Data-driven policing has been in Alameda for a year. Chief Joshi explains how it works.
Chief Nishant Joshi is all about the numbers. Since assuming command of the Alameda Police Department two years ago, the Chief has instituted data-driven policing policies to help predict and prevent crime. Departmental data are turned into figures, lists, and heatmaps showing concentrations of criminal activities. He sees it as “a total benefit to the department.”
Joshi came to Alameda after almost a quarter century working his way up the ranks, most recently as an assistant chief in Oakland. He recalled that while policing the streets more than two decades ago, despite making repeated arrests and issuing citation after citation during his shifts, the efforts had little impact on major violent crimes.
He related an example of previous misdirected efforts by police officers, describing a person who lives in “…a neighborhood where there’s heavy crime, but they happen to be on their way to a job interview. They get pulled over, maybe because they didn’t come to a complete stop at a stop sign. That is absolutely a violation. But… there’s a certain amount of time and resources that it takes to stop that person, issue a ticket, or issue a warning, confirm that their license is valid, confirm that they have ID—all those things take maybe 30 minutes. Well, if you think about that, if you have several of those types of [stops] and you write [each] person a ticket, what real impact did you have on shootings in a particular neighborhood?”
New Chief, new policy
Chief Joshi implemented data-driven policing in Alameda last June, which brought with it a steep learning curve. It took some time for staff to fully understand and embrace the changes, he said, but now everyone in the department is up-to-date on criminal activity and trends in the city. The biggest surprise from the analysis was the low level of violent crime in the city. “You have a 97% chance of not being a victim of violent crime” in Alameda, according to Joshi.
The Chief discussed how data-driven policies changed his work as a police officer. “Five years in a row, it worked. It reduced violent crime year after year after year, when I was an area commander.”
While he acknowledged that Alameda doesn’t see the same levels of violent crime as Oakland, Chief Joshi suggested the same principles apply to the island city. Theft-related crime, such as larceny, motor vehicle theft, robbery, and burglary, accounts for 95% to 97% of all of Alameda’s Part 1 crimes, which also include murder, rape, assault, and arson.
Studying crime data to find patterns
“If you truly want to have an impact on at least the quantitative data associated with crime, you look at the drivers of crime and then you focus on those things,” the Chief explained. “Theft is the priority here. We don’t have a lot of murders. So you look at all your crimes, look at which are the common trends, which are the frontrunners and then, to impact crime in Alameda, those are the ones that you focus on. Now, what do you do with data? You want to figure out the ‘who, what, where, when, why.’”
He explained that for criminals to be successful, they need to perfect their strategy and methodology. They figure out the gaps in police coverage and hone in on them to improve their chances of accomplishing their task and escaping without arrest. Reviewing the data of these successful crimes, is “A lead for us, as law enforcement, we need to be in these locations.”
As officers collate data from different crimes, citations, stops, and arrests, patterns become clear, and the data becomes actionable intelligence. “…This red Honda continues to do these auto thefts and it’s during this time and it’s at this place. So we then strategically will place police officers in those area… If you’re in the right place at the right time, [the] idea is that you could potentially stop a crime or take action.”
APD staff regularly prepares multiple reports to keep track of crime activity and statistics: a 24-hour report, a 72-hour report, a 96-hour report, a two-week report, and a four-week report. This variety of reports allows officers to compare both short- and long-term trends. They also compare data from previous years and other cities in the region to review trends.
Catalytic converter theft a major focus
Chief Joshi shared that although “Alameda is one of the safest cities in this region” and we have less crime than other communities, the crime rate is not decreasing. He pointed to catalytic converter theft, an issue of concern to many Alamedans, saying thefts in the region are up more than 100% because of how lucrative the sales of the stolen items can be. He compared that to a rise of 6% in catalytic converter thefts in Alameda.
“We specifically focus in on catalytic converters. We know where these crimes are occurring.” The Chief continued by alluding to some positive news he is planning to release in the coming weeks, explaining it was due to, “…using data and then using the tools that we have, like the license plate readers, and then putting that into action that is fine-tuned on a daily basis.”
Joshi pointed out that current staffing levels are 30% below what they should be, and for his data-driven policing policy to be successful, he needs a full complement of staff. “You can have all the tools, you can have all the toys, but operating them… People solve crimes. LPRs [license plate readers] give you leads, but it’s actually people who solve the crime.”
City divided into sections for equitable coverage
The Chief explained how he has divided the city into three sections to create a more manageable and equitable division of coverage. The West side of town accounts for 33% of service calls, middle island including Park Street produces 35% of calls, and 32% come from the East side, including Bay Farm Island. Each section has a Lieutenant responsible for everything in that section. “In essence, they’re chiefs of police of their [own] small little city,” he said.
The map further breaks down each section of the city into four beats, creating a numerical system that identifies each one with a two-digit number, like 12 or 31. This system is called geographic command. An example would be the location of South Shore Center, which is in sector 23, middle island, beat 3. “We have 12 beats, but at best we can have four to five police officers on so, of those, we make sure they’re spread out in these locations… Our response time is one minute and 40 seconds… The national average is under seven minutes. So, Alameda’s really good at responding to calls for service.”
