Race & Policing in America

Race and Policing in America is a research project analyzing the patterns and tensions between communities of color and police agencies across the United States. Race and Policing examines how people of color and entire communities are affected by over-policing, criminalization and excessive use of force.

The goal of this project is to examine the proliferation of fatal encounters with police and better understand the geographic distribution and community characteristics of neighborhoods where these events are prominent. Using publicly available data we examine trends, run statistical tests and perform spatial analysis to determine clustering and common characteristics among victims, police departments and neighborhoods.

By examining the geospatial distribution of these incidents as well as the neighborhood context we hope to learn more about how to combat these issues and use our findings to make impactful policy changes and create safer communities.

Our research to date includes scraping, cleaning and comparing the five most prominent datasets tracking civilian deaths at the hands of law enforcement between the years 2015-2020. After rectifying the data, we perform spatial analysis to find the cities, counties and law enforcement agencies that account for the greatest number of fatalities within the given timeframe. Once the top ten cities, counties and law enforcement agencies are identified we then zoom into these geographic areas to learn more about the neighborhood context, the police force’s history in that region, and the socio-economic and demographic statistics of that area. By incorporating United States Decennial Census data and American Community Survey 5 year estimate data we are able to learn more specific information about the neighborhoods and populations affected by this type of violence. Using Census level data allows us to incorporate very fine level information that helps run regressions, perform spatial analysis, and identify clustering and common characteristics across neighborhoods and cities.

The biggest challenge so far has been wrangling the data and using machine learning to identify clustering, de-duplicate repetitive entries and make sense of anomalies in the data. The research support provided by Stanford University can greatly improve our computational capabilities and help us think through some of the major data issues that we have been trying to solve over the past few years. When analyzing and cleaning these disparate datasets it can be very difficult to standardize and harness the power of the data to create the best statistical models. The Race & Policing project will benefit greatly from leveraging the tools, technologies and ideas provided by Digital Civil Society Lab and the Stanford Institute of Human-Centered Artificial Intelligence. In our data analysis and statistical modeling, our team of statistical models, our team has run into a variety of computational and methodological issues. Working alongside researchers within the Stanford community will help us overcome some of these hurdles by helping us build more robust and statistically sound models when analyzing patterns across the United States. 

In addition to collecting data on fatal encounters, our team has managed to scrape Twitter data, videos and images of conflicts between police and civilians during the past year of demonstrations and protests. This data is considered secondary to the primary research of fatal encounters, but is very much related to our overall emphasis on accountability for officers using excessive force. Our team is continuing to gather data related to this years' protests, clashes with police to better understand the relationship between fatal police encounters, use of force and civil unrest (see the app in progress:  www.blm-map.com ). A collaboration with Stanford researchers will allow us to harness the full capabilities of our datasets and, more importantly, ensure that all of our models and analyses are properly tested and vetted before releasing information to the public. Overall, this collaboration with Stanford’s expert researchers and Artificial Intelligence technologies will provide new perspectives and approaches to the research that will help us move towards a more innovative and impactful model for our research. 

This project is a massive undertaking because we are not only looking into two decades of fatal police encounters but are also researching information about specific Census variables for each of those years. Furthermore, we are hoping to examine additional information in the future regarding police department personnel, use of force settlements and policy initiatives. Therefore, I believe the success of this project should be measured on an ongoing basis. For us, one measure of success will be collating all of the available datasets to create the most comprehensive database of fatal encounters over the past two decades. Another measure of success will be when our team is able to visualize this data in a coherent manner to our stakeholders and community partners. This research will be part of a lifelong commitment to support communities of color and speak out against these injustices. If we are able to advance this research by standing up new datasets, creating visualizations, and providing quantitative analysis that can be leveraged for political change and action in even one community or one victim's case, then this project will have been successful.

Within the goal of attending to these issues, the data and methodology of calculating such incidents do vary amongst the three organizations. Therefore, there are three different totals among the datasets.

The differences are further substantiated when we examine the methodology - or, way of collecting and organizing data - of the three organizations.

Fatal Encounters - 6,958 Total Recorded Incidents

"Our paid researchers have several methods of getting information into the verification queue. First we aggregate data from other large sets like KilledByPolice or the Los Angeles Times’  The Homicide Report  and individuals like Carla DeCeros who have contributed their data to FE. They then research the missing information and double-check the information that’s included. When the record is complete, it’s moved over to the verification queue, where it is again checked against published sources yet again by the Principal Investigator of FE.

When an incident is reported by a volunteer—the crowd—every fact presented is compared to published media reports or public records to verify its accuracy. This information from any source–a hometown newspaper, for example–and submitted it through our form. Once submitted, it goes to a separate spreadsheet, where we verify its information against media sources.

We have also been conducting research by state and by date. These methods are intended to be redundant so that we catch as many incidents as possible. However, we know from experience that incidents have been missed, sometimes because the death was not reported at the time it happened, through human error, or just because of the vagaries of the internet. To address this issue, FE and our sister project, EncuentrosMortales.org, have made more than 2,300 public records requests of state, federal and local law enforcement agencies. This part of the process is extremely expensive, but the documents are useful as yet another level of redundancy. Other researchers, such as Lance Farman have also been testing the completeness of the database against FOIA requests and have found that this method yields a 97% completeness rate for 11 of the states that have been logged so far."

(Source -  Methodology - Fatal Encounters )

Fatal Force - 5,934 Total Recorded Incidents

"The Post is documenting only those shootings in which a police officer, in the line of duty, shoots and kills a civilian — the circumstances that most closely parallel the 2014 killing of Michael Brown in Ferguson, Mo., which began the protest movement culminating in Black Lives Matter and an increased focus on police accountability nationwide. The Post is not tracking deaths of people in police custody, fatal shootings by off-duty officers or non-shooting deaths.

The FBI and the Centers for Disease Control and Prevention log fatal shootings by police, but officials acknowledge that their data is incomplete. In 2015, The Post documented more than twice as many fatal shootings by police as had been recorded by the FBI. Last year, the FBI announced plans to overhaul how it tracks fatal police encounters.

The Post’s database is updated regularly as fatal shootings are reported and as facts emerge about individual cases."

(Source -  How The Washington Post is examining police shootings in the United States )

Mapping Police Violence - 6,631 Total Recorded Incidents

"This information has been meticulously sourced from the  three largest, most comprehensive and impartial crowdsourced databases  on police killings in the country:  FatalEncounters.org , the  U.S. Police Shootings Database  and  KilledbyPolice.net . 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 campaign to end police violence in our communities."

(Source -  About the Data - Mapping Police Violence )

{Call to action}

These interactive maps display fatal force used by police when encountering civilians between 2015-2020. Our project combines publically available datasets on police violence with socioeconomic data and spatially reinterprets it to understand the contexts where that violence occurs. These preliminary interactive maps aggregates data at the state, metro area, and county levels.

We also have a version of the Dashboard that allows you to interact with all three versions of the data. Check it out here: