Skip to main content
ABC News
An Ex-Cop Keeps The Country’s Best Data Set On Police Misconduct

When Talking Points Memo, The Wall Street Journal and The Washington Post needed data on how often police officers are charged with on-duty killings, they all turned to the same guy: Bowling Green State University criminologist Philip M. Stinson.

Phil Stinson.

Phil Stinson.

BGSU Marketing & Communications

Stinson, 50, has become an indispensable source for researchers and reporters looking into alleged crimes and acts of violence by police officers because he has built a database tracking thousands of incidents in which officers were arrested since 2005. His data has shown that even the few police officers who are arrested for drunken driving are rarely convicted and that arrests spike for cops who have been on the force 18 years or longer, contrary to prior research showing it was mostly new officers who were acting out.

The whole data-collecting operation is powered by 48 Google Alerts that Stinson set up in 2005, along with individual Google Alerts for each of nearly 6,000 arrests of officers. He has set up 10 Gmail addresses to collect all the alert emails, which feed articles into a database that also contains court records and videos.

It all adds up to a data set of alleged police misconduct unmatched by anything created inside or outside of government, which itself often uses Google Alerts to catch these cases.1 Yet Stinson’s database inevitably has holes because it relies on the media to cover every officer arrest, and because it takes immense effort to code each entry. The data set keeps falling behind.

Stinson’s path to police-misconduct expertise was a winding one. A high-school dropout, he worked for police departments in Arlington, Virginia, and Dover, New Hampshire, before getting his law degree, then turning in his license after mishandling client funds. He switched to criminology, getting his Ph.D. in 2009. He chose his research subject in part because of his experiences as a cop, seeing police officers get away with crimes others wouldn’t, and in part because he wanted to win a bet he’d made in his Master’s program. Now he’s the recipient of a National Institute of Justice grant to study police misconduct, host of an occasional podcast on the topic, and a go-to source for the media.

Stinson says he has no bias against his former profession. “This isn’t anti-police,” he said. “I hope to write papers and get things published that are helpful to law enforcement, not that bash law enforcement.”

The following is an edited transcript of my telephone interview with Stinson, with some followup questions by email.

Carl Bialik: You’ve worked with the media on stories about arrests for killings by police, and recently you’ve also been a source on arrests of police for DUIs. You’ve also studied sex crimes and corruption. How did you become the expert on police arrests?

Philip Stinson: The idea came up in the fall of 2004, when I was finishing my master’s at West Chester University of Pennsylvania. I was taking an ethics class. Somebody in the class — it was a bunch of cops in class, mid-career — somebody made a comment that cops don’t get in trouble much. I said, “That’s just absurd.” I started looking into it and realized there are no government statistics, and no government agency tracking it well.

I set up 48 Google Alerts, let it rip, and started printing pages out. Originally it was to win a bet in the master’s class.

CB: What did you bet?

PS: I don’t know. It was a beer or something. I wasn’t a betting man. It wasn’t anything big.

I had two and a half years, three years of data in my dissertation, covering 2005 to 2007, with 109 quantitative variables.

And then over time at Bowling Green, we now track 270 or so quantitative variables. Everything is automated now. Because data collection is real-time — you can’t use Lexis Nexis, NewsBank, all these other archival news databases, because lots of stuff has disappeared from the Internet — so because of that it’s very slow and time-consuming. It takes forever to do.

Some reporters wanted everything I had — everything. I was like, go fuck yourself. You get everything. I just spent 10 years on this.

Now we’re up to almost 11,000 cases involving almost 9,000 officers. We log these cases, then make Google Alerts on individual names so we can track cases through courts and the media.2

CB: How sure are you that The Washington Post’s count of 54 police officers charged with fatal shootings while on duty is a complete count?

PS: I want to make it very clear I’ve never claimed I have every case. It’s possible that’s not an exhaustive list. If anything, I think we’re missing just a handful of cases of killings, because those kinds of cases get news coverage. I’d have to hazard a guess that we do a better job of collecting data in smaller metropolitan areas and rural areas, because arrests there are so newsworthy.

CB: How much do you think you miss by relying on media reports?