“I have a lofty goal of once I get the people in these seats—which I’m hoping [will happen] by early next year—that the following year we are actually the safest city in Alameda County.” Currently the department has received 150 applications to fill 24 vacancies. Joshi plans to have an officer on every single beat when that happens.
Data comes from police reports, calls, and the community
APD uses data for its reports that comes from reports officers make, calls for service, and information from the community. Data from license plate readers are only reviewed for a specific case investigation. Those tools are considered reactive, and reports do not include their data. Per department regulations, license plate reader data is only kept for 30 days. Other data that informs reports can be kept for years.
A crime analyst, who started with the department at the beginning of 2023, oversees the data. She generates reports every few days using CompStat, software originally developed under Chief William Bratton by the New York Police Department. “He was many light years ahead of most other agencies,” by taking a business approach, said the Chief. He said the way investors research their stocks before purchase to improve their returns is analogous to how CompStat provides the opportunity to prevent and interdict crime.
Is the juice worth the squeeze?
“You can’t stop and enforce everything. You have to be smarter. You have to have a small footprint [with the] biggest impact… I can’t put my police officers and the reputation as an organization at risk by saying ‘Go out there and just write a bunch of tickets and arrest everyone.’” Said Joshi, “It’s ‘is the juice worth the squeeze?’”
He continued to explain that the shift toward data-driven policing came out of the sensationalizing of police work in TV and movies and marginalized communities learning their rights, coupled with police finally understanding their methods having a negative impact on communities. He also spoke about the benefit of having officers on the force today who have deeper critical-thinking abilities, and how humans are wired to want to get better.
Joshi displayed some of the department’s reports, each color coded to show the type of crime or offense, whether an arrest had been made, and which beat was affected. Some of the reports include heatmaps, showing the location and concentration of activity. This information is then used to determine staffing for each beat and the on-duty officers’ tasks for the day.
He provided an example. “Look at this—sector 2 and Beat 23 is generally high, but this time it accounted for 19% of the total crime. If I was a watch commander, I’m going to say I want all my resources focused on Sector 2 beat 23 during these times, right?”
Criticisms of data-driven policing
When asked about criticisms of data-driven policing, such as bias confirmation, and other downsides, the chief underscored the level of anti-bias training his officers receive, and that the data informs the outcomes. “We make sure our police officers are aware of their bias… You have to act on actionable intelligence. There’s also [a] legal threshold of probable cause or reasonable suspicion you have to meet. We’ve increased our level of oversight.”
“If there’s a use of force, depending on the type of use of force, it gets reviewed all the way up to the Captain, and then it goes across to another completely separate chain of command, and they review the circumstances of it to ensure that there wasn’t bias that was baked in…”
Body camera software flags bad and good police behavior
He went on to spell out the value of the recordings from the cameras all APD officers wear, the first department in CA to do so. “Our body-worn cameras record thousands of hours of footage… We got this software that…flags certain keywords, like you know if somebody uses bad language or somebody says, ‘Turn around and put your hands behind your back.’ Or ‘Thank you, sir.’ It flags good and bad behavior… We build on both.”
He detailed how the shift helps them find the correct suspect. “If a robbery happens involving a gray Honda Civic. Let’s say it happens at South Shore, at the Kohl’s, and officers flood the area. They’re going to stop [any] gray Honda Civic. And they’re legal and justified in making that stop. And if it’s an armed robbery, they may pull that person out at gunpoint and it’s all legal and justified. The problem with that is, what if that’s the wrong one? My son drives a gray Honda Civic, and if he were to get pulled out at gunpoint, I would say ‘I’m sorry that that happened, but it’s legal and it’s justified for the officers to take those precautionary measures.’
“Now we have this technology that will say, ‘Don’t stop that one.’ It records the license plate, and it will tell you which one to focus in on. I’d rather stop the gray Honda Civic rather than any gray Honda Civic… I think that this is the way to operate.”
Perceptions don’t always match reality
Perception is as important as data to Chief Joshi. He stated that he shares information about cases when he is looking for leads or wants to warn the community about a potential threat. He’s also keenly aware of community feelings about crime. When it comes to catalytic converter theft, for example, “We had a 35% decrease,” he said. “I’m very keenly aware of qualitative information and quantitative information. Perceptions are huge. I think there was this perception in the ’80s and ’90s that there was no crime occurring here. [Alameda] used to average 1200 burglaries a year back during those times. We now average two to three hundred. What’s different? Social media didn’t exist [then]. I have to be responsible, making sure I don’t push out information just for the sake of pushing out information.”
Misinterpreting the data or missing an opportunity to prevent a crime frustrates the Chief. “When a crime happens, you’re like, oh my God, where were we? You can’t predict them all.”
Alameda won’t ever be crime-free. Crimes may occur at a much lower rate than in surrounding communities, but they still happen. “There’s a draw here,” the Chief said. “But the folks that are engaged in criminal activity don’t like coming to this city because of the limited entry and exit points. And the chances of being caught are high. Clearly the fact that we still have crime suggests people are willing to take [the] chance. That’s where you use this information.”