratio, especially when controlling for these factors but these nine cities they’re particularly progressive, they use force less than the national average. What are we going to see here? Well, it turns out here we’re going to see a little less, black on top, white below. Instead of four to one, black to white, we saw 1.64 to 1 for the adults, which is usually what we’re looking at when we’re looking at national stats. But it was more than double.

Phillip Atiba Goff:

The racial disparity was more than double for black to white children. Now all of that is distressing to me, but none so much as this next finding, it’s a kind of finding that reminds me again of graduate school, forgive me for being back in here where we’re first looking at the black ape association, when understanding that in fact, people literally implicitly associated black folks with apes. It makes you sit and think because we thought that childhood would be protective just less so for black children than for white children.

Phillip Atiba Goff:

The next one I’m going to show you is the median use of force per 1000 residents within a race age category. I’m just going to show you two, I’ll show you white adults and black children. But what we see is that per 1000, we’ve got 12 use of forces for white adults and 14.6 for black children. What I’m showing you here is evidence that in these nine cities, whiteness is more protective than childhood. It’s not something I would have predicted.

Phillip Atiba Goff:

It’s not something that I think that folks who do developmental psychology or perceptions of children would have predicted. And yet we find it when we bother to look, which means we need to have innovative solutions for bigger than the problems that we collectively understand because we have not begun to investigate the problems at the depth at which they occur. Normally I’d just leave you like this, but it turns out I’m under time. So I get to talk to you just a little bit about something that’s good news, some promising work that the Center for Policing Equity has been up next to helping communities do in the last several years, particularly in this last year.

Phillip Atiba Goff:

So I’m going to start with harm reduction, which we take as language from the language of public health, right? So in public health we talk about harm reduction in the context of addiction, right? If you can’t get somebody to quit their addiction, they can’t cease taking an illicit substance that’s doing bad things to their health and their behavior. You want to keep them alive long enough that they get the chance to try over and over again, right? That’s harm reduction.

Phillip Atiba Goff:

In this context, we think about harm reduction as this nation’s addiction to punishment. And the context of that, the work we’ve been doing has been following what law enforcement has been doing for quite some time. Some of the greatest gains in crime reduction through policing have been through the simple application of performance management. Anybody who’s ever heard of COMPSTAT understands that that’s data police collect to hold themselves accountable to crime, and it turns out it does reduce crimes. It can be used to help reduce crimes.

Phillip Atiba Goff:

So we use the science that’s been done around this, particularly Andrew Gilman’s model along with Jeff Hagan to replicate that same performance management element, but with the goal not of reducing crime, but of reducing injustice. Because through these statistical methods, which again, I’m happy to show you all the extra nerd work at another point, what we can do is we can try and separate out the segment of racial disparities that belongs to the police behavior and police policy and the portion of that that belongs to education or employment or healthcare, the portion that they can do something about and a portion that they can’t.

Phillip Atiba Goff:

When you give that data back to communities and a law enforcement that can optimize to that in the same way that every organization optimizes to everything they’ve ever mentioned. We’ve got about 60 plus jurisdictions, I think we’re at 63 right now in partnership in this and on average where we have pre and post data, we see a reduction of arrests by 25%, of use of force by 26%, of officer related injuries by 13%. Now that’s just pre and post. So we’re looking forward to seeing in a causal way what portion of that is the intervention.

Phillip Atiba Goff:

I’m not talking causality here, but pre and post that’s better than nothing right now, better than a poke in the eye with a sharp stick, right? That’s the harm reduction side. Let me give you a little bit on the redesign side. This is the work on re-imagining public safety and how you move from re-imagining to redesigning. So this summer in July, we put out a roadmap for redesigning public safety, a bunch of things that folks could do to shrink the footprint of law enforcement in the places where both law enforcement and community said, “We don’t want to be.”