PS: If I had 1,100 arrest cases, I can’t believe there are another 1,100 out there. I doubt we’re missing half, but there aren’t none.

CB: You were a lawyer before going to graduate school for criminology. What happened?

PS: I fucked up in my law practice [Stinson Law Associates, PC, in Bryn Mawr, Pennsylvania]. I was charged with several crimes and the commingling of funds in 2002. It was the kiss of death. I paid a heavy price and took a heavy hit.

CB: Did you do what you were charged with doing?

PS: I did it. [“Chronic sleep deprivation exacerbated mental health issues that I was not aware that I had and led to impaired judgment in many instances, and for that I’m very remorseful and in treatment,” Stinson told the Philadelphia Inquirer in 2002.] Everybody was made whole eventually.

I don’t like to leave that out [that I did this]. It’s not something I’m proud of, and I don’t like seeing it in print. It really has nothing to do with why I am doing this research.

I certainly think I’ve done everything I can to rehabilitate myself.

If I were teaching theology, it would never come up, but in criminology, it does come up sometimes with students, so I usually walk into the classroom and lay it out there.

A photo of Stinson from summer 1986 when he was a police officer in Dover, New Hampshire. Taken in Dover, New Hampshire, outside his cruiser.

A photo of Stinson from summer 1986 when he was a police officer in Dover, New Hampshire.

Courtesy of Philip M. Stinson

CB: What was your experience as a cop and did it influence your work?

PS: When I went to New Hampshire, I saw some crazy shit. It really changed my outlook on things. When people were arrested, they would take them into the booking room, and sometimes the sergeant would come in and just beat the shit out of the guy while he was handcuffed — shit like that.

I was floored with my experience up there. In the two years I was up there, I saw all kinds of shit I did not know happened.

They faked reports, and there was creative report writing, to fit the arrest they wanted to have. There was evidence that was tampered with and overly suggestive court identifications to nail people with shit. It was quite an eye-opener. [When asked for comment, Anthony Colarusso, chief of Dover’s police department, said, “I feel strongly that the Dover Police Department has a high level of integrity and has had that since I joined the police department in May of 1985. In fact, our longstanding policy is that officers be terminated for any level of untruthfulness.” He added, “No police department is perfect, but we are very aggressive in holding ourselves accountable for our actions.”]

So I sort of had that kind of stuff in the back of my head for many years.

CB: Many who quote your stats use them to make larger points: Too many police officers break the law, or too few are punished for it, or police are too violent with citizens, or are biased against black people. Do you have a larger view based on the data you’ve collected?

PS: What we’ve seen is that courts are just very reluctant to convict cops who are charged with crimes. We see some interesting patterns with that. So juries, and even judges in bench trials, are not comfortable for prosecutors to prosecute cases.

We do look a lot at what gets them arrested, what gets them convicted or not convicted, and what gets them to lose their job. Two things that have come out that have just sort of blown me away:

1) Many cops are arrested for crimes that are serious and don’t lose their jobs. We have cops who were arrested in 2005 or 2006, and we’re startled to see they are still a cop — maybe still with the same agency, maybe not. That just blows me away. John Lewis in Schenectady, [New York,] is one example.

2) The other is, from prior research, it looked like if cops get in trouble, it will be early in their careers. What we found is that almost 20 percent of these cases involve officers with at least 18 years of experience.3

CB: The Washington Post investigation you contributed to found that in the last decade, just 54 officers were charged with fatally shooting someone while on duty. One open question is, what’s the denominator? How many people do you think police have killed overall in that time?

PS: I think it’s such a huge number. I don’t know how you could manage it. How do you verify it? We can at least go back and get the court records [for officer arrests]. It’s an impossible thing to do all of that [for the cases in which no one was charged].4

But we do have this file, that’s getting bigger, where we put articles that we come across where we’re just convinced they’re going to arrest the officer but they don’t. And the file just keeps growing. And we kind of scratch our heads. How does that happen?

CB: Seems like that’s more evidence that it’s much easier for police to avoid conviction.

