Outcome-oriented Policing

I had been summoned.

The Chief of Police whom I had been working with closely for several months asked me to swing by his office to discuss some concerns about our recent traffic crash reduction strategy. I was surprised to hear that anyone had “concerns” because there had been a recent public outcry over increasing traffic fatalities, the strategy was fairly simple and it involved fewer police interventions in the form of stops and tickets.The call ended with receiving no further information and I scheduled a time to drop by. 

The area of Town Hall where the police executives sit was usually fairly jovial. The administrative assistants, officers and executives had an easy-going manner about them which I enjoyed. When I walked into the reception area that day though, there was a noticeable air of tension. That sort of thing wasn’t unusual in the wake of a high profile controversy, during times of vitriolic community criticism or right after some law-enforcement related tragedy.  Traffic crash reduction though shouldn’t cause that thickness in the air, I felt like I had to be missing something. What in the world was I walking into? 

The receptionist waived me through and I walked past her desk and into the hallway of executive offices. I opened the door to the Chief's office and saw him seated across a table from the complainant- the Traffic Unit sergeant. I had very little interaction with this man before that day but I had always found him to be a cop’s cop in a very traditional sense. He kept his haircut high and tight, the same way he had it when he left the military some fifteen years before. HIs uniform was impeccable with dozens of medals for valor and excellence hanging off his chest.  In a certain light, his boots shined like a beacon. He had a stern, stoic expression but everyone who worked with him described him as calm, non-reactive, understanding and a paragon of professionalism. The officers he supervised loved him.  He was a hard worker who routinely led the department in proactive activity. He was a pillar of the community and volunteered his time to several local charities. If your grandmother had to call the police, this was the type of man that you wanted to show up. 

The three of us sat down at a small conference table. The sergeant and I exchanged pleasantries, although his stern demeanor didn’t crack.  I opened my computer, surely it would be needed for any discussion of our problem-solving project or analysis thereof. To my surprise, the sergeant also came armed. He opened a folder that had printouts of statistical reports I had generated over several weeks. Since beginning my work with this agency, I had taken dozens of requests for regular reports officers and decision-makers wanted to see. I automated these reports in the form of dashboards, spreadsheets, maps, and other products and published them to the agency’s intranet. Some of them I referred to regularly. Most of them were bespoke products for niche interests of officers and decision-makers and I didn’t invest much time tracking who was using what. The sergeant didn’t strike me as a particularly “Data-driven” guy, so to see him ready with months of reports printed out threw me even further into a loop. 

The reports he put on the table measured officer output over time and in various time windows. He calmly explained that he had taken over the traffic unit just over a year before and, as my reports proved, productivity had shot through the roof. He had taken a once stagnant unit and turned it into one of the most productive in the city. They had increased the number of traffic stops and tickets (to include warning tickets) many times over. His impact hadn’t just been in traffic either. His unit was responsible for serving hundreds of felony warrants, seizing guns and drugs and even intervening in a highly publicized human trafficking case that earned local praise from the media. “All of this” he explained “came as the result from stopping cars.” He put an exclamation point on it, “The more cars we stop, the more good we do.”

With that last remark, I finally understood the controversy. Several months prior, a fatal traffic accident had drawn the media’s attention to the growing number of similar tragedies city-wide. This was far from media sensationalism,  it was a serious issue at the time. Year to date fatal traffic crashes showed steady increases year over year for the previous ten years. I went to work studying the pattern of traffic crashes and found several root causes including new traffic patterns, construction, increased population and other factors. Fortunately for us though, this area’s traffic problems were fairly predictable. With a population made up mostly of commuters, most crashes occurred during rush hour periods. Monday through Friday, from 7am-9am and from 4:30pm to 7:00pm accounted for nearly all crashes. In addition, the unique roadway situation created predictable geographic locations for these crashes as well. 4 major intersections were host to a vast majority of traffic crashes and almost all of the fatal crashes. 