Phillip Atiba Goff:

So here, I think we’re at about 66 departments, jurisdictions that have used this through the Obama Foundation where we put this out. It was a way to start thinking about less systems of punishment at a time where we understand that’s good, but it’s scary because the murder rate is spiking across the country. What folks have done with that is that it’s empowered those who were worried about violence to reduce the footprint of law enforcement through an evidence based metric. It says, measure this first, this second, this third and if you’re still scared to go to the community, see what the community wants.

Phillip Atiba Goff:

We’ve been really excited to see people take that up, and particularly in places where we’ve worked even more closely, places like Berkeley, California. They’ve recently moved to eliminate low-level traffic enforcement by police, right? Not just to decline to prosecute, it been done in other places like Baltimore. And that’s our county where they’ve redesigned 911, they’ve removed officers from schools and they’ve ended the armed response to all non-violent infraction.

Phillip Atiba Goff:

In both of those places, we helped with those plans and we led with the evidence. And most important to me, I think maybe the professional accomplishment I’m most proud of happened this past April in Ithaca and Tompkins County. The state decided they wanted to have everybody come up with a new plan for re-imagining. If again, Tompkins County, they said, “We don’t want to do anything incremental. We want to do something that fits the moment.”

Phillip Atiba Goff:

So they moved for resolution to dissolve their police department and replace it with the department of community solutions and public safety. That new department will be civilian led, majority unarmed, they will not send any armed responders to nonviolent incidents, they will have a public health orientation and public health resources to create proactive investments in prevention. And the reason I tell you that story is not just because it’s a story of pretty significant change of a kind I never thought I would see but to this last point.

Phillip Atiba Goff:

Because there was widespread expectations of the police union as they often do, would sue to stop this plan. Of course, we’re going to dissolve the police department, the union would sue as they frequently do. But the Monday before the vote, the police benevolent association ended up publicly endorsing elements of the plan and not resisting it publicly in any way. I don’t expect we’re going to be able to do that again, but it does bear repeating. This is what can happen when community is armed with science, so they can trust a process when they can’t trust each other.

Phillip Atiba Goff:

So let me get off the stage and I give it up to folks who are both more compelling to watch and have interesting science results to get to. I just want to make sure that we get after Amadou Diallo, after that tremendously tragic incident where an individual holding up a wallet got shot at 41 times by NYPD because they thought it was a gun and it revolutionized the ability to talk about implicit bias.

Phillip Atiba Goff:

We’re past the point where just those split decisions are the things where psychologists owe responsibility to communities. We have to take the broader context of policing seriously and we can. Because it’s not all bias is implicit and racism is more than psychology, but as you’re going to hear today, psychology plays a role at how we reorient policing within that context. When we have the broader context in mind, we can specify what it is we’re supposed to do, and we can deliver on that in a way that doesn’t shrink the problem, but is responsive to it. All right, let me stop talking, give this back to Charles who’ll give this to Keith.

Charles Blue:

Thank you very much. A very engaging and thought provoking discussion. To keep us on time, I’m going to transition right now to Keith Payne. And again, if there are any questions, please send them to me in chat and we will get to them when everyone is finished. Keith.

Keith Payne:

All right, thank you. So let me make sure my slides are showing here. Anyone see those? Great. Well, my name is Keith Payne, I’m a psychology professor at the University of North Carolina Chapel Hill. I want to pick up on one of the themes that Professor Goff brought up and expand upon it a little bit in my talk. I’m going to not talk a lot about policing directly, but I’m going to take a step back and talk about some of the psychological and social dynamics that set the context for inequalities and bias in policing.

Keith Payne:

So to tell you what I’m going to tell you before I tell you, we’ve had a lot of broadening of the conversation in our society in the last couple of years around two terms, one, implicit bias, it’s become more of a household word and the other one is systemic racism. Before George Floyd’s killing, people outside of the academy didn’t use those words very often, but now they’ve percolated into a much broader kind of conversation.

Keith Payne:

What I want to argue is that these are not necessarily two entirely separate things. In fact, they might be two aspects of the same problem. So for the last 25 years or so, implicit bias has been studied in our labs and in lots of other settings, mainly as an individual attitude, as something in our heads like a belief or a preference that we carry around with us. What I’m going to argue is that in addition to that, it might actually be a better reflection of the social environments we inhabit, not just the thoughts in our heads, right?

