Why it’s so hard to find accurate policing data

Old IT systems and incomplete data sets make it hard to track things such as arrests and use of force.

Why it’s so hard to find accurate policing data
[Photo: DiMaggio/Kalish/Getty Images]

Americans are obsessed with data—from daily COVID-19 infections, to the presidential horse race, to unemployment numbers. But starkly absent from these daily accountings are policing figures. Instead of comprehensive data on arrests or use of force, we get individual stories about police officers abusing their power, using excessive force, or being fired from one department only to be hired by another.


It would not be misguided, given the current climate, to think that more data exists but is being concealed from the public. Surely there’s a database to show how pervasive these stories of abuse are, or which departments have a history of applying undue force in communities of color. And surely police departments have the budget to use the latest data tracking systems. Given the pictures of police spending excess—the $10 million for unused electric BMWs, the $700,000 Mine-Resistant Ambush Protected vehicle owned by an Iowa sheriff’s department—it seems like the police must have the money to adopt state-of-the-art IT systems. For better or worse, that picture couldn’t be farther from reality.

Across the nation, police departments are as backward as the rest of government when it comes to technology, having been bypassed by many of the private-sector innovations that occurred over the last 30 years. This has left the police with cumbersome systems and limited, often incomplete, data sets, making it difficult to adequately track arrests and use of force—as well as the outcomes of police intervention, both positive and negative.

The police suffer from many of the same problems that have kept government technology mired in the past, the biggest of which is the lack of in-house technical talent. The police are beholden to whatever the few big police IT contractors develop, often unwittingly ending up locked into a single contractor for all of their technology.

Government agencies are riddled with vendor lock-in, which means that vendors can charge government clients exorbitant rates for any small tweaks or changes to their existing technology. Imagine if the electrician who wired your house also charged a fee every time you needed to change a light bulb. Now imagine that you weren’t contractually allowed to talk to other electricians. To make matters worse, in some cases, vendor lock-in also means that vendors control the data stored in their system.

Regardless of which vendor they’ve gone with, most police departments have two separate computer systems. One system dispatches officers to a potential crime scene based on 911 calls. If someone calls and says they hear gunshots, the system records the call and logs it in a database. As one person familiar with police IT systems told me, the computer-aided dispatch system is usually “the best in 25-year-old technology.”

The other system is a records management system, which stores arrest reports and final 911 dispatches. So if the officer who went to investigate gunshots discovered that they were actually firecrackers, that is logged in this second system. Typically these are two different, totally separate systems, sometimes purchased from two different vendors. They very often don’t talk to each other. Which means each system only tells half the story. So when it comes to pulling data from the systems, it’s often incomplete.


[Photo: MattGush/iStock]
To make matters more complicated, the records management system relies on arrest data, which can vary widely by jurisdiction. In some cases, officers are required to include narrative descriptions of the incident, a description of the statute the person is charged with violating, location details, and demographic information. But arrest data can be tricky to capture correctly. Police officers might be simultaneously managing a crime scene and jotting down a sketch of the event as it happens. Only later, when they go back to their desk or pop open a laptop, are they able to fully transcribe the events, relying on their scribbled notes and recall of what may have been a highly charged, fraught situation. Depending on the technology their department uses, some officers might have a system for logging arrests in their car that isn’t cloud-based, so if they start writing an arrest record in the car, they then have to finish it in the car before they can save it and move on. Not ideal for creating clean, thorough data.

“The very nature of policing means that the data coming into the system, if it’s not automated or the system isn’t user-friendly, will always be lacking,” explains Clarence Wardell, who cofounded the White House Police Data Initiative in 2015, an effort to encourage police departments to use open data to increase accountability between communities and law enforcement.

His cofounder Denice Ross notes that police departments are beholden to vendors in the way they both collect and analyze data. Because police departments can only purchase what’s on the market, and because they’re often locked into long-term contracts, Ross and Wardell believe the push for more streamlined systems needs to come from vendors, who have a lot of power in the direction policing takes, in part due to the limited number of vendor options. If vendors made it clear how transparency benefited police departments, more departments might move in that direction.

But first, police departments need to see the value in upgrading their technology. This is another place where technological defects in policing mirror the rest of government. By and large, when police departments think about technology, they think about facial recognition, body cameras, tasers, or drones. At the industry’s annual conference, hosted by the International Associations of Chiefs of Police, people gravitate toward trying out the cool new toys. No one wants to look at the awesome new computer program that creates spreadsheets and pie charts—and no one wants to spend money on it either.

