Published using Google Docs
MPV Methodology (Public) | 10.03.22
Updated automatically every 5 minutes

Version: Oct 3, 2022

Mapping Police Violence


DATA & METHODOLOGY

mappingpoliceviolence.org  

Overview

This document includes varied methodology and process-related documentation to ensure transparency around our process and  allow the public to recreate the data and hold us accountable for the Mapping Police Violence (“MPV”) data platform.  

The following discussions with documentation are included below:

Preferred Citation:

 “Mapping Police Violence,” Campaign Zero, [Date accessed], https://mappingpoliceviolence.org/

A note on CZ Research + Data Products

We believe in exercising humility and course correcting when we make mistakes. Please contact us if you believe anything critically relevant is missing or misrepresented. While we’ve done our best to be comprehensive, we also have to balance accessibility and length. Furthermore,  we want to emphasize that we are research-driven, as opposed to exclusively data-driven. This allows us to center human experiences of people with  lived experience in the U.S. Criminal Legal System. We center co-creation over ego in all of our research and data initiatives.

Please email research.requests@campaignzero.org if you find any errors, problems, or have any questions about this documentation or dataset.

Contents

Police Violence Landscape        3

Research on Police Violence        4

MPV Definition of Police Violence        5

Mapping Police Violence        6

Context on Mapping Police Violence        6

Disclaimers & Limitations        8

Overview of Process        8

Cadence of Update        8

General Disclaimers        9

Incident Coding & Codebook        11

Process Map & Coding Instructions        11

Dataset Codebook        12

Codebook for Unique Variables        16

Armed/Unarmed Status        16

Alleged Threat Level        17

Fleeing        17

Encounter Types and Initial Reported Reason for Encounter        17

Helpful Coding Tips From Our Researchers        19

Second Review System        20

Replicating Google News Alerts        21

Existing Automations        22

Changelog        25

Methodology Revamp and Product Development        31

Limitations with Previous Methodology        31

Fatal Encounters & Mapping Police Violence Partnership        32

2022 Developments        32

Future Projects + Ongoing Initiatives        33

References        34

Appendix: Previous Data Methodology        35

Updating the Database        36

Definitions        36

Police Violence Landscape

Defining police violence

First, it is important to create distinctions between terms that have been synonymously employed to capture similar but different concepts. “Table 1” provides an overview of definitions  attempting to differentiate between commonly cited concepts.

Term

Definition

Source(s)

Police Violence

Currently, there is no agreed upon definition of police violence among scholars, the courts, and police officers (Harris 2017). Whereas some scholars define police violence as police use of excessive force during police-civilian encounters (Harris 2009), others argue that police violence may exceed the force required for an incident but may not be classified as excessive force (Alpert, Dunham, and MacDonald 2004). In Graham v. Connor (1989), the U.S. Supreme Court determined that police excessive force claims during investigatory stops, arrests, or seizures must be analyzed using the “objective reasonableness” standard under the Fourth Amendment.

Alpert, Geoffrey P., Roger G. Dunham, and John M. MacDonald. 2004. “Interactive Police-Citizen Encounters That Result in Force.” Police Quarterly 7(4):475–88. doi: 10.1177/1098611103260507.

Graham v. Connor (U.S. Supreme Court May 15, 1989).

Harris, Christopher J. 2009. “Police Use of Improper Force: A Systematic Review of the Evidence.” Victims & Offenders 4(1):25–41. doi: 10.1080/15564880701568470.

Harris, Christopher. 2017. “Reducing Violence and Aggression in Police Officers.” pp. 1–12 in The Wiley Handbook of Violence and Aggression. John Wiley & Sons, Ltd.

Police Abuse

Police abuse refers to police actions that limit citizens’ rights, typically receive minimal punishment, and may serve an economic and/or political function  (Bonner et al. 2018). Police abuse is generally characterized by civilian perceptions and complaints regardless of whether or not the police action is legally justified. Examples of police abuse include criminal acts as well as forms of disrespect, such as ethnic slurs, abusive language, and verbal threats (Albrecht 2017).

Albrecht, James F. 2017. Police Brutality, Misconduct, and Corruption: Criminological Explanations and Policy Implications. Cham, Switzerland: Springer.

Bonner, Michelle D., Mary Rose Kubal, Guillermina Seri, and Michael Kempa, eds. 2018. Police Abuse in Contemporary Democracies. New York, NY: Springer.

Police Misconduct

Police misconduct refers to police officers’ failure to comply with department regulations, policies, and procedures. Examples of police misconduct include, but are not limited to the following: criminal infractions, verbal abuse, excessive force, unwarranted stops, corruption, “chronic tardiness, disheveled appearance, routine unplanned absences, or excessive sick leave” (Weitzer and Tuch 2004; Albrecht 2017: 32).         

Researchers find that race is a significant predictor of perceptions of police misconduct in the United States, with African-Americans and Hispanics more likely to report experiences with police misconduct than their white counterparts (Weitzer and Tuch 2004).

Albrecht, James F. 2017. Police Brutality, Misconduct, and Corruption: Criminological Explanations and Policy Implications. Cham, Switzerland: Springer.

Weitzer, Ronald, and Steven A. Tuch. 2004. “Race and Perceptions of Police Misconduct.” Social Problems 51(3):305–25. doi: 10.1525/sp.2004.51.3.305.

Criminal Police Misconduct

Criminal police misconduct refers to police officers’ engagement in illicit activity that violates citizens’ Constitutional rights and/or obstructs justice. This may include criminal acts, such as excessive force, sexual misconduct, theft, false arrest, intentional neglect of serious medical needs of person in custody, preventing witnesses from reporting a crime, lying to federal, state, or local officials during an investigation, writing false reports, or concealing evidence (U.S. Department of Justice 2020).

U.S. Department of Justice. 2020. “Law Enforcement Misconduct.” Retrieved (https://www.justice.gov/crt/law-enforcement-misconduct).