Keith Payne:

So if that’s the case, then what it means is that we need to change our thinking from trying to educate or persuade or train people out of their individual attitudes and more toward trying to modify the social environments and the structures and the settings and the systems around us. So to begin, the standard narrative that we tell about implicit bias, both in the classroom within universities and when we talk to journalists go something like this.

Keith Payne:

So we’re all born in this society that has stereotypes and inequalities and we learn those stereotypes really early, soak it up from our culture as children. And because we learn that and practice it so much over the course of many years growing up, it becomes this really rigid bias, this attitude that’s difficult to change. We have lots of studies in psychology showing that it’s difficult to change them and if you do change them, they seem to snap right back.

Keith Payne:

And yet even though it’s difficult to change implicit biases, the story goes, they’re very important because there are individual differences in these biases and they tell us who’s likely to discriminate. So we assume that people who are higher on implicit biases and measures of implicit bias are going to discriminate a lot and those who are low in implicit bias are not. And so if we want to reduce discrimination, at least the part of discrimination that comes from unintended bias, then what we need to do is identify those individuals who are high in bias and either screen them out of decision-making roles or to train them and rehabilitate their biases in some way to change their attitudes.

Keith Payne:

There are a couple of problems with that view though. One is that over the last few years, lots of research has been showing that when you look at individual differences in implicit bias and try to correlate them with behavioral measures of actual discrimination, those correlations are really low. They’re not necessarily zero, but they’re quite small, meaning that if you look at somebody’s implicit bias score, whether it’s with something like the implicit association tests or some of the weapon oriented implicit bias measures that Professor Eberhardt talked about, it’s very difficult to predict who is going to actually discriminate and who’s not going to.

Keith Payne:

The second problem for that standard narrative around implicit bias is that implicit bias is not very stable at all. So in research studies, we talk about the test retest correlation. In other words, if I give you an implicit bias test today and I give it to you another time two weeks from now say, are you going to score in similar ways? And the answer is no. We can’t predict very much about how your score in a couple of weeks based on how you score today.

Keith Payne:

And that’s a problem if you’re thinking about implicit bias as an individual attitude that we carry around with us, because the people who score high today aren’t necessarily the people who are going to score high tomorrow. And if that’s the case, how do you expect to predict individual differences in discrimination? Now some researchers have looked at those two kinds of data as a criticism of implicit bias measures or a criticism of the implicit bias concept itself.

Keith Payne:

But there’s a very different way to look at these data that puts this in a very different line. So if we look instead of individual differences and trying to characterize a person as high in implicit bias or low in implicit bias, if we instead start looking at social contexts that are relatively high or low in bias, in other words, the average level of implicit bias in different places and contexts, these data look very, very different.

Keith Payne:

So what I’m showing you here is just two examples, on the left is a study that looked at the average gender bias linking male and female to science versus the liberal arts, and each data point on here is the average score for a country. And what’s on the vertical axis is the actual male, female gender discrepancy in standardized tests in math and science for those countries.

Keith Payne:

And what you find is the countries with a larger average implicit bias linking male to science have a much larger actual disparity in standardized test outcomes. And that’s a very strong relationship. On the right, you see state-level averages, so each data point here is one of the 50 states and that’s the state average on a race IAT measure of implicit bias. And on the vertical axis is the frequency of Google searches in that state for racist search terms or slurs.

Keith Payne:

And here it’s, again, an extremely strong correlation. You can predict with great accuracy based just on knowing what the average implicit biases in that state, how often people are Googling for racist search terms when they think they’re in the privacy of their own home and on the privacy of their own computer. Again, a very strong relationship. And the final correlation I want to highlight here is looking at actual racial disparities in police use of deadly force.

