How We Reported on Homicide Investigations in St. Louis

In February 2021, St. Louis Public Radio and APM Reports began an investigation — later joined by The Marshall Project — into the struggles of the St. Louis Metropolitan Police Department to solve homicides.

While St. Louis’ record-high homicide rate was well-known, the police department’s ability to solve those homicides had received less scrutiny. APM Reports and STLPR wanted to know what percentage of homicides police were solving and whether this figure differed by race. The quest for this data would end up lasting years.

The records request was modeled after a similar inquiry made in 2017 by The Washington Post. We asked for basic details about homicides in St. Louis, including the name, age and race of the victim, where they were killed, and the case’s clearance status, which, generally speaking, refers to whether the department considered the case solved.

Though police departments across the country routinely share these records, and the St. Louis police had shared this information with The Washington Post in 2017, the department refused to provide clearance information to STLPR and APM Reports. Instead, the department argued the clearance status for active cases was an “investigative record” and therefore the entire spreadsheet outlining clearance information was confidential under Missouri’s Sunshine Law. The city spent nearly two and a half years fighting to keep this information secret.

As a result, Minnesota Public Radio, the parent company of APM Reports, filed a lawsuit in November 2021 with the help of Lisa Hoppenjans, Tobin Raju and others at the First Amendment Clinic at Washington University in St. Louis’ law school.

After more than 18 months in litigation, the city and Minnesota Public Radio reached a settlement in June 2023. Police provided spreadsheets documenting clearance information for homicides from Jan. 1, 2018, to June 30, 2023. The department also provided the same information it had given The Washington Post: a spreadsheet outlining clearance information reported from 2001 to May 31, 2017. Since the department no longer updated clearances between 2001 and 2018 in this database, the city agreed to provide incident summaries for unsolved homicide cases listed in The Washington Post spreadsheet, along with data for every homicide that occurred between Jan. 1, 2017, and Jan 1, 2019.

In this series, APM Reports, STLPR and The Marshall Project relied on the police department’s designation of cases as homicides, which typically result in charges of murder or willful manslaughter. We similarly deferred to the department’s designation of homicides as “justifiable,” a term used when police officers or private citizens kill someone who was allegedly committing a serious crime. We excluded justifiable homicides from our analysis, as well as homicides labeled as “officer-involved shootings,” because there wasn’t enough information to determine how those cases should be classified.

Our analysis relied on the police department’s determinations of victims’ races. “Unknown,” non-white Hispanic or Latino, Asian and other people of color killed in St. Louis during the past 20 years made up less than 1% of all homicide victims.

We also relied on the department’s clearance designations. The FBI’s crime-reporting program, which most police departments report to, defines homicides as “cleared” when police have arrested a suspect, charged them and turned the case over for prosecution; or when police have fully identified a suspect but cannot take them into custody for any reason, including when a suspect has died or cannot be extradited.

In our reporting, we use the words “cleared,” “closed” and “solved” to refer to cleared cases.

We calculated clearance rates, which represent the ratio of cleared homicides to total homicides, differently than the St. Louis police department. The department uses a method set by the FBI and cited commonly by police departments and in news reports.

The FBI method divides the number of homicides police cleared within a time period, regardless of when that homicide occurred, by the number of homicides that occurred within that time period. For instance, if a 2019 homicide was cleared in 2023, the department would count that case toward its 2023 clearance rate. This makes it possible to report clearance rates that exceed 100%.

The FBI’s calculation can lead to obfuscation, said Jeff Asher, a data analyst who has worked for the CIA, Department of Defense and the New Orleans Police Department. “It doesn’t give us a real appreciation for the share of the year’s murders that you’re clearing,” he said.

Instead, we calculated clearance rates as the percentage of homicides occurring in a given year that were solved before Jan. 1, 2024. Unlike the FBI’s measurement, this figure reflects the share of unsolved homicides in a given year, a metric critical to our reporting.

Our analysis of homicide clearance data spans from 2004 through 2023 and is current as of Jan. 3, 2024.

We created the database by joining the spreadsheets provided by the department and removing duplicates, using case numbers and victim’s names as unique identifiers.

APM Reports then extracted text from incident summary reports so that the clearance status of homicides between 2001 and 2017 that had not been cleared as of 2018 could be updated as needed.

When incident reports could not be machine-read, we entered clearance data by hand. A separate group of journalists on our team reviewed the incident report classifications.

We then mapped the homicide locations. Sometimes, the department listed a cross street or highway as the incident location. When exact addresses were not provided, we geocoded the homicide to the nearest location based on information from news reports.

Our teams sought to account for homicides and clearances in St. Louis as accurately and comprehensively as we could. The data is not perfect, however. For example, the department initially stated 262 homicides took place in 2020 in its online reports, which it has since updated to 263 — the same number STLPR and APM Reports found in their analysis. But the department’s annual report, released in 2021, reported 264 homicides.

And on at least a few occasions, we found the police entered an incorrect case number or misspelled victims’ names across at least two sources of data provided to us.

Given these uncertainties, we have avoided providing exact counts of homicides and clearances when we couldn’t validate them.

The police department did not respond to requests to discuss our findings.

The data for this analysis can be found on GitHub.