Screenshot from the Washington Post’s interactive police shootings database. (TWP)

Inside the Washington Post’s police shootings database: An oral history

Reporters, editors and designers talk about what it’s like to gather data that has never been available before, identify untold stories and share them with readers.

Note: This post was originally published on Dec. 16, 2015 and reflects accurate data up to that point. At the end of the year, our database showed a total of 986 people fatally shot by police — more than double the average number reported annually by the FBI over the past decade. For the most up-to-date information, visit our database here.

After Michael Brown was shot and killed in Ferguson, Mo., by police officer Darren Wilson, pointed questions arose about police shootings: How often do they happen? Who are the victims? What happens to the officers?

Those questions couldn’t be answered because government data was incomplete. So, a team of reporters, researchers and editors at The Washington Post spent the past year building a comprehensive database of the country’s fatal shootings by police officers in the line of duty.

“We needed to do this because it wasn’t being done well,” said Post managing editor Cameron Barr. “People in a democratic society have a right to know the results of the state’s use of force in the enforcement of law.”

As of December 16, the database had more than 900 entries with more than a dozen details about each killing — including the race of the person who was killed, the circumstances of the shooting, and whether the person was armed.

From this large-scale undertaking came stories that drove an informed conversation about police shootings and led to critical policy reforms:

These significant findings caught the attention of law enforcement leaders across the country. Recently, in part as a result of our work, the FBI has committed to change the way it tracks and compiles data surrounding these incidents.

How did this start?

Wesley Lowery, national reporter
After the shooting of Michael Brown in Ferguson, as a reporter on the ground very early on, I had a lot of conversations with editors where they were asking: What kind of national context can we put this in? What kind of information is available about police shootings, and how many times this has happened?

We had activists telling us that this is happens all the time, that black men are being executed in the street, especially unarmed black men. Meanwhile, we had the police union in St. Louis saying that this never happens.

The natural journalistic desire is to answer that question: What’s right? Who’s right? … We realized very quickly that we couldn’t answer those questions; those statistics didn’t exist.

David Fallis, deputy investigative editor
I already knew about the lack of information nationally, because I dealt with that back in the early 2000s when we were diving into the police brutality out in Prince George’s County, Md. I knew that this would be a huge lift, and I knew that it would take a lot of work from a lot of different people.

Honestly, I was a little skeptical about the amount of time involved… I just knew how difficult it is to get information from police departments. When you really start peeling back the layers on an individual shooting, it can look like a lot of different things — there’s a narrative within a narrative within a narrative.

Kimberly Kindy, national investigative reporter
We tried to determine how often police officers, while on duty, fatally shoot and kill people — and we couldn’t find out. … We knew then that it was woefully undercounted, we just didn’t know by how much. I’ll be honest, I was one of the people who was very doubtful it could be done. It seemed like an overwhelming, very difficult thing to do.

How did you determine what the bounds of the database were?

Julie Tate and Jennifer Jenkins, news researchers
We arrived at the universe of all people being shot by police because it was a clean universe; when someone is shot and killed by a police officer there’s no confusion about what the cause of death was.

We first looked at two different places: Fatal Encounters and Killed By Police, which track all the people killed by police officers, meaning: tasers, hit by a vehicle, shot, died in custody. We looked at their sites and saw they were only collecting name, age, race, how the person was killed and one clip on the shooting. So we began by using that architecture to build our database.

We added originating incident, which looked at why the police were involved in the first place. Was it domestic violence? Was it a suicidal person? Was it a traffic stop? We also tracked whether or not there was a suspicion of mental illness. We tracked whether or not there was alcohol involved. Whether it happened in the day or the night. Obviously whether someone was armed and unarmed. Every time we marked somebody as armed we also marked down the kind of weapon that they possessed at the time. And then we have geocoding so we can look at census tracts to see patterns of where people are being killed. Of course we also tracked race, age, etc.

Linda Epstein, photo researcher
When I started this project, they presented me, at the time, with maybe 150 names on the list, and I had to go find photos of them. We wanted to bring faces to these data points. I got to be in touch with everybody that I knew in the photo world because I was contacting people all over the country.

Screenshot showing database entries for those shot by police, as recorded in the Washington Post’s police shootings database. (TWP)

How did you build the database?

John Muyskens, graphics editor
I started writing code right after a meeting in March. I used Django for the project, which is a web development framework. It was created by newsroom developers for newsroom developers, specifically to build database-centered web applications for telling news stories. It makes it really easy to set up a data entry tool that is directly tied to the fields in your database.

There was also a simple interface for any reporter in the newsroom to look at all of these records and to sort and filter them. It wasn’t pretty, but it allowed you to filter by any property in the database. This ended up being pretty powerful and helped us to identify some of the interesting stories without having to do some Excel magic or count by hand the number of cases in each category.

Ted Mellnik, database editor
One interesting thing about the way that data works, it’s like reporting. When you start off, you have some ideas about what you want, but you don’t know what’s possible until you start asking questions and compiling the data.

You have to adjust, which is a really interesting challenge when you are working with data, because by definition you have structure. I have a great deal of respect and appreciation for the flexibility that people were willing to implement in an environment that you don’t really think of as being flexible.

Kennedy Elliott, graphics editor
Given that it was a database, we relied so much on the reader to explore and find trends. … Our goal with this project, was to be really transparent and to show all the data we had collected.

But until the year is over we don’t have a complete data set. We are making findings throughout the year. We are constantly adding more into the database. … We were trying to publish what we knew and trying to always be processing and reflecting on it along the way.

What did you see when you started looking at the data?