Keith Payne:

So this is data from Eric Hayman and colleagues and what they find is that the average implicit association on the implicit association test at the city level linking black people and weapons relative to white people is strongly associated with the racial disparity and actual police use of deadly force in that city. Now, if you think about this, these are data that come from project implicit where millions of people go to take the implicit association test or other kinds of tests.

Keith Payne:

And what these data are showing is that the average level of implicit bias from people that have nothing to do with you predict what kinds of outcomes you might have in these different cities. So this is not police officers going to this website and taking these tests. So if you’re a black person in one of these cities that has high levels of implicit bias, based on the implicit bias of other people, you’re still more likely to be shot by police in that city than in a city with low levels of bias.

Keith Payne:

And what this all suggests is that we need to think much more systematically about how implicit bias relates to places like cities and counties and states and workplaces rather than just the people taking the tests. So my colleagues and I call this perspective, the bias of crowds model of implicit bias. We argue that implicit bias works something like the wisdom of crowds where an individual may not know the answer, but if you survey the answers to any kind of question like how tall does the Statue of Liberty, if you ask a hundred people for their guesses and that average number, it turns out to be really surprisingly accurate.

Keith Payne:

And the reason is that each person has a little bit of partial uncertain knowledge, and then uncertainty that’s kind of random, some people overestimate, some people underestimate. And if you average across those, what you get is an average answer that reflects the knowledge that’s in that crowd. So implicit bias can work that way too. And the reason is that implicit associations in our heads that link black or white social groups to various stereotypes and ideas, they can vary both chronically as a function of the person, we carry them around with us in our head to some extent, but also temporarily as an aspect of the content.

Keith Payne:

And what that means is that just like with the wisdom of crowds, this partial and fleeting knowledge in these associations, knowledge of stereotypes can create robust, stable aggregates that converge on the level of bias in that context even if no individual in that context actually holds that attitude or belief. And what that means is that implicit bias might be more informative about the context, in other words, the situations and the structures around us than the individuals that have it though.

Keith Payne:

And so that’s a major shift in the way that we think about what implicit bias means and what we ought to do about it. So just to give you a concrete metaphor, instead of thinking about it as something in our heads, it suggests that implicit bias is a little bit more like the wave in a stadium, right? So when the wave starts going around in a stadium, when it gets to your section, you’re likely to stand up and raise your hands regardless of what you think about standing and cheering, right? It’s not existing in our heads, it’s passing through us, it’s a true social phenomenon.

Keith Payne:

And so if we think about it like that, we need to stop asking or not just ask questions about why are some people more biased and others less, but what kinds of settings create bias? And we can ask that at lots of different levels and we need to be asking it at lots of different levels. I want to share with you some data that goes back to the most fundamental cause of this kind of bias that we could think of, which is the historical practice of slavery in the United States.

Keith Payne:

So what I’m showing you here is a map that President Lincoln used in planning and strategizing in the run-up to the civil war. This is data from the 1860 census mapped on to a map at a county by county level. And the counties that are shaded darker have a larger percentage of their population enslaved in the 1860 census. And what you can see here is it’s not some just broad north south difference, one county to the next varies a lot in terms the enslaved population based on all sorts of factors related to agriculture and geology and geography.

Keith Payne:

And now President Lincoln was using this to plan because he knew the states and areas that were more economically dependent on slavery were going to be more committed to secession and he turned out to be right. The states seceded in almost exactly the same order as based on the proportion of their populations that were enslaved. And he also knew that border states that had less slavery like Kentucky, Illinois, Maryland… sorry, Missouri and Maryland could be more likely to be persuaded to stay in the union because they were less economically dependent on slavery. All right.

Keith Payne:

So he was using it for the purposes of planning for the war, but it turns out that the economic dependence on slavery at a very granular county by county level as well as a state by state level turned out to predict not only how committed these regions were to seceding in the civil war, but those areas that were more economically dependent on slavery also created more intense Jim Crow laws, they created a more intense racial segregation.