In part, this lack of focus on the technology that could make policing more transparent, equitable, and honest is because the leadership doesn’t exist. In some larger departments, there might be a chief technology officer or a data scientist, but the overall ethos at a law enforcement agency is to bring order to chaotic situations, not crunch numbers.

A former fellow who worked with the police department in Vallejo, California, on technology modernization efforts reported being taken aback by how militarized the department appeared. An armored military truck was parked at the city’s only police station, and the department’s recruiting video featured glory shots of assault weapons and officers breaking into a residence with guns drawn. A lot of the police force were ex-military personnel.


“I believe that Vallejo’s police department wanted to make real changes, and to build trust with their community. But technology is just an implementation tool. Without changing the culture of the organization, technology only amplifies the vision that’s already in place,” the fellow said.

Not only does the lack of technical expertise in law enforcement make policing less transparent and less effective; it can also be dangerous. As a widely shared New York Times article reported, law enforcement has begun to depend more heavily on facial recognition systems, without an understanding of what the software’s algorithm is really telling them. In particular, the technology has difficulty correctly identifying people of color. But that hasn’t prevented law enforcement agencies from using it to make arrests.

“The use of facial recognition software by law enforcement will simply lead to them acting on the decisions made by the software. This has the effect of further removing law enforcement from any sense of accountability for their actions, since they can just say, ‘the software did it,'” said Brandon Bouier, a Black man who is a former IT analyst with the Seattle Police Department. “Facial recognition software in the hands of law enforcement puts my life in danger. Period. Full stop.”

There are vendors who are working to change the current situation; Wardell and Ross point to Mark43 as one of them. The startup’s cloud-based solutions are used by 80 departments around the country, and Jeff Bezos and General David Petraeus are among its list of funders and advisers. But it is still too early to determine the company’s efficacy with regard to policing.

Though Ross and Wardell are both strong advocates of police reform and recently authored an essay calling for radical transparency in police data, they both worry that calls to cut police budgets could end up setting police technology back. Police departments put most of their resources into personnel and see IT as a capital cost, so any cuts are likely to hurt the IT budget rather than reduce the number of personnel. And with COVID-19 decimating city budgets and major cities moving funds away from the police force, the IT modernization that could bring about more transparency and accountability likely won’t happen.

Police departments may not see the connection between improving their technology and improving their policing. But those that have been forced into improving their data collection systems have learned that the new system makes the work of policing easier. In 2001, the police department in Wallkill, New York, was placed under a consent decree. This happens when police departments have a record of consistent violations, and it puts the department under the Justice Department’s supervision, requiring the police department to collect and publish data that holds it accountable for improvements.


Among other requirements, the consent decree dictated that the Wallkill police department publish data on its traffic stops. In Wallkill, the traffic stop data was a mess, with fields only partially filled out or omitted completely. But because they were required to find a way to make the data more complete, the department purchased a system that allowed officers to swipe driver’s licenses to populate the fields. Suddenly their data got a lot better, as did their accountability. The department and the public could see who was being stopped and why. In 2006, the consent decree was lifted.

The coronavirus pandemic has highlighted how integral technology is to fulfilling government’s mission. When unemployment sites around the country crashed because of antiquated systems, the delay had a real, devastating impact on those who urgently needed the money to pay rent. Though we have yet to see any real changes yet, the complete failure of technology to serve people at both the federal and the state level at least functioned as a wake-up call. Turns out technology is important for things beyond texting and video games. Really important.

Now it’s the police department’s turn.

As cities across the country raise their voices against police violence, we learn more every day about how much policing suffers from a lack of transparency—whether it involves undue force or repeatedly hiring serial offenders. And while the opaque nature of policing may be dangerous to citizens, it isn’t working for police either. Applications to police departments across the country have dropped dramatically, even before this summer’s unrest. A 2019 study by the Police Executive Research Forum found that the skills police officers need to have today include fluency with technology and the ability to handle social issues such as mental illness or addiction, but that most recruiting videos still feature people rappelling down buildings and breaking down doors.

The first step is shifting the mindset inside policing. Police departments need to understand how integral technology is to effective modern policing. That understanding won’t come from spending money on whizzy new toys. It can only come from bringing tech expertise in-house. There is a real need for community-minded Americans with technology skills to help repair policing. Not with drones or facial recognition, but with care, transparency, and common sense.