Police Deviance

Police deviance includes an array of concepts that entail police involvement in criminal and/or noncriminal forms of misconduct. Police deviance may include police violence, police abuse of authority, police misconduct, criminal police misconduct, and police corruption (Albrecht 2017).

Albrecht, James F. 2017. Police Brutality, Misconduct, and Corruption: Criminological Explanations and Policy Implications. Cham, Switzerland: Springer.

Research on Police Violence

When examining data around police violence and outcomes of police killings, MPV data shows that Black residents in the US are approximately three times more likely to be killed in comparison to their white counterparts (Mapping Police Violence, 2022). However, it is important to note that police killings data represents policing outcomes and that the underlying interaction data is absent from this dataset. To better understand or make  inferences about the true nature, likelihood, or probability of police killings by demographics,  interaction level data must be considered. Research highlights how administrative data masks bias and therefore studies employing post-treatment data likely underestimate the severity of the bias (Knox, Lowe, and Mummolo, 2020). The lack of interaction-level data affects the outcome and assessment;the likelihood and probability of police violence could be severely underestimated.

Recent studies utilizing MPV data have found that the threshold for police killings of white people are much higher than those of Black people (DeAngelis, 2021). Similarly, other research using Fatal Encounters data finds that Black people are two times more likely to be killed by police “...even when there are no other obvious circumstances during the encounter that would make the use of deadly force reasonable” (Fagan and Campbell, 2020).

We encourage continued scholarship using the MPV, Fatal Encounters, and the Washington Post Police Shooting databases. For independent research, we encourage a combination of data sources to allow for validation and welcome any feedback, criticism, or issues that you come across with the data. Feel free to email info@campaignzero.org  with any questions.

MPV Definition of Police Violence

Any incident where a law enforcement officer (off-duty or on-duty) applies, on a civilian, lethal force resulting in the civilian being killed whether it is considered “justified” or “unjustified” by the U.S. Criminal Legal System.

A few disclaimers:

Mapping Police Violence

Context on Mapping Police Violence

In 2019, the FBI launched the data collection efforts for the National Use of Force database. However, participation is voluntary and granular-level agency data will only be released after passing a 80% threshold. Thus, to date, there is still no U.S. Government data source which captures all killings by US law enforcement with relevant contextual and demographic information of the victims.[1] 

The only governmental data source which could be used to identify police killings of civilians is the National Vital Statistics System (NVSS) maintained and monitored by the Centers for Disease Control (CDC). However, the dataset, consisting of birth and death certificate data,  suffers from severe undercounting and underreporting concerns. Specifically, the authors of one study, using several data sources, including MPV, find that NVSS undercounts police killings by 55%.

With the exception of a few recent state-led initiatives to improve data collection efforts (i.e. California) around police violence, law enforcement agencies across the U.S. have failed to produce transparent, reliable data updates tracking lives taken by police. This critical gap has been filled by three different non-governmental data sources. The most comprehensive and longest-running platform is Fatal Encounters (“FE”).  While FE fills a critical void, a core tenet in social science research is to be able to reproduce data to allow for further validation and rigor (Firebaugh, 2017). Thus, the need to continue this work and continue separate data collection efforts to allow for more vigilance, validity, and accessibility is critical given that there is no way to understand the true scale of police violence in the U.S.

MPV was launched in 2015 to continue data collection efforts on police violence.

There are three crowd-sourced datasets including MPV which are still active:

“Figure 1” includes a visual of the police violence data sources.[2] 

FIGURE 1

The MPV methodology does not depart significantly from the Fatal Encounters Methodology which can be found here. The primary difference between FE and MPV generally are as follows:

Disclaimers & Limitations

The following disclaimers and limitations are critical to note.

Overview of Process

We have a dedicated team of researchers who are assigned incidents to review on a daily basis through an automated process:

  1. We have automated the ingestion of Google and Meltwater alerts into our AirTable backend.
  2. Reviewers are assigned automatically every day and they indicate after an incident is reviewed.
  3. Second reviewers indicate after they have completed review which will automatically trigger the incident to be published onto the database.

Cadence of updates to the live tracker

It is important to note that incidents are reviewed at different times. This can be for several reasons, including delays between an incident occurring and being mentioned in the news, or second reviewers flagging incidents for further review due to possible reliability concerns. Thus, the date presented on the live tracker does not reflect the most recent internally reviewed  incident. Our research and product team do their best to keep the live tracker as comprehensive and up-to-date as possible.

A note on 2022 data.

Since transitioning and revamping our methodology, there will be a significant delay in the publishing of incidents due to the complete revamp, development, and improvement of the platform and coding process.  Further discussion about the revamp can be found below.

We believe data quality and accuracy should supersede speed of publishing given the sensitivity of the data.  

Second Reviewer Incidents

Given the initial accounts of incidents being police-centric and the significant differences in narrative and details released, we believe there is a responsibility to allow for a sufficient amount of time to pass prior to publishing incidents that may misrepresent the true details of the incident. This is both critical for data integrity and to honor the families and loved ones of victims to not present inaccurate, distorted, or false accounting of details.

General Disclaimers

The same data limitations noted by FE  also apply to MPV given the similarities of how the data is sourced and generated. As noted by Fatal Encounters,

“The vast majority of these records come from media sources and police records. That means only one version of the story—the police story—is generally told in the descriptions. Rarely do news media seek out family members and friends to balance or contradict police narratives. While we verify our data against media reports, sometimes the information presented is so wildly inaccurate that we instead include accurate but conflicting information. Also, be aware that this document is "living." We repair any errors or eliminate any duplicates as we discover them. Errors can be reported through the fatalencounters.org website.”