PS: Absolutely. You’ve got to really fuck up to get convicted as a cop, of anything at all. We’ve seen some really weird things with charging incidents, like a cop charged with domestic violence incidents, hauled out of the house, and then charged with DUI. Because if he were charged with a simple misdemeanor of domestic violence, he can’t carry a gun, can’t own a gun, can’t carry ammo, can’t own ammo. We have dozens of police officers with qualifying crimes who’ve still got their job, carrying a gun every day.

CB: Did you ever help a cop avoid a domestic violence charge?

PS: No. No. I never would have thought of doing something like that.

CB: Do you think it’s the job of policing that makes some people violent or break the law, or are some of the people who are drawn to the job prone to violence or lawbreaking?

PS: It’s a chicken-and-egg situation. I know this from my own experiences, even from my own experience as a police officer, that the gun and badge become part of you. When you lose that, something is taken away from you. There is such a power element to it.

Brian Gilmore, whom I went to law school with, is now at Michigan State. Every once in a while we get together in Ann Arbor, halfway between us. I told him about my research. He said one time he worked at a law firm in D.C. and represented union members who were police and firefighters when they got in trouble. He represented a ton of cops charged with domestic-violence crimes, but not one firefighter. That’s quite telling. [Gilmore, asked to confirm the story, said, “My experience is anecdotal but true.”]

CB: Seems like now there are so many ways that officers can be filmed — by dash cams, body cameras, surveillance cameras and citizens’ cameras. Do you have a sense of whether that’s a growing factor in these arrests?

PS: We haven’t coded for that, and we’re three years behind. We haven’t gone back to code them. We need more grant money. I just don’t have the money to hire the staff.

Anecdotally, absolutely it’s made a huge difference. That starts with Rodney King forwards. But in the last few years, everybody is a videographer. You whip it out, very easily and with no cost, and take videos. The problem is, they don’t start filming until something has caught their attention. The combination of dash-cam video, citizen video and body-worn camera video gives you different pieces you can put together.


The problem with body cameras is that they can turn it off. We’ve seen that in a few instances, the cop will turn it off, beat the shit out of somebody, then turn it back on.

CB: What do your former police colleagues think about your work?

PS: Somebody who was a supervisor I worked under in the Arlington, [Virginia,] police department, she posted something on Facebook in response to something I posted. She posted this thing asking if I’d been on any ride-alongs recently. I responded, I don’t think there are too many police officers who are interested in having me ride along with them these days.

CB: You’ve mentioned “we” a few times. Who’s “we”? Who are your collaborators? Grad students?

PS: I’ve collaborated with John Liederbach at Bowling Green. Steven L. Brewer at Penn State has done a lot of work with machine-learning type stuff and decision-tree analysis.

Now that I don’t have grant money, and we’re in a weird time in higher education where there’s a push for more grad students paying fees, it’s harder to get help without grants.

In the last one and a half to two years, I’ve started to see what I can get away with from undergraduates. I have two grad students, and then 10 undergraduates who work six to 10 hours a week. They don’t get paid, and they do a great job. Everyone has a GPA over 3.5.

A friend of mine, who found Vince Foster’s body, pointed out that I would never be qualified to work for me, based on the standards for the students.

CB: What was your GPA?

PS: It was really low until I got medicated for ADD. I saw a doctor and realized I was out of control with impulse control. It made a world of difference. I ended up with a 3.94 GPA in my Ph.D. program. It was not until my late 30s that I was medicated. Both of my ex-wives would tell you I’m far calmer than before.

CB: Are you married now?

PS: No, I’m between wives and dogs.

CB: How do you code, and how long does it take? Do you have to code all 270 variables for every case?

PS: For my grant, we had 12 months, and I had eight graduate assistants working for me at the time. Four of them were just working on coding cases. I think it was 20 to 30 a week they would get done. They all worked for me for 20 hours a week. It takes even longer now. There are so many documents in the cases.

I have been running the same 48 search terms5 in Google Alerts since the beginning of 2005. What we’re doing just now is getting the database as tight as we can so when we have more grant money, we can go back in and code all the cases. Two years ago, I started collecting videos. Now I have 3,000 videos. I don’t even know what we’re going to use a lot of them for.

CB: Why track only arrests as opposed to so many other things police do that don’t result in arrest?

