I’m no social scientist. My worst grade in college was statistics. But when I read about the recent study by Alex Piquero, a professor at The University of Texas at Dallas, among others, I wanted to understand. “Dissecting the Complexities of the Relationship Between Police Officer-Civilian Race/Ethnicity Dyads and Less-Than-Lethal Use of Force,” published in the July issue of the peer-reviewed American Journal of Public Health, is a mouthful. But when I read the actual study, the conclusion was succinct: “Although we observed significant bivariate relationships between race/ethnicity dyads and use of force, these relationships largely dissipated after we controlled for other factors.” Still confused? So was I. So I called up Professor Piquero to ask him what his conclusions about race and police in Dallas really mean. The answer involved a lot of tomato sauce.
This study feels like a big deal. What inspired it? The issue of the use of force. Remember, the use of force is not just deadly shootings. You’ve got to remember deadly shootings are the most severe but the most infrequently used of all police options. So, we have to start with that. Most of the academic research on this topic has only been focused on black officers, white officers, black citizens, and white citizens. And so we really didn’t know what other combinations would look like, especially with Hispanics and Latinos. So, we didn’t know if force was used differently when there was a black officer and a Latino citizen, or a Latino officer and a white citizen. We called these groupings in the article dyads. So we could intermix the race and ethnicity of officers and citizens to do those comparisons.
Why haven’t those comparisons been done before? The majority of police departments don’t have this level of detail available for people to look at. The Dallas Police Department, many years ago, was one of the leading police departments in this area. They’ve uploaded actual crime data from the DPD, including force data, on their website, so anyone—you and I could do it, our moms could do it, our brothers could do it—can go online and download it. We were fortunate that the former police chief David Brown was willing to invest the resources in providing those data to the public.
How did Chief Brown end up so far ahead of the curve? There are a few departments around the country that have been making their data publicly available. It’s certainly the exception and not the norm in big city police departments. But Chief Brown had always been interested in trying to be as transparent as possible to people. Some people may not have believed that, but I think providing the data to the public is one way of going about doing that.
Did you have any expectations going in about what the outcome would be? No. You know what? That’s a really good question. The reason why is because no one had ever done this. So it’s like you’re making a new recipe from scratch. You have no idea how it’s gonna taste cause you have nothing to compare it to. It’s similar to that. We had no idea what the results would look like.
I would have assumed that there was a connection between race and use of force in Dallas. What we had found, when you only looked at black versus white comparisons, was that the evidence—and people don’t like hearing this in the social sciences—the evidence is mixed. Some studies show that there’s a disparate use of force across race. Other studies don’t report that. So you had this mixture of evidence and all of these other studies that had looked the black-white focus. And because we had these other categories and were able to make these dyads and then also look at different uses of force, we really had no idea what to expect. And to be quite honest, from the researcher part of us, it didn’t matter which way the result came out. From an academic perspective, we just wanted to know the answer to the question, at least in these data for this time period.
OK, I get that, but were you surprised in any way by the outcome? No, not particularly. The overall balance of the research tends to show that there are not race differences or ethnic differences in the levels of use of force. Especially when you take into consideration all of these other characteristics of the situation, which is what we did in the studies. For example, we looked at whether there was a mental illness when the officer showed up, the type of call that the officer was responding to. When you take into consideration more of the factors of the incident, those initial relationships that people oftentimes are expecting, they disappear in our data set.
I’m no social scientist. So this is where you start to lose me. Let me make a food analogy. If someone was going to make a cake from scratch, you’re gonna have flour, you have sugar, you could have eggs, baking soda, maybe cinnamon, maybe nutmeg, depending upon what you’re making. I can adjust the taste of that cake if I put in other ingredients. It’s the same thing making pasta sauce. For a pasta sauce, you start with tomatoes, right? But at the end of the day, the pasta sauce doesn’t taste like just tomatoes. Because you’re adding other things in. And so the more things you add into the analysis, you now start to say, “Does the relationship of the race of the officer and the race of the citizen, for a particular type of force, does it still remain once you know all of these other variables that might affect the ability to use force?”
So the added ingredients in the sauce include things like gender, mental health, and service type? That’s exactly right. So, in table four, if you have a copy of the manuscript, that’s the table that has gender, tenure of the officer, gender of the civilian, and situational level characteristics. Now, those are what we had access to. It doesn’t mean that if you had access to even more other ones that the results could be different. We just don’t know the answer to that because there’s no perfect data source in the world for everything any human being wants to analyze.