Additional disclaimers and clarifications are noted below:

Cadence of Merge Update: 6 Months

Next Scheduled Merge: 12/2022

A note on the police violence data sources and replication

We posit that the data collected, coded, and made available belongs to the public. 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 the differences across datasets and improve data collection efforts when there are attempts to harmonize and study the data. For example, a recent study has found that while the three major datasets are similar, they have started to become more dissimilar over recent years (Comer and Ingram, 2022).

Thus, 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 can be a departure from orthodox academic norms and culture.

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 the initial development of the project and continues to advise and collaborate on the methodology.

Incident Coding & Codebook

Incident coding and unique variable coding are broken down into the following:

Process Map & Coding Instructions

The initial methodology for incident coding was developed by a team of Campaign Zero researchers led by Samuel Sinyangwe. D. Brian Burghart  helped play a critical role in providing initial instructions on how to code.

Description/Instructions

Variable(s)

Step 1.

Incident Identification

  • Meltwater & Google News Alerts are filtered into an Alerts Sheet which gets updated automatically every morning and automatically assigns alerts for every researcher to review.
  • When reviewing the alert, researchers need to ensure that the incident doesn’t already exist in the database or confirm it is not an alert related to police violence.
  • If it already exists, the researcher will check to see if there is any updated information, make any amendments/additions, and proceed to reviewing the next alert.
  • If it is not related, they will mark that they reviewed the alert and proceed to reviewing the next alert.
  • Link to news article
  • Incident Description
  • Article Publish Date

Step 2.

New Incident Info Entry

  • If it is a new incident, the researcher will “create a new observation” and start entering in Geographic and Demographic-level information
  • Employ Interpretive Analysis codebook to code out incident variables appropriately.
  • Check 1st review to indicate that the alert/incident is ready for second review.
  • Victim Demographics
  • Geographic-level information
  • Law Enforcement Agency & Officer Info
  • Armed/Unarmed
  • Encounter Type
  • WaPo Variables
  • FE Variables

Step 3.

Second Reviewer

  • Second Reviewer will look up name of victim (if available) on Google and Meltwater  to check for an updated article
  • If name is not available, the reviewer will utilize contextual information and prior link to see if there are follow-up articles available.
  • Reviewer will complete missing information and/or correct existing information
  • If there are any serious concerns, reviewer will flag in comments column for a third review
  • Second Reviewer
  • Updated Link

Step 4.

Auto Publish

  • Once completed, the reviewer will list their name for second review to trigger logic which will automatically publish the incident to MPV platform.
  • Second Reviewer
  • Updated Link

Dataset Codebook

The public version of the dataset can be found here and is updated regularly on a daily basis.

Field

Description

Manual/ Automated Entry

1

name

Name of victim killed by law enforcement

Manual

2

age

Age of victim at time of death

Manual

3

gender

Indicated gender by news/official reports

Manual

  • Male
  • Female
  • Unknown
  • Transgender Male
  • Transgender Female
  • Non-Binary/Gender non-conforming

4

race

Race of victim according to news/official reports

Manual

  • Black
  • White
  • Native American
  • Asian
  • Unknown
  • Native Hawaiian and Pacific Islander

5

victim_image

URL to an image of victim

Manual

6

date

Date of lethal action on victim

Manual

7

street_address

Address of where incident took place

Manual

8

city

City of where incident took place

Manual

9

state

State where incident took place

Manual

10

zip

Zip code of where incident took place

Automated

11

county

County of where incident took place

Manual

12

agency_responsible

Agency of law enforcement officer who took lethal action and responsible for killing

Manual

13

ori

Law enforcement unique identifier assigned by US Department of Justice (DoJ)

Automated

14

cause_of_death

Highest level of force used  by law enforcement officer

Manual

  • Gun
  • Taser
  • Taser, Gunshot
  • Unknown
  • Vehicle
  • Vehicle, Gunshot
  • Other Physical Force

15

circumstances

Description of situation/News extract surrounding situation

Manual

16

disposition_official

Case status against law enforcement officer

Manual

17

officer_charged

Outcome of case against law enforcement officer

Manual

18

news_urls

Url link(s) to news reports or official reports of incident

Manual

19

signs_of_mental_illness

Indicates whether official/news reports mentioned any signs of mental illness or mental health crises

Manual

  • Yes
  • No
  • Unknown
  • Drug or Alcohol Use

20

allegedly_armed

Indicates whether victim was armed according to news/official sources

Manual

  • Allegedly Armed
  • Unarmed
  • Unclear
  • Vehicle

21

wapo_armed

The Washington Post created variable indicating that the victim was armed with some sort of implement that a police officer believed could inflict harm

Manual

  • Gun

22

wapo_threat_level

The Washington Post created a variable indicating  incidents where officers or others were shot at, threatened with a gun, attacked with other weapons or physical force, etc. The attack category is meant to flag the highest level of threat. The other and undetermined categories represent all remaining cases.

Automated

  • Brandished Weapon
  • Used weapon
  • Sudden threatening movement
  • Undetermined
  • Other
  • Attack
  • None
  • Clear

23

wapo_flee

The Washington Post created variable indicating that the victim was moving away from officers

Automated

  • Foot
  • Car
  • Not fleeing
  • Other
  • Car, foot
  • Foot, car

24

wapo_body_camera

The Washington Post created a variable  indicating an officer was wearing a body camera and it may have recorded some portion of the incident.

Manual

  • No
  • Yes
  • Surveillance Video
  • Dash Cam Video
  • Bystander Video
  • Taser Video

25

wapo_id

A  unique identifier for each victim assigned by The Washington Post DB.

Automated

26

off_duty_killing

Indicates whether officer was working in an official capacity at the time lethal action was taken

Manual

  • Off-Duty

27

geography

Trulia developed geography measure based on zip code

Automated

This measure permits a more detailed and granular analysis of the geographic landscape of police violence, including differences between urban/suburban/rural zip codes within cities, counties or broader metropolitan areas that would more broadly be coded as urban areas using other methods.

