MAY 2021
STRONGERCOMMUNITIES.STRONGERAMERICA.

Special Advance Fact Sheet:

Deaths of People of Color
By Law Enforcement Are
Severely Under-Counted
Overview and Background
Beginning in early 2021, the Raza Database Project,1 a team of volunteer researchers,
journalists, family members of Latinos killed by police, and activists came together to
investigate a long-suspected undercount of the deaths of Latinos2 and other people
of color by or in the custody of law enforcement. The Project’s Director, Roberto
“Dr. Cintli” Rodríguez, himself a survivor of police abuse, began his inquiry into the
subject in 2016 by comparing well-known Hispanic surnames with the names of
individuals reported in the “White,” “Other,” and “Unknown” categories of national
databases of police killings that were created following the shooting of Michael
Brown in 2014. His initial inquiry concluded that deaths of Latino and Indigenous
people at the hands of police were under-counted in widely reported national
databases by a quarter to one-third. He also called attention to media narratives
that virtually ignored the killings of Latinos by law enforcement, even in Southern
California, the largest Hispanic media market in the country.3
UnidosUS has been working with Raza Database Project researchers because, as an
advocate for proposed reforms such as the George Floyd Justice in Policing Act,4 the
organization believes that effective policies to address law enforcement abuse rest
on accurate statistics and media coverage. In the absence of an official, standardized,
and transparent federal database system to collect, analyze, and publish data on
people killed by law enforcement, a number of private entities have stepped in to try
to fill the void.5 However, given limited resources, each of these data sets (as well as
comparisons between them) has limitations:
•

The kinds of incidents covered vary; some are limited only to police shootings,
while others cover all deaths while in the custody of law enforcement.

•

Some databases only cover specific years.

•

Hispanic/Latino, Indigenous, and Asian and Pacific Islander people are not counted
in a uniform way.

•

None of the databases attempts to distinguish between justifiable vs. unjustifiable
deaths, which is supposed to be the function of the judicial system; although with
very few exceptions, most deaths in police custody never become the subject of
judicial proceedings.

www.unidosus.org

PAGE

1

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

UNIIDSUS

•

The racial/ethnic identity of some unknown number of people dying in police
custody, as well as the circumstances of these deaths, falls into the category of
“information withheld by police,” often for lengthy and indeterminate periods.

•

Perhaps most importantly, race/ethnicity typically is first observed and recorded
by law enforcement; in the absence of standardized definitions, many Latinos and
other people of color appear to be counted in “White” or “Other” categories, and
discerning race/ethnicity often requires additional research, including reviewing
news media, social media, or direct reports by victims’ families.

Civil rights advocates understandably focus on disproportionate killings by law
enforcement of racial/ethnic minorities. But the data also reveal that all Americans,
including Whites, are at risk of being killed by police at rates Project researchers
believe no other developed country in the world even approximates.

Analysis
To address some of these limitations and produce a more accurate count of Latinos
killed by or who died in police custody, Project researchers conducted several analyses.
The first aggregated all names reported in national databases of all deaths by or in
the custody of police from the year 2000 through May 9, 2021, eliminating duplicate
names based on standard social science practices.6 The result of that analysis is in
Appendix I. Table 1, below, reports results from 2014 through early May 2021.
Table 1: Aggregated, Unadjusted Count
Deaths in Presence of/Killings by Police
By Race/Ethnicity 2014–May 9, 2021
Year

Total

Latino/
Hispanic

Asian/Pac
Islander

Black/African
American

Native
American

Middle
Eastern7

2014

1,869

322

31

492

20

2015

2,149

309

30

550

2016

1,998

268

27

2017

2,075

314

2018

2,108

2019

White

Other

2

854

148

-

26

7

1,077

144

6

511

29

3

989

163

8

20

521

33

6

930

234

17

325

35

504

29

1

883

307

24

2,073

276

40

536

19

4

819

357

22

2020

2,134

267

23

507

17

4

766

529

21

2021

679

58

11

145

5

-

218

213

29

2,139

217

3,766

178

27

6,536

2,095

127

All

15,085

Unknown

The aggregated analysis reveals several key findings:
•

There are many more people killed by or while in custody of police than even
well-informed observers may realize. For example, followers of the Washington
Post’s highly regarded Fatal Force Report website (which is limited to police
shootings) were advised in 2020 that 967 people were shot and killed by the
police.8 The Project’s aggregated analysis shows that more than twice that
number—2,134—were killed by or died in the custody of police during that year.