These simplistic insights led to our first attempt at intervention.  The Sergeant enthusiastically adjusted his team’s schedule to be  in those places, at those times. The officers complained, of course. Most of them had grown accustomed to working nights, weekends or day shift. The rush hour timing meant that many of them would have to adjust to a split-shift. That meant they would need to arrange for someone else to take their kids to school or pick them up. They lost time with their families. They had to drop training classes they had been previously enrolled in. They had to give up previously scheduled overtime shifts that brought in badly needed income. No, this data-driven adjustment was not easy for anyone but to this sergeant’s credit, he was unflinching in his decision. 

I was exploring the possibility of other problem-solving approaches- such as changing traffic light timing, trimming brush to increase visibility , removing on-street parking or even putting in speed bumps- but I really believed that the Traffic Team’s adjustments would show improvements in the short-term. To my surprise, the crash numbers kept rising. I had to see what was going on. I stepped away from my air conditioned office, my big leather chair, my bougie latte maker, my dish full of candy and my wall mounted TV that played “The Office” reruns on a loop and I went for a ride along with the Traffic Unit. 

They were exactly where we told them to be, exactly when we told them to be there. All of the officers on this team were go getters for sure. Within 5 minutes of signing on, the officer I was with had already stopped a vehicle and would shortly develop the probable cause needed to search and find drugs in the center console. Their group chat was a constant source of entertainment as each of them attempted to one-up the others with more stops, higher charges and better arrests. In my experience, the most pressing issue that cops face at the beginning of every shift is deciding where to eat. That becomes the focal point of many discussions and deliberations until the time comes to finally pull into a favorite restaurant and hope that the radio stays quiet. That was not the case with these guys- they worked hard the entire shift with absolutely zero loss of enthusiasm. It was encouraging to see. 

I did; however, see something that bothered me. Our four locations all had one unique thing in common. They were all flyover streets meaning that when an officer turned on their lights, most drivers would pull off the street via an exit ramp that led them down to ground level where there was less tightly condensed traffic. It was actually a pretty considerate thing to do, it made the traffic stop much safer for all involved. I saw it as a problem though. While stopping a motorist can impact that driver’s behavior, we were trying to affect an aggregate traffic pattern. We were trying to cause so many drivers to drive more carefully that fatal crashes went down. When an officer stopped a vehicle, they took themselves out of the line of sight of the rest of the cars in traffic who would inevitably commence driving like we all do when cops aren’t around. 

My “intervention” that had drawn so much eyre was a simple suggestion. During those rush hour periods, in those 4 locations, instead of conducting traffic stops, just sit in the narrow shoulder with your lights on. This would cause some delays of course, as people tend to tap their brakes when they see cops with their lights on in traffic, but I was hopeful the extra caution would also save lives. 

Then there was the other obvious consequence of this strategy. Traffic stops, tickets, warnings, seizures and arrests would go down.  This sergeant could no longer point to his productivity to demonstrate his effectiveness. He could no longer take pride in having taken a unit in disarray and turned it into one of the most productive in the city. He could no longer motivate his officers by gamifying how many warrants they could serve or guns they could seize. My “intervention” didn’t draw any criticism from the officers when it required them to give up their schedules, their family time, their hobbies, their side jobs or their career development. I only drew criticism when I suggested a focus on outcomes as opposed to outputs. In retrospect, I understand that this wasn’t such a simple suggestion. This sergeant had related his output to his professional identity and by asking him to disregard output, he interpreted that as me asking him to accept a lower standard, a failure, or perhaps even a disgrace. His integrity wouldn’t allow him to accept that, so there we sat, in a sea of my statistical reports, arguing about traffic stops. 

Analysis is easy. People are hard. Unfortunately for those who wish to drive change in law enforcement, spreadsheets, dashboards and maps don’t put on uniforms and investigate crime. I remember sitting in a classroom and reading about early attempts at data driven problem-solving in academic journals. They painted a shiny picture and a promise of tremendous crime reduction. For the most part, the early pioneers of Crime Analysis did something right because you wouldn’t be reading this blog today if they hadn’t. There is a stark difference though between partnering with a chief and running a research project as an outsider and being the analyst sitting at a desk from 9 to 5 every day trying to push an agency that has operated largely the same way for a century towards being truly data-driven. That type of work involves more than COMPSTAT, regression models, journals and data. It involves confronting closely-held beliefs and attitudes and when we dare to push on those soft spots, the pushback can be extreme.