28

mpv_id

Unique ID assigned to each MPV observation and matched with WaPo and FE incidents.

Automated

29

fe_id

Unique ID assigned to each FE observation.

Automated

30

encounter_type

The crime, if any, that caused the police to directly engage the victim.

Manual

  • Violent Crime
  • Other Crimes Against People
  • Person with a Weapon
  • Other Non-Violent Offenses
  • Mental Health/Welfare Check
  • Domestic Disturbance
  • Traffic Stop
  • None/Unknown

31

initial_reason

Initial reported reason(s) for police to be on the scene prior to using deadly force

Manual

32

officer_names

Name of officer responsible for highest level of force

Manual

33

officer_races

Race of officer responsible for highest level of force

Manual

34

officer_known_past_shootings

Indicates if officer has had a previous incident of police violence in the database

Manual

35

call_for_service

Indicates whether the incident was the result of a 911 call or an officer-initiated action

Manual

  • Yes
  • No
  • Unknown

36

tract

Indicates the U.S. Census Bureau unique identifier for the census tract

Automated

37

urban_rural_uspsai

Urbanization Perceptions Small Area Index (UPSAI) measure developed by the U.S. Department of Housing and Urban Development. Assigns urban/suburban/rural geographies to census tracts based on how residents classify their own neighborhoods in the American Housing Survey, a survey of 76,000 households in 2017.

Automated

  • Urban
  • Suburban
  • Rural
  • Undetermined

38

urban_rural_nchs

Urban-Rural Classification Scheme developed by the National Center for Health Statistics. Broadest in geography - assigning urban/suburban/rural geographies based on 2013 estimates of each county’s population and proximity to the Metropolitan Statistical Area’s principal city.

Automated

  • Large Central Metro
  • Medium Metro
  • Small Metro
  • Micropolitan
  • Large Fringe Metro
  • Non-Core
  • Undetermined

39

hhincome_median_census_tract

Median household income

Automated

40

latitude

The latitude of the incident

Automated

41

longitude

The longitude of the incident

Automated

42

pop_total_census_tract

Total population of the census tract of incident

Automated

43

pop_hispanic_census_tract

Hispanic population of the census tract of incident

Automated

44

pop_white_census_tract

White population of the census tract of incident

Automated

45

pop_black_census_tract

Black population of the census tract of incident

Automated

46

pop_native_american_census_tract

Native American population of the census tract of incident

Automated

47

pop_asian_census_tract

Asian population of the census tract of incident

Automated

48

pop_pacific_islander_census_tract

Pacific Islander population of the census tract of incident

Automated

49

pop_other_multiple_census_tract

Other/Multiple population of the census tract of incident

Automated

50

congressional_district_113

Congressional district where incident took place

Automated

Codebook for Unique Variables

Armed/Unarmed Status

Developed by: Campaign Zero, led by Samuel Sinyangwe

A person is coded as Unarmed/Did Not Have a Weapon in the database if they were one or more of the following:

A person was coded as having a Vehicle as a weapon if they were one or more of the following:

A person was coded as Allegedly Armed in the database if they:

Alleged Threat Level

Developed by: The Washington Post

Extracted from The Washington Post Methodology which can be accessed here.

MPV researchers are only responsible for coding incidents where a gun is not the highest level of force applied in the incident since WaPo data only includes observations with a gun present.

“The threat_level column was used to flag incidents for the story by Amy Brittain in October 2015.

As described in the story, the general criteria for the attack label was that there was the most direct and immediate threat to life. That would include incidents where officers or others were shot at, threatened with a gun, attacked with other weapons or physical force, etc. The attack category is meant to flag the highest level of threat. The other and undetermined categories represent all remaining cases. Other includes many incidents where officers or others faced significant threats.”

Fleeing

Developed by: The Washington Post

Extracted from The Washington Post Methodology which can be accessed here.

“News reports have indicated the victim was moving away from officers

The threat column and the fleeing column are not necessarily related. For example, there is an incident in which the suspect is fleeing and at the same time turns to fire at gun at the officer. Also, attacks represent a status immediately before fatal shots by police while fleeing could begin slightly earlier and involve a chase.”

Encounter Types and Initial Reported Reason for Encounter

Developed by: Campaign Zero, led by Samuel Sinyangwe

We believe it is important to continue the coding of this variable. The coding for these unique variables started in 2017. MPV has expanded the scope of data collection to include information on the initial reported reason(s) for police to be on the scene prior to using deadly force. This information is obtained from a review of existing media reports on each case as well as statements from police, prosecutors, and other officials. These initial reported reasons are grouped into broader Encounter Types that are standardized within the following taxonomy, ranked in order of severity whereby cases are coded according to the most severe encounter type:

If it is a Part I-related violent crime, then you would indicate beforehand.  Part I crimes include the following violent crimes: rape, aggravated assault, robbery, and human trafficking.

Helpful Coding Tips From Our Researchers

Last Updated: Mar 1, 2022        

Second Review System

The previous system employed Google Sheets to collect and code data. While Google Sheets offer advantages, it makes it difficult to deploy a systematic approach to allow for second-review the way AirTable permits.

We believe having a two-tiered review system is critical to both detecting human errors in addition to machine errors such as news alert concerns. This will provide further validity and improve data integrity. Nevertheless, if you find any errors or want to flag any concerns, please reach out to Campaign Zero. Your feedback and concerns are important and co-create a better platform.