Keith Payne:

To this day, they have more residential racial segregation, larger racial disparities in poverty and other kinds of modern day economic outcomes. So what we did is we took this data from this map and we basically created a digitized version of it. What you’re seeing on the top here is a digitized version of that same map from the 1860 census, and on the bottom is the intensity of implicit bias from the project implicit website among white residents today, and the areas that are in white simply don’t have enough data.

Keith Payne:

So I don’t know if you can see the relationship by just eyeballing it, but let me show it to you plotted a different way. On the left, you see a county by county association and on the right you see state level associations. And in both cases, you see that places that had a larger proportion of their population enslaved in 1860 have higher implicit bias today over 160 years later. So I think that is fairly striking evidence that it’s not just the individuals, none of whom are around from them that carry around these implicit biases with us, it’s the systems and structures around us.

Keith Payne:

And so what is it that connects these enslaved populations in 1860 to modern day implicit bias? Well, it has to be something about the structures around us that are long lasting and slow to change. And so what we found was that if you look at measures of systemic racism, in other words modern day disparities and poverty rates, modern day racial disparities in economic mobility rates and modern day housing segregation, these are the mediators, the factors that connect slavery in 1860 to implicit bias today. In other words, places that had a lot of slavery in 1860 developed more systemic racism that is still detectable today, and it’s those modern day systemic disparities that predict the average level of implicit bias today.

Keith Payne:

So just to wrap up, to the extent that these tests reflect social context, we need to rethink some things specifically the standard narrative of implicit bias as a personal attitude as opposed to systems and structures around us, training to change people’s individual biases which has become very popular in recent years, I think we need to rethink that and instead focus on interventions that change situations and not people. And that means focusing on policies, procedures, power differentials in the systems around us, not just attitudes in our heads. So let me stop there. Thanks all my collaborators and turn it over to you all. Thank you.

Charles Blue:

Thank you, Keith. Those again, very insightful with some important background and context, as we’re having this discussion today. To bring it all home is Tom Tyler who will be our final presentation of the day. Also, if any of the participants wish to ask a question, please send it to me by direct message and we will get into them immediately after Tom’s talk. Tom, go right ahead.

Tom Tyler:

Okay, well I may begin as many of the other people on this panel have begun by acknowledging the George Floyd killing and sadly the very many other killings that have gone in the same direction in recent years. As both Jennifer and Phil have mentioned, a focus on racism and racial justice is a natural response to the situation because African-Americans are more likely to be treated in the waves that they’ve laid out. But what I want to suggest is that policing in America is a problem for all Americans, even though black Americans are the most strongly impacted by the police.

Tom Tyler:

My main points for this talk are several. One, the warrior culture of policing addresses problems from an era when high levels of violent crime characterized America. But today violent crime is very low. So the warrior culture is a mismatch for the issues that contemporary police deal with. Furthermore, this culture hinders the ability of police officers to be effective in most of what they do today. So there are two solutions we might put forward.

Tom Tyler:

One is reorganized police agencies around this service model so that the police can better handle the tasks that they actually perform. The other is shrink the scope of policing and have many tasks that are now handled by the police handled by others. This is an ideal moment for us to think about this larger issue, to think about what the role of American police should be in our democratic society and we need to begin by asking the question what communities think their problems are and how they would like to solve those problems.

Tom Tyler:

And I think consistent with the general theme of this panel, psychological science can both support and inform this process. So the previous speakers have already made it clear that there are widespread calls for changing policing. Phil did a great job of pointing out the many things that are potentially being done in different areas. A lot of policy changes are being recommended, strengthening officer accountability, removing problematic officers, mandating training.

Tom Tyler:

I think we have to be realistic and we have to point out that at this moment at a national level, these policy changes are getting stopped and it’s not even clear if the George Floyd Police Reform Bill will pass Congress. That doesn’t mean the change isn’t happening, and I’m not going to repeat what Phil Goff said, I think we’re seeing change at a local level. What I think we want to emphasize is we need to move beyond surface policy changes, even if those changes are desirable and consider the underlying organization of policing.