Do the ingredients get weighted? I mean, are you adding tomatoes and oregano in equal measure? Now remember, we’re looking at the incident. In our analysis, we’re looking at the 5,000-odd actual cases that involved the use of force. Think about each row in an Excel file as incident one, incident, two, incident three, and each of those columns has all of these other variables. So incident one, you have the race of the citizen, race of the officer, gender of the citizen, gender of the officer. And then tenure and all that stuff are the columns, and then each of the rows are the actual incident. When you do this analysis, we actually weight them at the exact same level. So, no variable in this model gets X more oomph by putting it in there in a certain way. We put them all together simultaneously. So think about the pasta sauce analogy. If I dump everything in at the same time, that’s what we do. So in this case, we didn’t say, “Let’s put in the tomato sauce. Oh, then let’s put in the oregano. Oh, then let’s put in some heat, put in some cayenne pepper.” That’s not how we did it in this study. We put everything in at the same time, which is the traditional norm in the social sciences.
But let’s say I am a bad-acting cop, and I get more aggressive with, let’s say, Hispanic civilians that I come into contact with. If that’s my primary focus, then none of the other factors matter, right? Let me answer it this way, if that relationship held, which is the one you initially elaborated on with the race and ethnicity of the civilian and the officer, then if that still was strong enough, it would still show up after you include all of these other variables.
Were you able to ascertain if there are other factors that were more important in the use of force than race? Not necessarily more important, but they were important. So for example, knowing that the civilian was under the influence of something, that led to a higher likelihood of the officer using an intermediate weapon. Another example would be when the civilian was male, officers were more likely to use an intermediate weapon.
Did you specifically look at excessive use of force, as opposed to appropriate use of force, in these scenarios? No. The police department doesn’t create a code for whether or not it’s an inappropriate use of force versus not.
Presumably, if you were able to separate out excessive use of force cases, you could have a different result for the data. You could. But we do not look at that question. That’s a different question–an important one and an interesting one–but we do not look at that one.
At the conclusion of your study, is your sense that the biggest issue for the Dallas Police Department is a problem of perception as opposed to a race-based use of excessive force? I think that that’s the problem for every police department in the country. As you know, perception guides about everything that people believe. And the technology that we have–you and I remember when there was no cell phones, much less cameras and videos on them. That technology’s very good, but it’s also limited. And we have to be careful when we use that technology that the snippets that we see may not capture the entire event. We may not see the 30 minutes before that or that 30 minutes after that. We may see 10 seconds of that. So I think that there’s a narrative about what police use of force looks like, and I think that when officers misuse force, they should be reprimanded and punished accordingly. But I also think that we need to treat officers as fairly as we do anybody else and amass all of the information before we can develop any sense of finding in one way or the other. And that takes time. A lot of people want an instantaneous decision about why an officer did this or an officer did that. We can’t be quick to rush to judgment. We have to be able to get all of the data, all of the information. That’s not a five minute decision. That’s a very painstaking process, which is unlike an officer’s split second decision to use force.
Why did you exclude deadly force from your study? It’s too small a data set, and it’s also handled independently of how this stuff works. It goes through the special investigations unit, so it’s an entirely different event. So we do not examine those in these data. But it’s a very rare event, and that’s something that’s true not just in Dallas, but it’s also true in many police departments. Deadly force is not the norm when police use force. Most officers never shoot their weapons over the course of their careers, much less kill someone. I’m not diminishing when those things are done, and when a life is lost—especially when a life is lost for illegitimate reasons. That’s a horrific event. But we have to bear in mind that those acts are very infrequent, and that’s true among the majority of police departments in the United States.
One thing that concerned me about the study is that people may look at it and say Dallas doesn’t have a problem with race in its police force, so things like implicit bias training shouldn’t be a priority. Or maybe the training they’re doing is what’s paying off. Police departments, and not just DPD–Vegas, Charlotte, and most big city police departments–have a plethora of different types of programs. Because these use of force incidences that we have seen on video, and the terrible cases that have come out of them, have prompted departments to now seek out these types of training curricula and actually implement them. So I think one of the good things that has come out of these really horrific events and tragedies is that departments are changing the way they’re training their officers.
Obviously, since this study was just completed, there’s no way to compare your findings to how DPD operated prior to its current training regimen. Hard to say. We just don’t know the answer to that.
So, if racial bias tastes like tomatoes, and I add enough ingredients so that I can no longer taste the tomatoes in the sauce, does that really mean it is no longer a biased sauce? First off, it is not correct to call it racial bias. Bias is much more intentional. The literature on both arrests and use of force, when it does find an imbalance in a relationship regarding race/ethnicity, indicates it as disproportionality. We do not know if it is biased per se, because we are not in the person’s head. That is a very big difference.