How we did it

Most data on police shootings is scattered, focused on fatal shootings when it’s available at all and even then rarely contains the level of detail we were looking for: How often do police shoot people who are unarmed? How often do shootings occur after pursuits? Why did officers feel they had to use deadly force?

The only agency privy to the details of every shooting in the county is the Los Angeles County District Attorney’s Office. When an officer shoots someone, prosecutors review evidence to determine whether it was legally justified.

If prosecutors decide not to file charges, the district attorney sends a letter to the head of the law enforcement agency summarizing the facts of the case and spelling out the reasons a criminal case won't be pursued.

Because the D.A. hasn’t filed charges against an officer for an on-duty shooting in 15 years, the letters comprise a rare collection of every shooting in the county. They are the basis for KPCC's analysis.

KPCC requested letters for all shootings that occurred on or after Jan. 1, 2010. All but two shootings in 2015 were still pending at the time of publication, so we ended our period of analysis at Dec. 31, 2014. The district attorney released 359 letters that KPCC determined were on-duty officer-involved shootings. These documents show 375 people were hurt or killed in those cases. We did not review shootings that involved off-duty officers or D.A. reviews of suspects who died in custody from causes other than shooting. At the time of publication, the D.A. had not rendered a decision on at least 29 on-duty shootings that occurred during the five-year period we reviewed.

KPCC staff — with help from interns at NPR in Los Angeles — read these letters to build a database that included more than three-dozen factors for each shooting. We had to make a number of choices when cataloging their details. Among them:

• We found that we could reasonably place only about 68 percent of the 359 incidents on a map because the locations of the rest of the shootings were too vague in the letters.

• If an individual possessed a firearm, airsoft pistol, BB gun or other weapon, we deemed them to be armed. If an individual did not possess a weapon at the time of the shooting we deemed them to be unarmed, even if an officer feared the person was trying to use a vehicle to injure or kill the officer. However, in 11 cases where the individual was carrying a non functional, toy or replica firearm, we categorized them as neither armed nor unarmed. Additionally, one bystander who was shot was categorized neither armed nor unarmed.

• Our findings regarding signs of mental illness and/or impairment from drugs or alcohol are based on details from the D.A.’s summary of the shooting. Those range from knowledge that a person had been involuntarily committed for treatment to a family member's anecdote. In fatal incidents, toxicology tests at times provided insight.

• When we refer to a pursuit or fleeing suspect, that includes any shooting where an individual ran from police — whether on foot or fled in a vehicle — at some point during the incident.

• We determined an individual "reached for their waistband," "ignored officer commands" or an officer "couldn't see their hands" when those phrases or variants appeared in the D.A. letters.

KPCC used multiple sources of information to supplement the details, including interviews, court records and a number of other data sources, including:

• The Los Angeles County Medical Examiner released data for shootings during the five-year period of review in which there were “law enforcement-related circumstances.” We used this to determine gender, age and race or ethnicity of people who were shot and killed by police.

• PDFs from the Los Angeles County Sheriff's Department provided some details on deputy-involved shootings between Jan. 1, 2010 and July 20, 2015.

• California Department of Justice data detailed the number of officers assaulted or killed in California. Separately, we consulted that agency’s data on in-custody deaths between 2010 and 2014, some of which are shootings.

• Police officer and sheriff's deputy employment data came from the Commission on Peace Officer Standards and Training, the Los Angeles County Sheriff's Department, the Long Beach Police Department and the Los Angeles Police Department.

• Arrest data came from the California Department of Justice's Monthly Arrest and Citation Register from 2010 to 2014.

Frank Stoltze, Annie Gilbertson and Rina Palta reported the stories. Chris Keller and Aaron Mendelson provided research and contributed to the fact-checking and reporting.

Sean Dillingham and Katie Briggs designed the series' visual aesthetic and crafted the user experience. Dillingham also served as the principal developer. Vijay Singh managed the project's development and design roadmap.

Maya Sugarman and contributor Susanica Tam provided photography.

Kristen Lepore and Ashley Alvarado led engagement and promotional efforts.

Will Craft, Aaron Bloom, Taylor Haney, Christian Brown and Quinn Owen entered data from district attorney records.

Evelyn Larrubia edited the project.