Recent social movements for police reform have highlighted the lack of data collected or publicly available on a host of important policing practices—most notably, the dearth of information on officer use of force incidents resulting from a lack of transparency and consistency on disciplinary processes and protections put into place by police unions. However, though far from perfect, there are data currently available. The following post offers a summary of some of the more commonly cited sources used to better understand the trends of police use of force over time.

Past and ongoing efforts to collect reliable and valid use of force data from the roughly 18,500 federal, state, local, and tribal law enforcement agencies have shown how incredibly complex and difficult the task is. The longest-running data collection effort is the FBI's National Incident-Based Reporting System (NIBRS), which began in 1988 and is used to report crime and law enforcement agency (LEA) data, including officer use of force statistics. Throughout the years, efforts have been made to increase the number of agencies reporting data to NIBRS, and in January 2021, the system became the mandatory method for LEAs to share criminal incident-level data with the federal government. However, many local agencies across the country have struggled to transfer to the NIBRS system from the antiquated uniform crime reports, leading to gaps in data being reported and made available for public consumption and analysis.

The federal government has routinely estimated police contact and use of force levels through its nationally representative sample of respondents in the National Crime Victimization Survey. These data estimate that in 2018, roughly 61.5 million people 16 or older had at least one contact with the police in the prior 12 months, with approximately 1.3 million (2.0 percent) reporting that they experienced threats or force by an officer. Although this information provides reliable data on uses of force, it is neither local nor specific to agencies, and the information from official bulletins is typically released years apart. In response to these issues, the FBI established a new national dataset of police use of force in 2019; however, participation, which is voluntary, is low. In 2022, the data represented 53 percent of sworn officers employed by federal, state, local, and tribal agencies across the nation, from roughly 6,750 of the 18,500 agencies in the country, or only 37 percent.

In comparison to officer use of force, the collection and reporting of police use of deadly force is somewhat more reliable but has similar challenges. In recent years, a wide range of entities, including federal, media, and private organizations, have made notable efforts to document police use of deadly force data by, for example, creating crowd-sourced or web-scraped databases. The figure below details the current efforts to document police use of deadly force by source and across the years.

Figure. Yearly count of police use of deadly force, by data source.
Note: As of 06/13/2022
Yearly count of police use of deadly force by data source

Federal data collection efforts of deadly force are housed under the FBI's Summary Reporting System Supplemental Homicide reports and its NIBRS program, and the Center for Disease Control and Prevention's (CDC) Underlying Cause of Death database and its National Violent Death Reporting System (NVDRS). The Biden administration also recently signed an Executive Order that, in part, requires all federal LEAs to collect and submit data on a monthly basis on incidents involving the use of deadly force.

Another grouping of sources comes from major efforts by notable media organizations. Specifically, the Washington Post and the Guardian have publicly available data, while Reuters has a public-facing online feature detailing police uses of deadly force but has not released the data associated with it. The final source category includes crowd-sourced or web-scraped data. The two main crowd-sourced resources are Fatal Encounters and Mapping Police Violence.

The figure shows the significant variation in the number of incidents of police deadly force across these databases, mainly as a result of how a "death" is defined. The federal data sources are relatively conservative in the definition. The FBI supplementary homicide reports and NIBRS account for all justifiable homicides by a law enforcement officer with a weapon-justifiable homicide being defined as an instance where an officer kills a community member in self-defense or in defense of another person. The CDC's mortality statistics include all Y35 ICD-10 Codes (medical codes for diagnoses, symptoms, and procedures) for "legal intervention," excluding those for legal executions (i.e., a death sentence). The NVDRS defines the use of deadly force as a community member's death by a law enforcement officer; however, its major caveat is that the number of participating states has been growing since its deployment in 2003. As such, only seven states reported data in 2003 and this number has grown to 41 states by 2019, resulting in increased counts of police use of deadly force coinciding with increased participation.

The data sources from media organizations differ broadly in their definitions of police use of deadly force. The Washington Post's database is a highly reliable account of deaths that resulted from police officers firing their weapons and killing an individual. The Guardian attempted to identify all people killed by the police but collected and reported data only for 2015 and 2016. Reuter's database takes a different approach and focuses specifically on identifying all deaths that resulted from police officers using tasers on individuals.

Finally, counts from the two open-sourced databases on deaths by police also vary. Data from Mapping Police Violence closely overlap with those from media organizations because Mapping Police Violence accounts for all people killed by police while the officer was acting in a law enforcement capacity. Fatal Encounters reports much higher numbers of deaths by police because it defines such deaths as any occurring in any instance where a police officer was present or involved. This can include such factors as whether the officer was directly involved, as in a shooting, or more obscure instances, such as an officer responding to a suicide attempt or drug overdose that resulted in a death.

Much of the public's opinion about the police is based on the available data, how the data is presented, and the narrative around the data. By the same token, LEAs are rendered unable to make data-driven decisions or create a transparent relationship with the communities they serve if the data are not being tracked or managed in a way that makes data high quality and readily accessible. The currently available data on police use of force leads to both the public and law enforcement agencies not being able to make important decisions concerning policies, leadership, funding, or day-to-day choices that can keep the public safe and improve police-community relations. Furthermore, more rigorous and empirical analyses could be conducted with more detailed data that allow better insight into the effectiveness of the policing policies. Ultimately, the findings of this examination emphasize the lack of police data available on the subject, and the need for new approaches to better define, collect, manage, and analyze police officer uses of force at the agency level across the country.

Daniel Lawrence is a Research Scientist Center for CNA’s Justice Research and Innovation.

Hannah McLaurin is a CNA System Engineer for CNA's Center for Data Management and Analytics.