PS: I wanted to have something where somebody, some magistrate had signed off on an arrest warrant. So you gotta be charged — however you come in, whether arrested, arrested then indicted, indicted then arrested — somehow formally brought in and booked. That’s just to give it some rules.

CB: You mentioned before that you don’t like sharing the whole database.

PS: It’s my intellectual property.

CB: But do you think the government itself should be compiling a similar database, doing the work itself?

PS: Oh, absolutely. They haven’t figured out how to do it on their own. Survey research is not going to get the right answers. It’s seemingly hard to track for some reason.

CB: Are there any geographical trends you see? Are police officers especially likely to be arrested in certain states or cities, or in rural or urban areas?

PS: It’s everywhere. We looked at it at the county level. One variable is county FIPS [Federal Information Processing Standard] numbers. We mapped these cases and graphed them.

officer_arrest_rate_per_100000_pop.v4 (1)

We see it in rural areas. We see it everywhere. It’s amazing to me. Everywhere across the country we see this kind of stuff, all 50 states and the District of Columbia.

CB: Cops are under lots of scrutiny and criticism. They’re increasingly filmed on the job. Has it become a tougher job since you had it?

PS: Over the last 10 or 15 years or so, a lot of police departments have had trouble hiring qualified people, especially during the height of war. Some police departments lowered educational standards. I don’t know if it’s a harder job. It feels like a more violent job.


  1. The Bureau of Justice Statistics has used what it calls open-source methods, including Google Alerts, since 2010 to monitor deaths in police custody. “BJS, like all Federal statistical agencies, is interested in and will use various types of ‘big data’ (e.g., unstructured and unverified data from web searches) but will use them in a manner that is consistent with the nature of the data,” Michael Planty, chief of victimization statistics for the BJS, said in an email. “For example, open source can be a good source for ‘nominating’ cases that then require follow-up with the affected entity for verification.”

  2. Stinson and collaborators started setting up Google Alerts for individual officer names in 2011, and have set up them up for roughly 6,000 of the cases in their database.

  3. That study covers the years 2005 to 2007. At my request, Stinson updated this analysis using all the cases he and his colleagues had coded from 2005 to 2011. He was able to reply in minutes by querying his database. The findings were consistent with his earlier study: Nearly 20 percent of the 4,806 cases of police-officer arrests for which he had information on years of service came at least 18 years into the arrested officer’s career, with another spike at 25 years of service.

  4. Some groups are trying. One group whose data we used in a recent article, Mapping Police Violence, estimates that in 2014 and March 2015, about 25 percent of people killed by police were unarmed. The Washington Post’s analysis of cases against police charged with fatal shootings while on duty between 2005 and 2015 found that 51 of 54 — 94 percent — officers were charged in cases where the victim was unarmed. That suggests that police officers who kill an unarmed person have about a 50 times greater probability of facing criminal charges than do police officers who kill someone who may have been armed.

  5. Those terms: “agent was arrested,” “agent was charged,” 
”agent was convicted,” “agent was indicted,” “deputy was arrested,” “deputy was charged,” “deputy was convicted,” “detective was arrested,” “detective was charged,” “detective was convicted,” “detective was indicted,” “detectives were arrested,” “detectives were charged,” “detectives were convicted,” “detectives were indicted,” “officer charged,” “officer was charged,” “police chief was arrested,” “police chief was charged,” “police chief was convicted,” “police officer was arrested,” “police officer was charged,” “police officer was convicted,” “police officers were indicted,” “police officers were charged,” “police officers were convicted,” “sheriff was arrested,” “sheriff was charged,” 
”sheriff was convicted,” 
”trooper was convicted,” 
”trooper was arrested,” “trooper was charged,” 
”trooper was convicted,” 
police “captain was arrested,” police “captain was charged,” police “captain was convicted,” police “captain was indicted,” police “lieutenant was arrested,” police “lieutenant was charged,” police “lieutenant was convicted,” police “lieutenant was indicted,” police “officer is charged,” police “sergeant was arrested,” police “sergeant was charged,” police “sergeant was convicted,” police “sergeant was indicted.”

Carl Bialik was FiveThirtyEight’s lead writer for news.