Instructions for second-reviewers

  1. Second Reviewer will look up the name of the victim (if available) on Google and Meltwater to check for an updated article
  2. If name is not available, the reviewer will utilize contextual information and prior link to see if there are follow-up articles available.
  3. Reviewer will complete missing information and/or correct existing information
  4. Paste new article link in news_url field under previous link(s), separated by newline (press enter/return).
  5. If there are any serious concerns, reviewer will flag in comments column for a third review

Who qualifies as a second-reviewer

Second reviewer qualifications:

Replicating Google News Alerts

There are two ways to develop replicate the news results and would like to cite D. Brian Burghart for his expertise and collaboration:

Key Words

Key Words: authorities shot killed;  chase drowned;  custody OR handcuffed died;  deputy, chase, killed;  deputy, shot, killed;  died taser OR "stun gun" OR "stungun";  law enforcement pursuit kill;  law enforcement shot killed;  officer involved shootings;  officer police intersection chase killed;  officer-involved;  officer, police, crash, killed;  officers justified OR cleared;  police standard administrative leave;  police-involved;  police, chase, killed;  standoff dead;  standoff self-inflicted;  trooper-involved;  trooper, chase, killed;  trooper, shot, killed

Replicating News Feed

The 21 alerts are as follows:

  1. https://www.google.com/alerts/feeds/09528353354863773528/7526401424772565040
  2. https://www.google.com/alerts/feeds/09528353354863773528/3986613816392777988
  3. https://www.google.com/alerts/feeds/09528353354863773528/12387128514723811524
  4. https://www.google.com/alerts/feeds/09528353354863773528/6042935056065544825
  5. https://www.google.com/alerts/feeds/09528353354863773528/2495735278215740846
  6. https://www.google.com/alerts/feeds/09528353354863773528/7457298848418186377
  7. https://www.google.com/alerts/feeds/09528353354863773528/14298085485813281829
  8. https://www.google.com/alerts/feeds/09528353354863773528/7457298848418189098
  9. https://www.google.com/alerts/feeds/09528353354863773528/15552000386701586242
  10. https://www.google.com/alerts/feeds/09528353354863773528/15552000386701583880
  11. https://www.google.com/alerts/feeds/09528353354863773528/16180856559082508744
  12. https://www.google.com/alerts/feeds/09528353354863773528/16180856559082511399
  13. https://www.google.com/alerts/feeds/09528353354863773528/17499577270006445413
  14. https://www.google.com/alerts/feeds/09528353354863773528/2179455100157334934
  15. https://www.google.com/alerts/feeds/09528353354863773528/2179455100157337042
  16. https://www.google.com/alerts/feeds/09528353354863773528/2179455100157334247
  17. https://www.google.com/alerts/feeds/09528353354863773528/8145116068315731327
  18. https://www.google.com/alerts/feeds/09528353354863773528/6108069055156697586
  19. https://www.google.com/alerts/feeds/09528353354863773528/9230368609021506778
  20. https://www.google.com/alerts/feeds/09528353354863773528/8494202513583889355
  21. https://www.google.com/alerts/feeds/09528353354863773528/470642475239085179

Existing Automations

When coding, which fields should be entered manually? Anything with a “yes” in the table below. Other fields will be automatically populated.

Field

Manual input required?

Notes

Manual input priority

1

name

yes

high

2

age

yes

moderate

3

gender

yes

moderate

4

race

yes

high

5

victim_image

yes

moderate

6

date

yes

high

7

street_address

yes

high

8

city

yes

high

9

state

yes

high

10

zip

11

county

sometimes

*Required if no city is available (for non-urban areas), and no street address is available (for multi-county cities)

12

agency_responsible

yes

high

13

ori

14

cause_of_death

yes

high

15

circumstances

yes

moderate

16

disposition_official

yes

17

officer_charged

yes

18

news_urls

yes

Append any new urls that inform any change to data in any field. Separate URLs by newlines.

high

19

signs_of_mental_illness

yes

20

allegedly_armed

yes

21

wapo_armed

yes

this will be split into mpv_armed and wapo_armed

22

wapo_threat_level

23

wapo_flee

24

wapo_body_camera

yes

this will be split into body_camera and wapo_body_camera

25

wapo_id

26

off_duty_killing

yes

27

geography

28

mpv_id

29

fe_id

30

encounter_type

yes

31

initial_reason

yes

32

officer_names

yes

33

officer_races

yes

34

officer_known_past_shootings

yes

35

call_for_service

yes

36

tract

37

urban_rural_uspsai

38

urban_rural_nchs

39

hhincome_median_census_tract

40

latitude

41

longitude

42

pop_total_census_tract

43

pop_hispanic_census_tract

44

pop_white_census_tract

45

pop_black_census_tract

46

pop_native_american_census_tract

47

pop_asian_census_tract

48

pop_pacific_islander_census_tract

49

pop_other_multiple_census_tract

50

congressional_district_113

Changelog

[draft / projected] Feb 28, 2022

Summary of Changes | Data moved from Google Sheets to Airtable. Column names have been shortened, made lowercase, and don’t include spaces. Where appropriate, information contained in previously existing column names moved to column descriptions. Data types for some columns have been changed, for data integrity and validation purposes.

For all columns that were changed to the single-select data type, single-select options were cleaned so that:

Field

Type

Change

1

name

Single Line Text

field name change | Victim’s name → name

added description | → Victim’s name

2

age

Number (integer 2)

field name change | Victim’s age → age

added description | → Victim’s age

3

gender

Single Select

add option | Transgender Female

add option | Transgender Male

add option | Non-Binary

field name change | Victim’s gender → gender

added description | → Victim’s gender

4

race

Single Select

field name change | Victim’s race → race

added description | → Victim’s race

5

victim_image

Single Line Text

field name change | URL of image of victim → victim_image

added description | → URL of image of victim

6

date

Date - Local (2/16/2022)

field name change | Date of Incident (month/day/year) → date

added description | → Date of Incident (local)

7

street_address

Long Text

field name change | Street Address of Incident → street_address

added description | → First line of address where incident occurred (eg, 123 Main Street)

8

city

Single Select

field name change | City → city

added description | → City/town/village where incident occurred (eg, Brooklyn)