PAGE

2

www.unidosus.org

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

UNlll:6 US

•

Similarly, the number of Latinos killed by or who died in police custody nearly
doubles when fully accounted for. While the Washington Post reports that 1,058
Latinos were shot and killed by police over the 2015–2020 period, that number
nearly doubles to 1,759 when all databases and all causes of death in police
custody are included.

The team also aggregated and analyzed the causes of death reported in national
databases. While shootings are by far the most common cause of death for all people
in police custody, the data also show large numbers of people who died due to pepper
spray, physical restraint (as a jury found killed George Floyd), tasering, in or by a
vehicle, and a suspiciously high number of people who purportedly experienced a
“medical emergency” while in custody (see Appendix III).
Raza Database Project researchers conducted additional analyses to identify Latinos
whose deaths may have been recorded in the “White,” “Other,” or “Unknown”
categories in national databases. Their initial analysis, reported in Appendix II, ran the
names of all people reported killed by or in the custody of law enforcement from 2000
through May 9, 2021, against the 2010 Decennial Census of Population and Housing:
Surname datasets.9 The results of that additional analysis for 2014 through May 9, 2021,
are reported in Table 2, below.
Table 2: Aggregated Adjusted Count
Deaths in Presence of/Killings by Police
By Race/Ethnicity, 2014–May 9, 2021
Year

Total

Latino/
Hispanic

Asian/Pac
Islander

Black/African
American

Native
American

Middle
Eastern10

White

Other

Unknown

2014

1,869

343

194

431

25

1

760

113

-

2015

2,149

346

170

491

39

1

992

104

6

2016

1,998

311

182

438

41

2

905

112

7

2017

2,075

388

187

456

45

3

837

148

11

2018

2,108

411

181

434

52

1

784

222

22

2019

2,073

342

233

465

33

2

729

248

21

2020

2,134

406

195

456

28

3

675

353

18

2021

679

106

85

134

9

-

189

133

23

All

15,085

2,653

1,427

3,305

272

13

5,871

1,433

108

Notwithstanding the limitations of surnames as a proxy for race/ethnicity,11 the
analysis produces significant findings. After aggregating across all databases and
adjusting for surname:
•

The number of Latinos killed by or in the custody of police over the 2014–present
period increased from 2,139 to 2,653, suggesting that such deaths are underreported in commonly cited databases by about one-quarter.

•

Deaths of Asian/Pacific Islanders increased by 1,210, or more than six times, in the
adjusted analysis.

www.unidosus.org

PAGE

3

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

UNIIDSUS

•

Reported deaths of Native American people grew to 272 from 178 in the adjusted
analysis, an increase of more than 50%.

•

The largest arithmetic change came from a reduction in the “White” category from
6,536 to 5,871, a decrease of 665 deaths.

•

The “Other” category also decreased significantly, by 662, a reduction of nearly
one-third.12

This analysis permits several other conclusions. First, people of color, who together
constitute less than 40% of the U.S. population, comprise more than 60% of all
people killed by or who died in the custody of the police. By comparison, Whites, who
constitute more than 60% of the population, comprise less than 40% of all deaths
over the 2014-2021 period. Second, even the adjusted numbers likely understate
the disparity, since significant numbers of Latinos, and likely many Asians as well,
have “European” surnames not captured by the surname match. Both communities
experience relatively high rates of intermarriage, the majority of which, especially
between Asians and Whites, involve women of color. Their children may be identifiably
Asian, Latino, or Indigenous, but those who have European surnames would not be
identified by this analysis. Finally, in the absence of an official government database,
additional research is required to produce a more accurate identification of Latinos
and others not identified by existing datasets.

About UnidosUS
UnidosUS, previously known as NCLR (National Council of La Raza), is the nation’s
largest Hispanic civil rights and advocacy organization. Through its unique
combination of expert research, advocacy, programs, and an Affiliate Network of
nearly 300 community-based organizations across the United States and Puerto Rico,
UnidosUS simultaneously challenges the social, economic, and political barriers that
affect Latinos at the national and local levels.
For more than 50 years, UnidosUS has united communities and different groups
seeking common round through collaboration, and that share a desire to make our
country stronger. For more information on UnidosUS, visit www.unidosus.org, or follow
us on Facebook, Instagram, and Twitter.