Tom Tyler:

And you can tell from my saying that, that I’m a social psychologist, but the point I think in the larger sense is that America has never had a discussion about what it really wants from its democratic policing. This is a discussion for everybody, not just for minorities. Well, what’s true of the police today? The police define their mission as harm reduction through crime control. They’re trained and equipped to use force to compel compliance. They’re basically warriors and they describe themselves that way.

Tom Tyler:

But this framing of the police flows from an era of high crime and the police have now changed to reflect the fact that crime is very low now. And I’m going to show you a figure. This is the rate of violent crime from 1990 to 2014, dramatic decline. Now I know that you will have heard that there’s a spike in violent crime, which there is. So if you look over on the side of the slide, I noted in 2020 there is a spike, but look where it puts violent crime compared to where it was in 1990, it puts it slightly above. So let’s not overstate the spike. Crime is at historic low levels and it remains that way.

Tom Tyler:

So that leads us to a mismatch where we have a warrior skillset that doesn’t match what the police do all day. Their jobs seldom require them to be warriors. Estimates are at 4% of their day involves tasks that require armed officers that are capable of dealing and deploying force dealing with violence. What their job does require them to do is to be mental health workers, social workers, mediators, and lifestyle dispute managers.

Tom Tyler:

So what are the problems? One, the police don’t have the training, the skillset to manage those problems the one they actually counter and their training as warriors makes these situations worse. Because they bring a dominance force based orientation to all the situations they deal with, which spirals conflict and leads to many of the violent situations that we’re all aware of today. So just to give you an example of recent research, what the police do all day in different cities, you’ll see that across these cities about 4% of the time the police are dealing with violent crime and all the other things don’t require an armed police presence.

Tom Tyler:

Well, what’s the consequence of this warrior focus? What we see is that public trust in the police has dropped. And in 2020, a majority of adult Americans expressed distrust in the local police, only 48% of Americans say they trust their local police. And trust among black Americans is 37% lower, this is not an anomaly. We’ve seen for decades that about 60% of adult Americans trust the police, now it’s down to about 50%. So the strategy that’s being used is not building police legitimacy.

Tom Tyler:

All right, well, what do we want the police to do in response to this problem? One approach is for the police to change so that the officers can actually manage these other tasks that they actually perform. This requires culture change, the police need to rechange their mission, they need to rethink and move towards a service model, a different style of policing is required. So we need retraining, new policies and practices and different reward and promotion approaches.

Tom Tyler:

All right, well, how can psychology help with this? Psychology has trust-building models, the current hierarchical model of policing concentrates authority at the top, and then pushes down that authority on the community. But psychological research suggests the value of inverting this authority structure considering how people experience policing and what people think is desirable police behavior.

Tom Tyler:

The police have assumed that the style of policing that gets them approval is performance, but we’re very clear in our research, we psychologists, in showing that what people primarily want is they want fair procedures from the police. They want to have voice, they want evidence of neutrality, they want treatment with courtesy and respect, and they want to feel that the officers that they deal with sincerely care about them and are trying to do the right thing.

Tom Tyler:

So we need to change police culture and develop this new model of policing, build public trust, train officers differently and have a more consensual model. All right. But an alternative approach that also should be considered, and it’s more like the Ithaca model that Phil already mentioned is re-imagining the concept of public safety. So an alternative approach is to lessen the police footprint in the community. We could bring the police into police stations and only have them deal with violent crime where we actually needed armed police presence and replace them with social workers, mental health workers, again, something that Phil mentioned several communities already doing.

Tom Tyler:

I think one core problem with this is that although crime is down substantially, police forces have not downsized. So they have a lot of resources that we would have to reallocate, and this is a political issue. Let me just give one example here. As you saw, crime has dropped dramatically. Violent crime is about 25% what it was in 1990. The number of police officers in America is the same as it was in 1990. So a lot of resources are going to create a service for a problem that doesn’t really exist anymore and those resources are the resources that could be reallocated to produce the social services that Phil was talking about.