9

state

Single Select

modify option | NB -> NE

field name change | State → state

added description | → State (or DC) where incident occurred

10

zip

Single Line Text

field name change | Zipcode → zip

added description | → 5-digit Zip Code where incident occurred

11

county

Single Select

field name change | County → county

added description | → County where incident occurred (e.g. Kings)

12

agency_responsible

Multiple Select

field name change | Agency responsible for death → agency_responsible

added description | → Agency responsible for death

13

ori

Multiple Select

field name change | ORI Agency Identifier (if available) → ori

added description | → ORI Agency Identifier (if available)

14

cause_of_death

Multiple Select

field name change | Cause of death → cause_of_death

added description | → Cause of death

modify option | separate all options by comma

modify option | Beaten/Bludgeoned with instrument -> Beaten

modify option | Chemical Agent/Pepper Spray -> Chemical Agent

15

circumstances

Long Text

field name change | A brief description of the circumstances surrounding the death → circumstances

added description | → A brief description of the circumstances surrounding the death

16

disposition_official

Single Select

field name change | Official disposition of death (justified or other) → disposition_official

added description | → Official disposition of death (eg pending investigation, justified, charged, etc)

17

officer_charged

Single Select

field name change | Criminal Charges? → officer_charged

added description | → Was/were the officer(s) charged with a crime?

18

news_urls

Long Text

field name change | Link to news article → news_urls

added description | → Links to related news articles

19

signs_of_mental_illness

Single Select

field name change | Symptoms of mental illness? → signs_of_mental_illness

added description | → Symptoms of mental illness? (yes, no, drug or alcohol use, unknown)

20

allegedly_armed

Single Select

field name change | Armed/Unarmed Status → allegedly_armed

added description | → Was the victim alleged to have been armed? (allegedly, unarmed, vehicle, unclear)

21

wapo_armed

Single Select

field name change | Alleged Weapon (Source: WaPo and Review of Cases Not Included in WaPo Database) → wapo_armed

added description | → Alleged Weapon (Source: WaPo and Review of Cases Not Included in WaPo Database)

22

wapo_threat_level

Single Select

field name change | Alleged Threat Level (Source: WaPo) → wapo_threat_level

added description | → Alleged Threat Level (Source: WaPo)

23

wapo_flee

Single Select

field name change | Fleeing (Source: WaPo) → wapo_flee

added description | → Was the victim fleeing? (Source: WaPo)

24

wapo_body_camera

Single Select

field name change | Body Camera (Source: WaPo) → wapo_body_camera

added description | → Body Camera (Source: WaPo)

25

wapo_id

Number (integer 2)

field name change | WaPo ID (If included in WaPo database) → wapo_id

added description | → WaPo ID (if it exists in https://github.com/washingtonpost/data-police-shootings)

26

off_duty_killing

Single Select

field name change | Off-Duty Killing? → off_duty_killing

added description | → Off-Duty Killing?

27

geography

Single Select

field name change | Geography (via Trulia methodology based on zipcode population density: http://jedkolko.com/wp-content/uploads/2015/05/full-ZCTA-urban-suburban-rural-classification.xlsx ) → geography

added description | → Geography: rural/suburban/urban (via Trulia methodology based on zipcode population density: http://jedkolko.com/wp-content/uploads/2015/05/full-ZCTA-urban-suburban-rural-classification.xlsx )

28

mpv_id

Number (integer 2)

field name change | MPV ID → mpv_id

added description | Unique identifier for each incident in this dataset (Mapping Police Violence)

29

fe_id

Number (integer 2)

field name change | Fatal Encounters ID → fe_id

added description | → Fatal Encounters ID (if it exists)

30

encounter_type

Single Select

field name change | Encounter Type (DRAFT) → encounter_type

added description | → Encounter Type

31

initial_reason

Single Select

field name change | Initial Reported Reason for Encounter (DRAFT) → initial_reason

added description | → Initial Reported Reason for Encounter

32

officer_names

Single Line Text

field name change | Names of Officers Involved (DRAFT) → officer_names

added description | → Names of Officers Involved

add text | Catalin Panov

33

officer_races

Single Select

field name change | Race of Officers Involved (DRAFT) → officer_races

added description | → Race of Officers Involved

delete text | Catalin Panov

34

officer_known_past_shootings

Single Line Text

field name change | Known Past Shootings of Officer(s) (DRAFT) → officer_known_past_shootings

added description | → Known Past Shootings of Officer(s)

35

call_for_service

Single Select

field name change | Call for Service? (DRAFT) → call_for_service

added description | → Call for Service? (yes/no/unavailable)

36

tract

Number (integer 2)

field name change | Census Tract Code → tract

added description | → Census tract code (“TRACT”)

37

urban_rural_uspsai

Single Select

field name change | HUD UPSAI Geography → urban_rural_uspsai

added description | → HUD UPSAI Geography (urban/suburban/rural/undetermined)

38

urban_rural_nchs

Single Select

field name change | NCHS Urban-Rural Classification Scheme Codes (https://www.cdc.gov/nchs/data_access/urban_rural.htm) → urban_rural_nchs

added description | → NCHS Urban-Rural Classification Scheme Codes (https://www.cdc.gov/nchs/data_access/urban_rural.htm)

39

hhincome_median_census_tract

Number (integer 2)

field name change | Median household income ACS Census Tract → hhincome_median_census_tract

added description | → Median household income ACS Census Tract (B19013_001)

40

latitude

Number (decimal)

field name change | Latitude → latitude

added description | → Latitude

41

longitude

Number (decimal)

field name change | Longitude → longitude

added description | → Longitude

42

pop_total_census_tract

Number (integer 2)

field name change | Total Population of Census Tract 2019 ACS 5-Year Estimates → pop_total_census_tract

added description | → Total population of census tract 2019 ACS 5-year estimates (DP05_0070)