PAGE

4

www.unidosus.org

UNlll:6 US

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

Endnotes
1

The Raza Database Project is a network of some 50 researchers, scholars, journalists, activists, and family
members of victims killed by police. Roberto “Dr. Cintli” Rodríguez, an author and former Associate
Professor at the University of Arizona, is Project Director. Ivette Xochiyotl, a community-based advocate
of Mental Health and Civil and Human Rights issues, serves as Project Manager. Richard Casillas, Joanna
Laguna, Wendy Lujan, Maria Gomez Murphy, Alfred Porras, Dr. Paul Ruiz, and Valentina Zapata served as
volunteer researchers. The analyses reported here were conducted by former Census Bureau Statistician
Demographer Jesus M. García, CEO of La Cresta Demographics. These analyses represent initial findings
from the Project, which intends to delve more deeply into these and other data to identify with greater
precision and accuracy the actual number of Latinos killed by or in the custody of law enforcement.
Other planned/ongoing efforts include: studies of local media reports, social media, testimonials from
affected families, and other sources to address the misclassification of Latino deaths; carrying out
similar analyses of people killed by or in the custody of U.S. Border Patrol or Immigration and Customs
Enforcement; and deeper investigation of the reported causes of deaths by or in police custody. At
UnidosUS, Senior Director of the Racial Equity Initiative Viviana López Green, Civil Rights Policy Analyst
Claudia Ruiz, and Senior Cabinet Advisor Charles Kamasaki contributed to this Fact Sheet.

2

The terms “Hispanic” and “Latino” are used interchangeably by the U.S. Census Bureau and throughout
this document to refer to persons of Mexican, Puerto Rican, Cuban, Central and South American,
Dominican, Spanish, and other Hispanic descent. According to the technical definitions used by the
Census, Latinos may be of any race. This document uses the sociological construct of “race” whereby, at
least historically, most Latinos were treated as a distinct racial group, regardless of ethnicity. UnidosUS
also occasionally refers to this population as “Latinx” to represent the diversity of gender identities and
expressions that are present in the community

3

Roberto Rodríguez, “The Miscounted of 2016 and False Media Narrative of Police Abuse,” Diverse Issues
in Higher Education, January 10, 2017, https://diverseeducation.com/article/91022/.

4

https://www.congress.gov/bill/117th-congress/house-bill/1280.

5

Databases consulted by the Project include the following sources: Washington Post, “Fatal Force Report
Police Shootings Database, 2015-2021,” https://www.washingtonpost.com/graphics/investigations/
police-shootings-database/ (accessed May 20, 2021); The Guardian, “The Counted,” 2016, https://www.
theguardian.com/us-news/series/counted-us-police-killings (accessed May 20, 2021); Fatal Encounters,
https://fatalencounters.org/; Killed By Police, “Police Shooting and Crime Database,” https://robarguns.
com/crime-and-police-shootings (accessed May 20, 2021); Ana Srikanth, “National Database on Police
Killings Tracked 1,127 Deaths Last Year,” The Hill, March 17, 2021, https://thehill.com/changing-america/
respect/equality/543712-national-database-on-police-killings-tracked-1127-police (accessed May 20,
2021); Mapping Police Violence, https://mappingpoliceviolence.org/; Terry J. Wilson II, “Police Killings:
United States of America 1/01/2000-5/26/2020,” https://services3.arcgis.com/ZOm3lyRvRxR6535t/
arcgis/rest/services/FATAL_ENCOUNTERS_DOT_ORG_SPREADSHEET_Form_Responses/FeatureServer
(accessed May 20, 2021); Whitney Kotlewski, “People for the People Official Election Toolkit,” https://
services9.arcgis.com/q5uyFfTZo3LFL04P/arcgis/rest/services/Fatal_Police_Killings/FeatureServer/0
(accessed May 20, 2021); and “US Police Killings 2013 to 2019,” https://services9.arcgis.com/
uPQyVUNozOtWbtrl/arcgis/rest/services/US_Police_Killings_2013_to_2019/FeatureServer (accessed
May 20, 2021). Notwithstanding their limitations, the Raza Database Project and UnidosUS commend
these entities for their work to help Americans gain a better understanding of the scope and severity of
police killings/deaths in the custody of law enforcement in the United States.

