Data and Methodology

Data last updated 07/15/2026

July 13th, 2026, Mapping Police Violence updated its coding framework. While most fields that users engage with through the main site remain unchanged, the following fields have been amended: use_of_force (formerly cause_of_death), allegedly_armed_with (formerly, wapo_armed and allegedly_armed), call_for_service, and incident_footage (formerly wapo_body_cam). Please see more details in the change log.

Two fields will be updated in real time starting with the most recent records, due to analyst capacity. These fields are threat_perception (formerly wapo_threat_level) and incident_footage_public_release (new field). Please continue to use our legacy variable, wapo_threat_level, for data prior to July 2026. New records will only be coded for threat_perception, not wapo_threat_level.

To download the data for your own use offline, tap the “...” button at the top of the white portion of the table (below the light yellow background and “Mapping Police Violence” title), and then tap “Download CSV”.

We encourage users to review the full methodology which is accessible here or via web.

We have included extracts to summarize our methodology below:

MPV definition of police violence

At this time, our project tracks fatal police encounters, defined as: Any incident where a law enforcement officer (off-duty or on-duty) applies force resulting in the death of another person, whether it is considered “justified” or “unjustified” by the U.S. criminal legal system.

Additional details:

1. Definition of Law Enforcement Officer. Any person employed or authorized by a government entity (e.g. federal, state, or local) to enforce laws, maintain public order, or detain individuals. This includes police officers, sheriffs and deputies, state troopers, and federal agents. We generally do not include corrections officers, whose authority is typically limited to supervising incarcerated people within a facility. However, when corrections officers participate in enforcement actions outside that custodial context — such as a search operation involving a non-prisoner — we treat them as law enforcement officers for the purposes of this database, because they are acting under the authority of the state in a law enforcement capacity.
2. Vehicular Force. Cases when police have used a vehicle as a form of force, whether intentionally or non-intentionally, are included in our tracking (e.g. accidental strikes, PIT maneuvers). However, people who were killed by a civilian driver involved in a police chase, or an occupant of a car who crashed during a police chase, but the crash was not directly caused by police are not currently tracked according to our current criteria. Campaign Zero has separately started to internally track fatal vehicular chase incidents and will continue to collect and analyze that data.
3. Suicide by Cop. Cases where someone is killed by a law enforcement officer when they may be intending to die by police action (sometimes referred to as “Suicide by Cop”) are included as incidents of police violence.
4. Off-Duty/On-Duty Incidents. If a law enforcement officer caused the death of another person, it is considered a trackable incident of police violence according to our criteria whether or not they are “on-duty” or “off-duty”. The majority of all law enforcement contracts include policies and expectations that officers retain their authority and their responsibilities as police officers, both on-duty and off-duty, 24 hours a day. Additionally, it is worth noting that 97% of the incidents in MPV occurred while a police officer was acting in a law enforcement capacity. Thus, while MPV does include data on off-duty officers, it does not include killings by vigilantes or security guards who are not off-duty police officers. This is a gap in the field that should be further investigated, especially at the intersection of “Stand Your Ground Laws.”
5. Date of Incident. Our data is tracked against the date of the fatal police encounter, and not the date that the victim died. MPV tracks an incident whenever a victim dies as a result of actions by law enforcement, regardless of whether the death occurs during the encounter or subsequently, whether it may be days or even years later.

Data collection

MPV gathers data from a number of sources. While we consult official data sources from local and state government agencies, we believe it is important to continue collecting data from publicly accessible media sources. This allows us to identify gaps in government data and further triangulate and validate the data.

We use Google Alerts as our primary method for detecting news media mentions of police violence. We continue to develop and improve our own automated systems to filter out irrelevant news articles, maximize the comprehensiveness of the articles we do detect, and reduce potential human coding errors.

The steps we now follow to collect data are:

• 1

  Journalists report incidents

  Local or syndicated press reports publicly on details of a law enforcement encounter which is available online.

• 2

  News articles ingested to data lake

  Retrieve articles using a broad set of keywords relating to police violence using Google Alerts; store them in the data lake.

• 3

  Articles filtered for relevance

  Automations developed in-house scrape full article text, remove duplicate articles, and identifies and removes irrelevant articles.

