Using Data to Improve Public Safety: An Interview with Kim Smith

The Economic Mobility Catalog’s Strategy Guide on Evidence-Based Policing provides guidance for local leaders seeking to improve public safety through institutionalizing data-informed decisionmaking, partnerships with community-based organizations, and innovative policing tactics. It was created in partnership with experts and practitioners in the public safety field, including Kim Smith, Director of National Programs for the University of Chicago Crime Lab and Education Lab.

In this edited interview, Kim shares findings from the Crime Lab’s research that are improving law enforcement practices, supporting data-driven operational strategies, and increasing public safety.

What are some of the most effective ways that local governments can use data to improve public safety?

Having the data is great, but once you have it, what do you do with it? What’s really promising are programs that use data to identify points for warm handoffs from law enforcement to other social service providers, offering a supportive rather than punitive intervention. We’ve seen this with the Narcotics Arrest Diversion Program (NADP) in Chicago. The NADP changed how police responded to substance use calls or related calls like loitering. Basically, instead of arresting the individual, they’re passed on to a treatment provider located in the police district. This program connects the individual to medical treatment or housing assistance, while still addressing the community’s 911 call and saving police time and resources.

Another way to leverage law enforcement data to connect individuals to supportive programming is in creating warm handoffs with violence prevention programs. The Crime Lab and several of our research affiliates analyzed 10 years of crime and victimization data from the Chicago Police Department to predict an individual’s future risk of gun violence involvement — not just being a victim, but participating in it — so they could be prioritized for social services. In our case, we referred them to READI Chicago, a violence prevention community organization. READI found these individuals’ last known address, tracked them down, then offered them job training, a place to live, or other support services. This doesn’t address the larger problem of disinvestment and guns coming into these communities, but it’s a way to redirect people at the highest risk of violence involvement without direct intervention from law enforcement.

What are some insights from your research on police management?

In response to the 2016 surge in gun violence in Chicago, I was involved in the implementation of Strategic Decision Support Centers. This intervention embedded civilian analysts in the Chicago Police Department and helped them implement new management strategies in high violence districts so commanders could use their resources more effectively.

The SDSCs helped commanders create patrol assignments based on data and prioritize violence reduction at a time when there was a lot of pressure to reduce shootings. This process started with daily briefings at 6AM where the district commander would review gun violence data from the past 24 hours, seven days, and 28 days. Using that data and other intelligence, they would set "missions," specific assignments for officers, like patrolling a high school between 2 and 3PM due to recent fights, and then following up to ensure the mission was completed.

Beyond Chicago, our team has also looked at leadership across large police departments nationally and found that who's in charge really does impact what happens in a city. A commander’s vision, whether they intentionally convene community stakeholders, or if they hold their officers accountable, directly impacts policing outcomes. An evaluation of the SDSCs found that in the first three months of the intervention, the highest violence police districts saw statistically significant declines in shooting victimisations (-21%) and violent felonies (-8%), with no detectable changes in overall arrests, traffic stops, or uses of force.

How does your research on cognition and decisionmaking help improve policing?

Policing is a risky business, and on a given day, officers must make several high stakes decisions in a short amount of time, often with limited information. They might be primed, based on past experiences, to assume worst-case outcomes in certain situations–for example, when engaging a suspect, they might assume that a gun is present. These assumptions inform how police act, and their actions can have life or death consequences. The question then becomes: how do you manage officers' approaches to risks?

Working with Oeindrila Dube, Anuj Shah, and Sandy Jo MacArthur, our team at the Crime Lab partnered with the Chicago Police Department to evaluate Situational Decision-Making training, or Sit-D training. This training gives individuals the tools to reassess their approach to risk and identify alternative pathways in navigating high-stakes situations. Officers learn to recognize stressful situations, cognitive biases, and emotional triggers. We’re not trying to rewrite the police training manual – Sit-D aims to help officers make better decisions faster.

In our evaluation, the Sit-D training was effective in reducing discretionary arrests for black residents for things like disorderly conduct and obstructing an officer, and in reducing the use of force overall, all without decreasing desirable outcomes like firearm recoveries.

Many local leaders are navigating the intersection of addiction and law enforcement. What does your research say about how communities can confront these challenges?

This is a significant issue for communities, and I want to go back to the Narcotics Arrest Diversion Program (NADP) to explain it more because I think it offers a promising model for how law enforcement can handle substance use issues.

The real issue NADP addressed is that lower level narcotics arrests were a strain on police resources. The police were required to respond to resident calls about drug use and loitering, which often led to an arrest of individuals experiencing addiction. This resulted in officers having to sit with individuals going through withdrawal at a hospital before processing the arrest — a process that wasn't helping anyone.

With NADP, the officer still responds to community calls for assistance, but instead of arresting someone, the person is connected to a local service provider who is much better equipped to deal with substance misuse and addiction. When we evaluated the program, we found it to be effective in reducing violent crime and re-arrest rates. Plus, it freed up officer time and successfully connected individuals with treatment.

What is something in the public safety field that you’re excited about?

The good news is that we are seeing a move towards more precision in public safety, not just with local law enforcement but with non-profits and others who work in public safety. Increasingly, organizations are using data to identify high-risk places and individuals. The challenge now is to combine this precision approach – the “who, where and when” – with efforts to implement evidence-based interventions in the places and among the populations most in need – “the what”. I’m also interested in how we might leverage existing data and machine learning (AI) in this space – whether it be to better understand public perceptions of public safety strategies, and to proactively identify officers who might be at risk of adverse outcomes for themselves or the residents they serve.