6

A small number of duplicates likely remain, attributable to misspellings of names in one or more databases
or uses of different names for the same person. For example, the use of different names occasionally
occurs with some Asians for whom English translations may reverse first names and surnames, or with
some people of Hispanic descent who may be identified with maternal (i.e., appellido materno) in addition
to paternal surnames in a database. Other factors, such as inconsistencies in a victim’s age or date of death
among datasets, may result in the same person being counted more than once.

www.unidosus.org

PAGE

5

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

7
8
9

UNIIDi US

The “Middle Eastern” category is only rarely used or reported, and researchers believe these data are
unreliable; they are included herein to maintain fidelity with the design of the databases listed above.
See Washington Post, Police Shootings Database 2015-2021.
U.S. Census Bureau, Frequently Occurring Surnames in the 2010 Census, 2010 Decennial Census of
Population and Housing: Surnames, 2016, https://api.census.gov/data/2010/surname.html; https://
www.census.gov/topics/population/genealogy/data/2010_surnames.html. These datasets contain
over 150,000 spellings of surnames that occurred 100 times or more in the decennial Census records,
including surnames of people of Hispanic/Latino descent, non-Hispanic American Indian and Alaska
Native, non-Hispanic Asian and Native Hawaiian and Other Pacific Islander, and non-Hispanic Black or
African American. See also David L. Word and R. Colby Perkins, Jr., “Building a Spanish Surname List for
the 1990’s—A New Approach to an Old Problem,” U.S. Census Bureau, March 1996, https://www.census.
gov/content/dam/Census/library/working-papers/1996/demo/POP-twps0013.pdf; D. E. Knuth, SOUNDEX
Function, The Art of Computer Programming, Volume 3, Sorting and Searching (Reading, MA: AddisonWesley, 1973); U.S. Census Bureau, “Census API: Datasets in /data/2010/surname and its descendants,”
accessed May 25, 2021, https://api.census.gov/data/2010/surname.html; and Joshua Comenetz,
“Frequently Occurring Surnames in the 2010 Census.”

10 The “Middle Eastern” category is only rarely used or reported, and researchers believe these data are
unreliable; they are included herein to maintain fidelity with the design of the databases listed above.
11 The phenomenon of “surname clustering” makes the use of surname as a proxy for race/ethnicity
far more accurate than might be assumed at first glance; see U.S. Bureau of the Census, “What’s in a
Name?,” December 15, 2016, https://www.census.gov/newsroom/blogs/random-samplings/2016/12/
what_s_in_a_name.html. Nevertheless, there are many non-Hispanics, e.g., from the Philippines, who
have Spanish surnames, which may lead to an over-counting of Latinos based on surname matching
alone. However, there are also many people of Mexican origin with “Anglo” surnames; European surnames
occur frequently among people from Panama, Puerto Rico, and other areas with a long-standing U.S.
presence, and German and Italian surnames are especially common among people with origins in Chile
and Argentina. In the United States, most of these individuals would likely view themselves as Latino or
Hispanic. The researchers believe that, in sum, the latter outnumber the former, such that if anything,
the adjusted data in Table 2 and Appendix II likely still significantly under-reports the actual number of
Latinos killed by or who die in the custody of the police.
12 UnidosUS believes that the apparent drop in Black/African American deaths reported in Table 2 is
attributable to a technical and definitional issue, not to any over-reporting of Black deaths in existing
databases. Technically, the researchers used the standard Census definition of “Black, non-Hispanic” in
conducting the analyses. UnidosUS believes that the reduction of 461 Black deaths is entirely attributable
to “Afro-Latinos,” who legitimately should be included in both the Black/African American and Hispanic
Latino categories.

PAGE

6

www.unidosus.org

UNlll:6 US

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

Appendix I: Aggregated, Unadjusted Count*
Deaths in Presence of/Killings by Police
By Race/Ethnicity 2000–May 9, 2021
RACE/ETHNICITY
YEAR