Lori Montgomery, America editor
When the graphics team actually started to create the infrastructure so that we could see and sort the information, it was really cool. I remember the meeting with some editors, when we realized, wow, now we really have something. Something that is deep and rich and new. We had this body of information that no one has ever had before, you can ask it any question and come up with an answer.

Keith L. Alexander, local reporter
Looking at the data and trying to find a story in it was like drinking from a fire hydrant that has been opened in the middle of August. You think you find a small story and … you realize this small story is a bigger story, it’s a bigger story, it’s an even bigger story. That was the most challenging part — you think you’ve found a small kernel of something and you keep digging into the database and you see a pattern, you see repetition.

Kimberly Kindy, national investigative reporter
About four months in was when one of the first big gets became clear, one of the first surprises. Half of the fatal shootings were in some sort of domestic disturbance. Sometimes it was domestic abuse, but sometimes it was people in a family in the middle of a mental health crisis. That half of the database included some of the most tragic cases that we saw. Many times people reaching out to the police, trying to get help for someone who was not a criminal and they end up dead. It was heartbreaking.

I remember this one mother who called the police because her schizophrenic son was out in the cold at 2 or 3 in the morning wearing nothing but underwear and she was trying to get him to come in. She called the police — she had before — and asked them to please come and help her get her son inside. He was out swinging a broomstick around trying to get away from the officers, and he ended up dead. When the investigators came back later that evening to talk to her about it, they brought her a six-pack of Coke. Some of the cases are so tragic and so sad, you can’t believe it when you see how routine it can sometimes become.

Screenshot from Lavall Hall’s records in the Washington Post’s police shooting database. (TWP)

The police officers, we learned, are not always getting the best training. It’s very difficult to deal with people who are in a mental health crisis. The traditional training where an officer goes in, they rush in fast, they yell at somebody, they tackle them, they take control, is the very opposite of the way to try to negotiate with someone in a mental health crisis.

That was a very important get because we were able to identify not just deaths and how often they happen, but a subsection of deaths that are preventable. It started a conversation nationally about the need to change training in this area and more and more departments are acknowledging that they need to improve and are doing it.

What are your biggest takeaways from this undertaking?

Cameron Barr, managing editor
It made sense to us to embark on a project to count every single one of those instances over the course of a year and see what we could learn about the way that force is applied. And the results have been remarkable.

We’ve seen in the course of this year, the government point to our work and the work of the Guardian, and then launch a new government effort to count these deaths more rigorously going forward. Thats gives us a great sense of satisfaction that we have contributed to the public debate to public understanding of how police officers do their jobs and who the casualties are along the way.

Ted Mellnik, database editor
A lot of times journalism is kind of a solitary exercise — a reporter goes out, digs up stuff, writes a story, or sometimes people bring in stuff for one or two people. … Just the scale that people worked together across departments and built this one thing, I was really proud. There was a lot of heroic editing and reporting in this whole project, but the data infrastructure was the foundation of it all.

David Fallis, deputy investigative editor
For me, I’ve been at The Post since 1999, and this was a first — it’s a completely different way of reporting, gathering and presenting information. By using the news itself to gather these incidents, and pull them up and repackage them and put them in a vehicle to do analysis, we were able to look at patterns. The power of that is that you have a real-time understanding of the universe.

Kimberly Kindy, national investigative reporter
It would have been really terrific if we could have tracked down the name of every single police officer who shot and killed somebody. … It’s just shameful that you immediately get the name of the person who is dead, sometimes within minutes of their death, often accompanied by their complete criminal history — even if it’s just a bunch of petty thefts. What needs to reformed and changed is that the names of police officers who shoot and kill people should be revealed more readily.

Lori Montgomery, America editor
The other thing that’s a little frustrating is that you can’t forget we are building this database based on police reports. For example, when assessing the level of threat — the cops always say that there was a threat, that’s why they shot. That’s one of the most important but one of the murkiest areas; right now, for a lot of these cases, all we have is the police department’s word for it.

So, now that we have all of this information, it would be great to go back and take a deeper look at some of these cases and figure out what really happened. It would be nice to go back and put some pressure on those things. There are still a lot of questions.

Wesley Lowery, national reporter
I think now we can bring facts, we can bring real information and real analysis to a conversation that otherwise had been about emotion and anecdote.

Police are some of the most powerful individuals in the U.S. — they have the right to kill you, and they have the right to do it in the name of the taxpayer. It’s important for us to be asking hard questions and to be holding police departments to account. That’s not to say that many of these shootings are not justified, but it is to say that we need to ask that question because if we don’t ask that question, no one else is asking that question.

Explore the database

Read the stories

This effort was led by editors David Fallis, Lori Montgomery, Ted Mellnik and Maria Glod. The data was collected by Julie Tate and Jennifer Jenkins. Photo research was done by Linda Epstein, Robert Miller and Wendy Galietta. John Muyskens, alongside Ted, built the internal database. Kennedy Elliott built the user-facing version. Several reporters found and shared important stories that surfaced from the data: Kimberly Kindy, Wesley Lowery, Keith Alexander, Amy Brittain, Sandhya Somashekar, Steven Rich, Kimbriell Kelly, Marc Fisher, Scott Higham, Cheryl W. Thompson and Mark Berman. Reporters from our video team included: Whitney Leaming, Zoeann Murphy, Alice Li and Jorge Ribas. The project was copy edited by Beth Hughes, Brian Cleveland and J.J. Evans. Matt Callahan contributed design, and Alex Laughlin worked on audience engagement.

This oral history was compiled by Allison Michaels.