• 4

  Articles assigned to researchers

  Custom Airtable automations developed in-house assign filtered articles to researchers for review.

• 5

  Two-step incident review

  1st researcher enters new incident information into Airtable from assigned articles. 2nd researcher checks for accuracy and fills in missing/updated info. Incidents then made public automatically.

Cadence of updates to the live tracker

It is important to note that incident records are created internally earlier than they are published to the MPV website. This is so that a general consensus of detail and data accuracy can be reached before it is included in the dataset. Thus, the date presented on the website’s “live tracker” does not reflect the most recent internally reviewed incident but instead shows the date of the most recent published data.

When we publish incident data

Initial accounts of incidents often come solely from police sources. Witness accounts and other differences in narrative details take more time to be released to the public, therefore we consider it our responsibility to allow sufficient time to pass prior to publishing incidents. This is both critical for data integrity, as well as to honor the families and loved ones of victims by not presenting inaccurate, distorted, or false accounting of details.

We do our best to maintain a 2-week lag between when incidents occur and when they are published on the site in order to keep the dataset both comprehensive and as up to date as possible. Data quality and accuracy should supersede speed of publishing given the sensitivity of the data.

General disclaimers

1. We treat each victim and incident record as a “living” document and make updates and additions to them as soon as we have more accurate information. We make corrections to any errors as soon as we discover or are made aware of them.
2. The data we gather comes almost entirely from media sources and public reports. We historically have not filed FOIA requests for records, and we do not include the details of non-public or sealed documents. This means that the MPV data is limited to the information released publicly and to the reporting work of our news media. For example, our data may occasionally not reflect the details or perspectives of key witnesses if these accounts were not sought out and reported on.
3. MPV captures the names and race of all involved officers in each case, whenever possible. We also track the outcomes of criminal investigations into the officers’ actions, including when officers are charged with a crime or are cleared of criminal charges, as well as any administrative discipline an officer faces, like suspension or a written reprimand. Where possible, we also track civil lawsuits related to these fatal encounters and their outcomes. This data is collected and updated as part of our ongoing monitoring of media reports.

A note on the police violence data sources and replication

With the exception of a few recent state-led initiatives to improve data collection efforts (e.g., California) around police violence, law enforcement agencies across the U.S. have failed to produce transparent, reliable data tracking lives taken by police. Mapping Police Violence was launched by Campaign Zero in 2015 to fill this critical gap.

There used to be two additional notable crowd-sourced datasets that were also working to track police violence:

1. Fatal Encounters. Fatal Encounters had been the longest-running dataset of police killing civilians, as well as of fatal incidents where an officer is present, such as suicides. While the criteria for whether an incident qualifies for inclusion in the dataset differ, the MPV methodology otherwise does not depart significantly from the original Fatal Encounters Methodology (found here). Fatal Encounters is no longer active, having concluded at the end of 2021.
2. The Washington Post. The Washington Post “Fatal Force” project tracked incidents where police killed someone with a firearm. The project stopped collecting new data at the end of 2024.

MPV is now the most comprehensive and longest running platform documenting fatal police violence in the United States. We believe that the data we collect, code, and make available should be public and free to access. It is not and should not be owned by any platform. We encourage communities, organizations, and other individuals to develop their own databases. This only strengthens the integrity and validity of the data by allowing us to study differences across datasets and improve data collection efforts when there are attempts to harmonize and analyze the data.

We are always open to partnerships and collaborations to improve the methodology and/or share best practices. We center collaboration and co-creation over ego, which is a departure from orthodox academic norms and culture.

Previous methodology

Our previous methodology primarily sourced data from Fatal Encounters, The Washington Post, publicly accessible media sources, and official data sources from local and state agencies required to report this data (e.g., California Department of Justice).

We want to acknowledge and thank Samuel Sinyangwe, a Campaign Zero co-founder for leading this work and the development of the initial methodology. We also want to acknowledge and thank D. Brian Burghart, founder of Fatal Encounters, who advised in the initial development of the project and collaborated on the methodology.

We are continually working to make our methodology more rigorous and generate automated ways to collect and validate data. Please contact us if you have any suggestions or would like to partner in any way.