Tom Tyler:

So we need to shift the conversation to building vital communities. If the police manage the conversation, then they talk about how to control crime. But the absence of harm is not the same thing as the presence of vitality. Community members need to be asked what the needs of their community are and how they would like to see them addressed. This can be collaborative with the police, but the police can’t control it as they so often do. And it has to include traditionally excluded voices.

Tom Tyler:

So what can psychology contribute? Models of deliberation, we know how stakeholders can work together to re-imagine their communities. We can demonstrate the value of community deliberation. If we want economic, social and political vitality, we need engagement and identification. Those things are built by collective deliberation. And finally, we need to identify models of wellbeing. We need to talk about what a good community looks like, not just the absence of crime, but the presence of vitality. Psychology has those models.

Tom Tyler:

All right. So finally then how can we combat racism in policing, the theme of this panel? Change the style of policing so that the police connect with the community, they’re less likely to express bias and less likely to use force. Emphasize positive goals for the police beyond harm reduction and change the role of the police in the community, recognizing that you cannot arrest your way out of crime. The police have to work with everyone else for community development as a longterm mechanism for lowering crime. Thank you.

Charles Blue:

Thank you very much. I’d like to thank Tom and the rest of the panelists for their contributions. I did have a few questions come in which I’ll pass along, but if anyone else has additional ones, please feel free to enter them in the chat. I’d like to start off with a general question for the panel, and that is what role does incentivizing arrests have on police behavior? The, I guess, rewards or metrics that are given for more arrests, is that something that is tied into this, that there seems to be a benefit for police to be this warrior that I believe Tom was talking about because it’s considered the metric of their job?

Tom Tyler:

Well, I’ll just start with that and then other people could also respond. The whole system of policing incentivizes police officers to focus on trying to arrest people. Because if you want to be successful as a police officer, you need an arrest record, that’s how people are viewed as good police. And in particular, the whole warrior mentality is you’re fighting crime. So how do you show that you’re good at your job? Is through arrests.

Tom Tyler:

So I think it’s absolutely true that one of the things we always talk about with culture changes, instead of the police officer with the most arrests being the one whose picture is on the wall as officer of the month, how about the police officer who has the best relationship with the community? That would change what police officers think a good police officer looks like, that person should be promoted and then officers will start to see that the way to get ahead as a police officer is through building good relationships with the community, not through arresting people.

Charles Blue:

I do want to just interrupt that. Phil has to go and I asked him if he has any final comments that he would like to add after hearing the rest of the presentations.

Phillip Atiba Goff:

As always, I am smarter for having listened to my colleagues. And to both your question to all of us and following up on Tom’s response, I don’t think we can overstate how crushing it can be to live in a community where it’s obvious that the people are disposable. I listen, oftentimes the communities where we work and they talk about being referred to as throwaway people. And one of the ways that we can combat that is simply to do as Tom was suggesting but even on a grander scale, measure the things that actually matter.

Phillip Atiba Goff:

In the context of public safety, we’ve only ever measured crime, we’ve never measured anything like justice. That’s why our interventions are well received by communities is because for the first time we’re saying, “Oh, we’re going to measure what the state does to you.” We have never measured once on a national level, what the armed representatives of the state empowered to take away life and liberty do in the communities that are majority made up of the children, the descendants, the formerly enslaved people.

Phillip Atiba Goff:

If we cared about the legacy of what we have done to black folks in this country, there’s no way we wouldn’t even bother to measure the thing. It’s not what we as psychologists often think about, but if we remember that we’re supposed to be doing not just attitudes, but behaviors as well which you’ve been hearing throughout all four presentations, we could rededicate ourselves to getting some of those measures of behavior right and figuring out the right ways to use those to measure impact.

Phillip Atiba Goff:

So I just, again, for the psychologist listening if you can hear the sound of my voice, behavior out in the world it matters a great deal. And from Kurt Lewin to Jennifer Eberhardt, we shouldn’t shrink away from going and doing the difficult work of finding the ways to measure that and putting that in within our theories, giving that back to communities. I’m so sorry I have to go, but this is a wonderful, wonderful event. I’m so glad to have been a part of it. Thank you all.