The sergeant’s feelings are not uncommon in law enforcement. The idea that police can and should be responsible for reducing outcomes (Crimes, Traffic Crashes, etc) is still relatively new. This was the hallmark of the NYPD’s COMPSTAT process that started in 1994 and it was completely novel at the time. The sudden requirement to affect crime rates did not sit well with police commanders who had never been accountable for outcomes before. Over two-thirds of the commanders in NYPD were gone 18 months after the start of COMPSTAT. Reports have since come out that many Commanders altered their data in order to avoid the vitriolic form of accountability that characterized the first COMPSTAT meetings. 

Getting police to focus on outcomes opposed to output is an on-going battle. Putting together long-term problem-solving projects, partnering with the community, CPTED and consistent accountability are hard. Its much easier to surge police presence and appear on TV with a table full of seized drugs and guns. When a hot spot pops up or the media runs a story of spiking crime and disorder, it's easy to just go kick ass for a weekend but its unlikely to affect sustainable crime reductions. Its also tempting to suggest that the answer is to obtain a shiny new toy. While fusion centers, real-time crime centers, shot detection systems, LPRs and other technology can be used to effectively impact outcomes, all too often they are simply engines that further drive outputs. 

I would like to think the remedy for this problem is strong leadership but when I think about it, I just can’t stop thinking about the influence of buzzwords. When I walk through the vendor halls at any conference, you can see the hottest buzzwords plastered on booths, signs and fliers. Companies advertise “real-time analytics” that incorporate “Artificial Intelligence” in a “solution” that provides “actionable insights.” These words “work” in a marketing sense largely because people don’t really understand them but they’ve gained general acceptance as something approximating a “best practice.” I have read Requests for Proposals (RFPs) where large law enforcement agencies have described their desired systems with several nonsensical paragraphs that read as if they drew random buzzwords from a hat and stitched them into sentences. I frequently find myself reading several pages only to be no closer to figuring out what the potential customer wants than I was when I started, but I know that they want it to be “real-time”, “actionable” and incorporate “machine learning”, “fabrics” and “the cloud” but also to be installed “on-prem.” Imagine my shock when I walked into a large police department who had recently won a million dollar grant and wanted to enlist my services to execute the grant requirements. In their application, they had stated that they would use risk-terrain modeling to identify priority offenders. Those familiar with risk terrain modeling know that its a technique to identify problematic environmental features, not a practice that prioritizes human beings. Still, shoving that term into a grant application was all that was needed to see a seven figure check land in their account. 

Buzz words also seem to be an engine that drive academics and many of them seem to make their careers turning their theories into buzz words or iterating the buzz words created by others. My favorite example of this is the PANDAS model. Most of us are familiar with the SARA model but in an attempt to create a “new” model, these steps were iterated upon into a new, fun acronym. How different are these approaches exactly? The “S” in SARA stands for “Scan” but the “P” in PANDAS stands for “Perform Scan.” Of course, they’re not the same, if they were, one of them would be pointless. 

Academia has fed us a lot of buzzwords. What model of crime reduction is your agency running? Is it “Problem-oriented Policing”, “Intelligence-led Policing”, “Community-Oriented Policing” or perhaps “Stratified Policing?” Some of these are great frameworks and others are clearly an attempt to rephrase the same idea with a new buzz word.  Some of them are relatively obscure but others are burned onto the lips of police chiefs across the country who will repeat their buzz word of choice every chance they get. I am neither a Marketing genius nor an academic so I am issuing an impassioned plea to those that are- use your powers. Publish a journal, write a book, make a spam video that pops up on my Tik Tok, or whatever it is you do to make the phrase “Outcome-oriented Policing” as commonplace as some of these others.