Stepping Into A Data Void: How We Reported ‘Collateral Damage’
The WAMU 88.5 series “Collateral Damage” focused on the impact of the D.C. police department’s aggressive focus on confiscating illegal guns. The investigation explored how tactics used by police to search for guns are also angering and alienating residents, especially in D.C.’s predominantly black neighborhoods where police focus these efforts.
In order to investigate these tactics, we needed more than anecdotes and cell-phone videos. We needed data, documents and records.
All of these were hard to get — and challenging to interpret.
Here, because we think showing our work is the right way to go, we explain what it took.
The existing stop-and-frisk data was either missing or incomplete.
One of our first questions when we launched this reporting project was: How often does a stop-and-frisk led to a gun recovery? In New York City, for example, guns were recovered in less than 1 percent of all stop-and-frisks, according to data obtained by the ACLU.
We were also curious about how often people were patted down, and the reasons police gave for stopping someone.
Unfortunately this information does not exist in D.C.
Despite a 2016 law requiring the police department to keep track of detailed stop-and-frisk data, D.C. police have not collected this information. The police department has released some preliminary stop-and-frisk data but not the type of comprehensive, detailed information we needed for our investigation.
So, we decided to look at records that we knew existed: gun possession charges at the courthouse.
Gun possession cases can provide a window into police tactics
We wanted to look at cases that would give us a window into how police are making stops. When someone is caught with an illegal gun through a stop-and-frisk or a traffic stop, they’re typically charged with gun possession. So we decided to focus on these cases.
But getting access to these cases on file at D.C. Superior Court turned out to be difficult. For starters, there’s no easy way to search court records. You can search by defendant name or case number, but it’s impossible to search by charge. And we wanted to find all of the gun possession cases — not one specific case or defendant.
To solve this problem, we had to improvise. We built an automated web scraper to pull the records from the court’s database. We used this data — charges filed, case number, defendant name, date of filing, and docket information — to create a searchable database hosted on a remote server.
Our goal was to look at gun possession cases to see how police are finding illegal firearms. We selected cases that included a gun possession charge. The most common we found were:
- Carry A Pistol Without A License
- Possession of an Unregistered Firearm
- Unlawful Possession of a Firearm or Destructive Device
We then decided to filter out gun possession cases that included other criminal charges unrelated to gun possession — such as attempted robbery, drug possession, or other violent crimes. We wanted to focus exclusively on simple gun possession cases.
We were left with 1,713 cases over 5 years, from 2010–2015. We assigned each case a number, and then selected a random sample of nearly 500 cases to analyze in depth. (We chose this sample size to give us a 95 percent confidence level that our sample was random, with a margin of error of plus or minus 3.7 percentage points.)
But our challenges weren’t over. You can’t download these cases or access them online.You have to go to the courthouse to look at the documents or print them individually.
Because we wanted as much detail on the type of stop in the case in the case, we focused on the affidavit that was written by the police officer at the time of the arrest. The affidavit is a narrative of the facts of the case: why police officers were on the scene, the interaction between police and the defendant, why officers say they had reasonable suspicion or probable cause to search the person or make an arrest.
These affidavits, also known as Gerstein affidavits, are full of incredible details. We scanned the affidavits and case summaries into machine-readable PDFs.
A sample gun case featuring plainclothes officers making a pedestrian stop-and-frisk.
In this case, officers reported the defendant “looked nervously” before asking him to lift up his shirt and show his waistband.
We analyzed the cases by reading a sample of nearly 500 cases, including the affidavits and case summaries. The findings from this analysis were crucial to helping guide our reporting — but we also needed to be careful about the data’s limitations.
We worked with a team of graduate students at the Investigative Reporting Workshop at American University to carefully read the affidavits and case summaries. We double-checked our work by having multiple people read each affidavit. We entered this information into a database.
For our analysis findings we looked at several data points:
- Basic case info: Where the arrest took place, the date, the address, the race and age of the defendant.
- The outcome of the case: Was the case dismissed in court? Was the defendant found guilty or not guilty?
- The type of stop: “Suspicious activity”, “traffic stop”, ”radio run”, ”tip or confidential informant”, ”search warrant” or other.
We found from our analysis of affidavits that pedestrian stop-and-frisks — the hands-on confrontations between residents and police were the most common type of stop, accounting for more than a third of the cases. This finding allowed us to push back on police testimony that most stops originated out of tips or calls from neighbors.
This was important because these stop-and-frisks seem to spark the most anger in the community, like the controversial pedestrian stop-and-frisk outside Nook’s barbershop. As our story explores, the cellphone video that captured a confrontation between plainclothes police and young men outside a barbershop in Northeast D.C. sparked a series of D.C. Council hearings about the tactics police are using to get guns off the street.
We also found that nearly 4 in 10 of the gun possession cases in our sample were ultimately dismissed in court. (This rate roughly mirrored the overall dismissal rate for the overall sample of 1,713 cases, falling within 3 percentage points).
This was a crucial finding. Experts told us that this high dismissal rate suggests police could be making unjustified stops or searches — which could lead to cases being dismissed. This finding seemed to back up what many residents told us regarding stop-and-frisks and other police tactics.
But we also learned that cases can get dismissed for other reasons too. As prosecutors told us, some cases can get dismissed as part of plea deal in a separate case, or due to court rulings on D.C. handgun laws.
And prosecutors in D.C. are not required to say why they moved to dismiss a case.
From our story:
“If police find a gun on the ground or in a car with multiple people, it can be hard to tie the gun to a single suspect. There can be issues with the credibility of a witness or the officer. Gun cases can also be dismissed because attorneys fear police made an unjustified stop or search. In the worst of cases, police could be violating Constitutional rights, as some alleged at the July D.C. Council hearing.”
So we used our data finding to help guide our reporting — to shape our questions to police and prosecutors — while still carefully highlighting the limitations and caveats in our data.
Using court data to guide our reporting on guns stops
There was no blueprint for this type of investigation. It took months to figure out how to extract and analyze the court records. There were false starts, leads that didn’t pan out — hours spent poring over documents that we ended up not using.
But we knew that the thousands of pages of court documents — the docket updates, the case summaries, the affidavits of gun stops — had value. The fact that so many gun possession cases were ending up dismissed in court was important. Experts told us it suggested there were problems with how police were making stops.
Ultimately, our data-driven approach made the on-the-ground reporting much stronger. The power of radio is its intimacy — hearing the emotion in someone’s voice. We were able strengthen our narrative storytelling with important findings from the court records analysis.
Finally, its critical to show your work. The biggest hurdle we faced was the lack of record keeping and transparency in D.C’s criminal justice system. This investigation hopefully shows the value in trying to open up criminal justice data to ask tough questions about how police and prosecutors tackle gun violence.
Read our story here:
Getting illegal guns off the street is important in reducing crime, but when the trust of the community is lost in the…wamu.org
And definitely check out our video of how this tension over gun stops is playing out in one D.C. neighborhood:
Finally, please feel free to reach out to me or anyone else on our team if you have questions or want to go down a similar path in your community.
Power & Influence Reporter, WAMU 88.5