43

pop_hispanic_census_tract

Percent (non-negative)

field name change | Hispanic Percent of the Population ACS → pop_hispanic_census_tract

added description | → Hispanic (any race) percent of the population from ACS (DP05_0071)

44

pop_white_census_tract

Percent (non-negative)

field name change | White Non-Hispanic Percent of the Population ACS → pop_white_census_tract

added description | → White alone Non-Hispanic percent of the population from ACS (DP05_0077)

45

pop_black_census_tract

Percent (non-negative)

field name change | Black Non-Hispanic Percent of the Population ACS → pop_black_census_tract

added description | → Black or African American alone Non-Hispanic percent of the population from ACS (DP05_0078)

46

pop_native_american_census_tract

Percent (non-negative)

field name change | Native American Percent of the Population ACS → pop_native_american_census_tract

added description | → American Indian and Alaska Native alone Non-Hispanic percent of the population from ACS (DP05_0079)

47

pop_asian_census_tract

Percent (non-negative)

field name change | Asian Percent of the Population ACS → pop_asian_census_tract

added description | → Asian alone Non-Hispanic percent of the population from ACS (DP05_0080)

48

pop_pacific_islander_census_tract

Percent (non-negative)

field name change | Pacific Islander Percent of the Population ACS → pop_pacific_islander_census_tract

added description | → Native Hawaiian and Other Pacific Islander alone Non-Hispanic percent of the population from ACS (DP05_0081)

49

pop_other_multiple_census_tract

Percent (non-negative)

field name change | Other/Two or More Race Percent of the Population ACS → pop_other_multiple_census_tract

added description | → Other and multiple races percent of the population from ACS

50

congressional_district_113

Single Select

field name change | Congressional District → congressional_district_113

added description | → Congressional District (State abbreviation + “CD113FP” FIPS code, eg AL1. This version represents the 113th congress boundaries from the 2012-2022 redistricting cycle)

Methodology Revamp and Product Development

We believe that the data should not be owned by any entity or individuals. Moreover, we believe that this project needs to live beyond us and is critical that it can continue without the reliance on any single individual or group of individuals. This required the start of developing a system, processes, and applications that would allow this project to not only scale but become reproducible and sustainable.

In February 2022, Campaign Zero launched a revamped Mapping Police Violence intended to allow for more accessible use of the data. Prior to the launch and since the revamp, CZ has worked tirelessly on developing the best ways to improve data collection, coding, and dissemination of incidents of police violence. While there have been changes to the cadence of updates and a delay in the publishing of incidents to allow for experimentation of different methods, this brief provides an overview of how the innovative new system allows for more transparency, rigor, and scalability. This was for several reasons, including staff transitions which revealed that the methodology being used beforehand had an overreliance on a small number of researchers and involved several manual processes that created serious concerns around data credibility, sustainability, and scalability over the long-term.

We’re working with data organizations to co-develop a methodology that represents different intersections of identities in order to ensure that we’re properly capturing experiences and honoring victims. This includes university partnerships, and most notably, our most recent partnership with Fatal Encounters, the first dataset on police killings in the US that dates back to 2020. We started working with FE’s founder to continue to refine our methodology and in September 2022, we formalized a partnership where Fatal Encounters would merge the data efforts under the Mapping Police Violence arm to allow for more rigor, transparency, and sustainability.

Limitations with Previous Methodology

Fatal Encounters & Mapping Police Violence Partnership

In 2021, Fatal Encounters and the CZ Research & Data team operating MPV came together to build out a plan to build out a platform that is not only public, replicable, and rigorous, but something that would be sustainable to support the movement to help inform solutions that bring us to a world beyond policing and ending police violence.

During the transition of Mapping Police Violence platform, we have worked closely and in partnership with Fatal Encounters, led by D. Brian Burghart. Not only was Fatal Encounters the first database of its kind, but it was the foundation and origins of Mapping Police Violence. While initially the partnership was focused on improving methodology, accessibility, and transparency of the data generation process, we have been grateful to take our partnership to the next step. We realized while revamping the methodology that the previous methodology was heavily reliant on single researchers or small groups of researchers which introduced concerns around scalability, sustainability, and concerns around data norming and integrity.

We believe this work is larger than any one individual or organization, it is about building a sustainable, rigorous, and scalable product that will live beyond all of us.

2022 Developments

Future Projects + Ongoing Initiatives

References

Alpert, Geoffrey P., Roger G. Dunham, and John M. MacDonald. 2004. “Interactive Police-Citizen Encounters That Result in Force.” Police Quarterly 7 (4): 475–88. https://doi.org/10.1177/1098611103260507.

Comer, Benjamin P., and Jason R. Ingram. 2022. “Comparing Fatal Encounters, Mapping Police Violence, and Washington Post Fatal Police Shooting Data from 2015–2019: A Research Note.” Criminal Justice Review, January, 07340168211071014. https://doi.org/10.1177/07340168211071014.

DeAngelis, Reed T. 2021. “Systemic Racism in Police Killings: New Evidence From the Mapping Police Violence Database, 2013–2021.” Race and Justice, October, 21533687211047944. https://doi.org/10.1177/21533687211047943.

Fagan, Jeffrey, and Alexis Campbell. 2020. “Race and Reasonableness in Police Killings.” B.U. L. Rev. 100 (January): 951.

“Fatal Police Violence by Race and State in the USA, 1980–2019: A Network Meta-Regression.” 2021. The Lancet 398 (10307): 1239–55. https://doi.org/10.1016/S0140-6736(21)01609-3.

“Graham v. Connor, 490 U.S. 386 (1989).” n.d. Justia Law. Accessed April 14, 2022. https://supreme.justia.com/cases/federal/us/490/386/.

Harris, Christopher J. 2009. “Police Use of Improper Force: A Systematic Review of the Evidence.” Victims & Offenders 4 (1): 25–41. https://doi.org/10.1080/15564880701568470.