TOTAL

Latino

NLAsian/Pac
Isld

NLBlack

NLMiddle
East

NLNative
Amer

NLWhite

OtherUnknown

Unknown

-

-

-

-

-

-

-

-

All

32,542

4,333

480

7,425

51

326

11,571

8,229

127

2000

859

56

12

122

–

13

125

531

–

2001

942

76

15

95

3

3

127

623

–

2002

1,011

97

12

133

1

6

147

615

–

2003

1,128

123

18

213

–

12

239

523

–

2004

1,105

146

17

193

–

6

265

478

–

2005

1,211

150

21

234

5

13

327

461

–

2006

1,318

167

19

271

1

13

358

489

–

2007

1,292

169

13

272

2

9

368

459

–

2008

1,236

165

15

262

3

9

334

448

–

2009

1,285

176

23

276

1

9

387

413

–

2010

1,302

155

19

301

4

15

461

347

–

2011

1,424

193

22

365

–

10

522

312

–

2012

1,477

215

23

420

1

17

565

236

–

2013

1,867

306

34

502

3

13

810

199

–

2014

1,869

322

31

492

2

20

854

148

–

2015

2,149

309

30

550

7

26

1,077

144

6

2016

1,998

268

27

511

3

29

989

163

8

2017

2,075

314

20

521

6

33

930

234

17

2018

2,108

325

35

504

1

29

883

307

24

2019

2,073

276

40

536

4

19

819

357

22

2020

2,134

267

23

507

4

17

766

529

21

2021

679

58

11

145

–

5

218

213

29

* NL= Non-Latino

www.unidosus.org

PAGE

7

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

UNIIDSUS

Appendix II: Aggregated Adjusted Count*
Deaths in Presence of/Killings by Police
By Race/Ethnicity, 2000–May 9, 2021
RACE/ETHNICITY
YEAR

TOTAL

Latino

NL-Asian/
Pac Isld

NLBlack

NLMiddle
East

NLNative
Amer

NLOther

NLWhite

OtherUnknown

Unknown

-

-

-

-

-

-

-

-

-

All

32,542

5,671

3,057

6,620

27

528

5

10,384

6,142

108

2000

859

117

79

121

–

22

–

111

409

–

2001

942

144

98

83

2

14

–

110

491

–

2002

1,011

178

96

121

1

11

–

130

474

–

2003

1,128

186

91

209

–

17

–

218

407

–

2004

1,105

216

109

175

–

13

–

240

352

–

2005

1,211

218

104

217

3

19

1

297

352

–

2006

1,318

242

120

241

–

20

–

328

367

–

2007

1,292

220

117

244

1

23

–

336

351

–

2008

1,236

216

114

241

2

15

–

300

348

–

2009

1,285

229

137

239

1

13

–

350

316

–

2010

1,302

219

110

278

3

26

1

411

254

–

2011

1,424

243

131

319

–

16

–

466

249

–

2012

1,477

265

144

374

–

26

–

488

180

–

2013

1,867

325

180

453

1

21

–

728

159

–

2014

1,869

343

194

431

1

25

2

760

113

–

2015

2,149

346

170

491

1

39

–

992

104

6

2016

1,998

311

182

438

2

41

–

905

112

7

2017

2,075

388

187

456

3

45

–

837

148

11

2018

2,108

411

181

434

1

52

1

784

222

22

2019

2,073

342

233

465

2

33

–

729

248

21

2020

2,134

406

195

456

3

28

–

675

353

18

2021

679

106

85

134

–

9

–

189

133

23

* NL= Non-Latino

PAGE

8

www.unidosus.org

UNlll:6 US

DEATHS OF PEOPLE OF COLOR BY LAW ENFORCEMENT ARE SEVERELY UNDER-COUNTED

Appendix III: Aggregated Count*
Causes of Death Reported in National Databases
By Race/Ethnicity, 2014–May 9, 2021
All
RACE/ETHNICITY
YEAR

All

TOTAL

32,542

Latino

NLAsian/
Pac Isld

NLBlack

NLMiddle
East

-

-

-

-

-

-

-

-

-

5,671

3,057

6,620

27

528

5

10,384

6,142

108

NLNLNative
Other
Amer

NLWhite

OtherUnknown
Unknown

CAUSE OF DEATH
Beaten

194

46

16

52

1

3

–

44

32

–

Burned/
Smoke
Inhalation

42

5

6

3

–

–

–

13

15

–

Drowned

203

32

15

46

–

4

–

47

59

–

69

18

3

9

–

1

–

20

18

–

23,664

4,146

2,209

4,725

22

415

1

8,065

3,973

108

Medical
Emergency

576

101

51

170

–

7

–

148

99

–

Other/
Undetermined

167

39

14

47

–

6

–

39

22

–

Pepper Spray

40

5

4

12

–

–

–

7

12

–

Physical
Restraint

325

61

27

99

1

2

–

82

53

–

Stabbed

50

17

3

4

–

–

–

12

14

–

1,008

174

75

304

–

14

–

279

162

–

4

1

–

–

–

–

–

1

2

–

6,200

1,026

634

1,149

3

76

4

1,627

1,681

–

Fell from a
height
Gunshot

Taser

Unknown

Vehicle

* NL= Non-Latino

www.unidosus.org

PAGE

9