Keith Payne:

Thank you Phillip.

Jennifer Eberhardt:

Thanks Phil.

Charles Blue:

I do have a question and this may be for Keith, but certainly anyone can jump in. And the question is, can we say that physical context, i.e monuments or neighborhoods that still reflect segregation are part of the social context that contributes to systemic racism and implicit bias? Is it possible that these physical contexts provide cues that influence police prejudice actions?

Keith Payne:

Right. So what we can say right now is that they’re correlated. So we have… I showed you one study showing the physical aspects such as housing segregation are correlated with the level of implicit bias in the community. We’ve done other studies looking at implicit bias across different college campuses. And we looked at what we considered markers of sort of concrete markers of systemic racism at a campus level. And those things that were associated with the level of implicit bias on most campuses were things like the presence of confederate statues, so part of the built environment and also the amount of faculty diversity.

Keith Payne:

So the representation of diverse faculty members, but also people with power on campus who are people of color. So those concrete aspects of the setting that we’ve looked at so far, or at least correlated with the level of implicit bias, I don’t think we can say yet that they’re necessarily the causes because it’s bound up with lots of other things going on in the culture. But this is all fairly new research and that’s one of the big questions out there, what are the aspects of the literal context around us that are transmitting those biases?

Jennifer Eberhardt:

Another researcher at Buju Dasgupta has looked at this in terms of gender, where you have… you’re looking at students who are at an all women’s college and you find that they have gender bias that is pretty similar to the rest of the population when they walk in the door as freshmen, but then over time, as they spend more time in this other context where they see women professors, they see women leaders in stem and so forth, that bias changes the longer they stay. So that’s another example of the same kind of thing we’re talking about but looking at gender.

Charles Blue:

Seeing as that we are running a bit over time, I want to finish off with one question and I’ll begin by answering it, but this was actually presented to us. So that is what steps would you suggest individuals take in response to this, to learning about your research findings? And I’ll start off by saying that on our website psychologicalscience.org, we do have research topics, one of them is policing where you can look at some background and additional research done on this topic. But to wrap things up, would each panelist just like to recommend kind of where to go next for information and where would you recommend that journalists look for the latest research?

Tom Tyler:

Well, I was probably remiss when I started by not saying that I’m a member of the Justice Collaboratory at Yale Law School, which has a website, which is a perfect repository for a lot of this information. If anybody wants to see articles, wants to see different policy statements, I would recommend going through the Justice Collaboratory website at Yale Law School as an example.

Jennifer Eberhardt:

I also am a part of a group called SPARQ, which Charles mentioned earlier on. And we are a group of researchers who work in collaboration with practitioners to address significant social issues, and one of those issues has to do with criminal justice and policing. So going to the SPARQ website is S-P-A-R-Q .org, you can get information there. And then I believe there are going to be some of the members of SPARQ who will be involved with producing a backgrounder for this and so there’ll be more information for others who want to know more there. Thanks

Keith Payne:

Now pointing out because Jennifer is too modest, she has a fabulous book called Bias today. You can see that summarizes much of her research on this and broader topics. I also have a book to plug, it’s called The Broken Ladder. It talks about the psychology of inequality broadly, it includes racial inequality as part of that. So check those out as well.

Charles Blue:

With those wonderful references in mind, I would like to thank our panelists and participants for today. There will be a transcript of this conversation and also a video that will be available on the website and I will send an email to all participants and panelists when that’s available. And if you have any questions, you should all have my email, feel free to contact me directly, I’m happy to put you in touch with anyone or to follow up with anything that needs additional information. So thank you very much. Again, I appreciate your time.


APS regularly opens certain online articles for discussion on our website. Effective February 2021, you must be a logged-in APS member to post comments. By posting a comment, you agree to our Community Guidelines and the display of your profile information, including your name and affiliation. Any opinions, findings, conclusions, or recommendations present in article comments are those of the writers and do not necessarily reflect the views of APS or the article’s author. For more information, please see our Community Guidelines.

Please login with your APS account to comment.