Knox, Dean, Will Lowe, and Jonathan Mummolo. 2020. “Administrative Records Mask Racially Biased Policing.” American Political Science Review 114 (3): 619–37. https://doi.org/10.1017/S0003055420000039.

“Law Enforcement Misconduct.” 2016. September 26, 2016. https://www.justice.gov/crt/law-enforcement-misconduct.

Weitzer, Ronald, and Steven A. Tuch. 2004. “Race and Perceptions of Police Misconduct.” Social Problems 51 (3): 305–25. https://doi.org/10.1525/sp.2004.51.3.305.

“Wisconsin Trooper Faced down a Gunman Who Planned to Go out Fighting | The Washington Post.” n.d. Accessed April 14, 2022. https://www.washingtonpost.com/sf/investigative/2015/10/24/on-duty-under-fire/.

Appendix: Previous Data Methodology

As of 12/31/2021

Law enforcement agencies across the country have failed to provide us with even basic information about the lives they have taken. And while the Deaths in Custody Reporting Act mandates this data be reported, its unclear whether police departments will actually comply with this mandate and, even if they do decide to report this information, it could be several years before the data is fully collected, compiled and made public.

We cannot wait to know the true scale of police violence against our communities. In a country where at least three people are killed by police every day, we cannot wait for police departments to provide us with these answers. The maps and charts on this site aim to provide us with some insights into patterns of police violence across the country. They include information on over 9,000 killings by police nationwide since 2013. 97 percent of the killings in our database occurred while a police officer was acting in a law enforcement capacity. Importantly, these data do not include killings by vigilantes or security guards who are not off-duty police officers.

This information has been meticulously sourced from official police use of force data collection programs in states like California, Texas and Virginia, combined with nationwide data from the Fatal Encounters database, an impartial crowdsourced database on police killings. We've also done extensive original research to further improve the quality and completeness of the data; searching social media, obituaries, criminal records databases, police reports and other sources to identify the race of 90 percent of all victims in the database.

We believe the data represented on this site is the most comprehensive accounting of people killed by police since 2013. Note that the Mapping Police Violence database is more comprehensive than the Washington Post police shootings database: while WaPo only tracks cases where people are fatally shot by on-duty police officers, our database includes additional incidents such as cases where police kill someone through use of a chokehold, baton, taser or other means as well as cases such as killings by off-duty police. A recent report from the Bureau of Justice Statistics estimated approximately 1,200 people were killed by police between June, 2015 and May, 2016. Our database identified 1,104 people killed by police over this time period. While there are undoubtedly police killings that are not included in our database (namely, those that go unreported by the media), these estimates suggest that our database captures 92% of the total number of police killings that have occurred since 2013. We hope these data will be used to provide greater transparency and accountability for police departments as part of the ongoing work to end police violence in America.

Updating the Database

The Mapping Police Violence database is updated with new cases every weekend. Since it takes roughly a week to find and code cases, the data and analysis on this site usually includes all killings by police that occurred up to the end of the prior week. Additionally, we conduct quarterly reviews of cases from prior years to ensure our database remains complete and up-to-date.

Definitions

Police Killing: A case where a person dies as a result of being shot, beaten, restrained, intentionally hit by a police vehicle, pepper sprayed, tasered, or otherwise harmed by police officers, whether on-duty or off-duty.

A person was coded as Unarmed/Did Not Have a Weapon in the database if they were one or more of the following:

A person was coded as having a Vehicle as a weapon if they were one or more of the following:

A person was coded as Allegedly Armed in the database if they:

Urban/Suburban/Rural Geography Measures:

In order to assess changes in the geography of killings by police over time, the Mapping Police Violence database includes three different geographic measures that each have their own strengths and limitations:

Encounter Types and Initial Reported Reason for Encounter:

Since 2017, Mapping Police Violence has expanded the scope of data collection to include information on the initial reported reason(s) for police to be on the scene prior to using deadly force. This information is obtained from a review of existing media reports on each case as well as statements from police, prosecutors, and other officials. These initial reported reasons are grouped into broader Encounter Types that are standardized within the following taxonomy, ranked in order of severity whereby cases are coded according to the most severe encounter type:

Data on Officers and Case Outcomes:

The Mapping Police Violence database includes information on the officers involved in each case (both the officers who killed the person and any other officers on the scene during the use of deadly force), including the names and race of the officers, any prior deadly force incidents involving that officer that have been reported by the media, and whether the case resulted in any administrative discipline, civil suits and/or misconduct settlements. This information is among the most difficult to obtain, especially in states that restrict or prohibit information about police misconduct or discipline from being made public. As such, we expect this information to provide important, albeit incomplete, insights into the officers who engage in fatal police violence. We will continue to add more data on these topics over time as more info are reported by officials and the media.

Additionally, Mapping Police Violence tracks cases where officers have been charged with a crime related to an incident of fatal police violence. These data are collected initially from ongoing monitoring of media reports related to each case in our database. We also worked with Professor Phil Stinson at the Henry A. Wallace Police Crime Database, the most complete database of officers who’ve been charged with crimes nationwide, to ensure that all known cases where officers were charged with a crime related to the use of deadly force since 2013 in their database are also included in the Mapping Police Violence database. Data on case outcomes and other details about each case are collected and updated on an ongoing basis, with the date of latest update noted in the Official disposition of death column of our database. This data includes what charges officers faced, whether any officers were convicted of a crime, and the sentence imposed, where such information are available.


[1] In 2019, the U.S. The Department of Justice (DoJ) launched the National Use-Of-Force Data Collection. However, one investigation found that the FBI was falling short in data collection efforts. More recently, a recent announcement from the DoJ  indicated that the program may be shut down given the lack of participation from law enforcement agencies, requiring a 60% participation threshold to release data.  

[2] It is also worth acknowledging that The Guardian created a comprehensive database called “The Counted” but only examines a single year of data.