June 12, 2020

Editor's note: This is a cross-posting of an interview with researcher Rodrigo Canales that originally appeared on the Yale School of Management's Yale Insights blog about his ongoing work with IPA Mexico.  


For the last three years, Rodrigo Canales, an associate professor of organizational behavior at Yale SOM, has led a project studying police forces in Mexico and testing approaches to building more effective and trusted departments. We asked him what the research says about how to prevent racist and violent policing in the United States.

Photo courtesy of Rodrigo Canales

Photo courtesy of Rodrigo Canales

What have you learned about building effective and trusted police organizations in your work in Mexico?

There are two sets of insights that seem especially relevant for the current moment. The first is the importance of conceptualizing police forces as organizations rather than as a collection of individuals. The structural configuration, processes, routines, culture, and values of a police organization greatly determine police officer behavior. Of course, there is still variation in behavior across individuals, but structural factors shift the entire distribution of available, expected, acceptable, and common behaviors. Average cop behavior becomes good. Good cop behavior becomes great. And most important, bad cop behavior is greatly mitigated. But organizational practices tend to be sticky and tend to influence each other, such that changing police behavior requires you to look at the organization as a whole and change not only specific practices, but rather collections of practices, processes, and organizational linkages that create coherent expectations.

For example, our research shows that training police officers in procedural justice significantly improves not only how individual police officers think about policing but also how they behave on the street. But we also find that the effects for beat cops are much smaller if their managers have not also been trained in procedural justice. We also found that any improvements in behavior are constrained by how police officers (and especially their sergeants) are evaluated: if you train cops to behave in procedurally just ways but you only reward them for arrests and reductions in violent crime, then not surprisingly cops have to focus on arresting people rather than on building citizen trust. So if you are serious about improving police officer behavior, you also have to rethink how you evaluate and reward police officer behavior, what you define as individual, team, and organizational goals, and what values you continually reinforce (e.g. “we are warriors who are tough on criminals vs. we are protectors, trusted by our neighbors”).

This is related to the second key insight. We have found that different police forces implicitly or explicitly have identified different actors as their “main customer.” When a police force explicitly establishes that citizens are their central customer, they become much more focused on a) understanding how citizens define and understand their (hyper-local) safety problems, b) building trust with citizens, especially in marginalized communities, which can lead to c) establishing collaborative relationships with neighbors to design better strategies that target those problems, and d) creating mechanisms of bidirectional accountability so a true contract and working relationship is established between citizens and their police. This focus is naturally reflected in routines, protocols, training, and evaluation metrics. And I cannot emphasize the role of leadership in this enough. Ultimately, the chief of police is the voice of its organization, both towards police officers and the outside. Who we choose for positions of leadership, the messages they send and how they send them has a disproportionate impact.

If you are serious about improving police officer behavior, you also have to rethink how you evaluate and reward police officer behavior, what you define as individual, team, and organizational goals, and what values you continually reinforce.

In contrast, we have found that other police forces implicitly define local government as their “customer” (indeed, in general we find that not identifying citizens as the most important constituency leads to this de facto definition). In this perspective, what matters most are not the lived experiences, perceptions, and behaviors of citizens. Rather, police forces become entirely focused on the aggregate metrics of crime: crime rates, arrests, clearance rates, “police efficiency.” Notice that in this narrative, the main focus of the police is on “criminals” and crime. All processes, protocols, training, metrics, and evaluations are designed around that very small minority of people (some of whom are dangerous!). And the more this happens, the more police forces turn inward in an “us vs. them” rhetoric.

I want to be very clear: I am in no way suggesting that metrics and evaluations of performance are not important. Rather, what we observe is that police forces have different objectives, many of which are in unavoidable tension (e.g. we need to have the ability to deploy force but we can only do that effectively if citizens trust us). And when those tensions are not explicitly acknowledged and then woven into the processes and systems of the organization, then one (usually the deployment of force) tends to dominate over others (usually citizen trust and accountability).

Do you see instances of police violence and racism as stemming from the actions of a few bad apples or as systemic problems?

This dichotomy is extremely prevalent. But it is also false. First, it is empirically true that practically all instances of police misbehavior are driven by a very small number of problematic police officers. The overwhelming majority of police officers do their work carefully, respectfully, and with a sincere commitment to the citizens they serve. (Incidentally, the same is true of crime—most crime, and especially violence, is typically driven by a very small number of high-risk individuals, who tend to belong to a small number of high-risk groups, in few high-risk places). At the same time, it is only within permissive systems that a small number of bad cops can continue to behave poorly. Research has shown that most instances of severe police misconduct (e.g. the murder of an innocent black man) are conducted by police officers who have received numerous complaints throughout their career, including for violent behavior (the police officer who murdered George Floyd had 18 recorded complaints in his record).

Police work is extremely difficult—probably the toughest job I have seen. It is to be expected that some people who look good on paper and in the academy will not know how to use their power and discretion well. Police organizations, therefore, must have the systems and protocols to routinely identify officers who are not living up to their mandate. Some may need additional support to get back on track (e.g., every police officer I know has significant trauma; very few ever seek help for it, and for some of them it turns into a severe hindrance to their work). Some may need to be removed from the organization. And the organization, as a system, must show absolutely no tolerance for police misconduct. If it does not, then the message it sends to all its officers is extremely loud and extremely clear.

There are no quick fixes. There are no simple, one-off interventions that are going to fix this.

There is a telling moment in the video that shows the arrest and murder of George Floyd. One of the younger officers on the scene is clearly uncomfortable with what is happening. We can see the hesitation in his face. But he does not take action. And that tells you a lot about the organization that he is a part of. As Chris Rock used to joke, “bad apple” is a very generous name to give to a murderer. “I have eaten bad apples. They are tart. They don’t choke me to death.” And there are certain jobs where you simply cannot allow “bad apples.” Police and airline pilots are two examples. The consequences are just too grave.

Are there lessons from your research for addressing racism and violence from the police in the United States?

Yes. Change is possible. We have seen police agencies radically transform themselves in relatively short periods of time. We have seen them go from oppressive, corrupt, violent organizations to true civic agencies that have the trust and collaboration of the citizens they serve. But this has only happened when a) there is a fundamental reframing in the mission of the organization (who is our central “customer”); b) the organization is conceived as a system of interconnected practices, protocols, systems, culture, and values that need to be in coherence; c) there is significant involvement from civic society in a constructive, collaborative approach (including local business and civic leaders, neighbor groups, educational institutions) that nonetheless has clear mechanisms for accountability; and d) there is political support from the top leadership to ensure change and continuity across political cycles. There are no quick fixes. There are no simple, one-off interventions that are going to fix this.

Police forces should never tolerate police misconduct and we should hold them accountable to that. But police officers should not be framed as the enemy either. Just as police officers should, unequivocally, be held accountable for bad behavior, police forces and citizens must also make it clear that we will always have the backs of those police officers who act in good faith and in accordance with our shared values. We can only truly achieve change when both sides, the police and the citizens they serve, see the other as a necessary partner. We can only do this from a place of mutual respect and empathy. We can only do this by listening to and working with each other. We can only do this together.