Race and Extreme Police-Citizen Violence Report

    RAC^ AND KX'l'M'F’MF 

POLICE-CITIZEN 

VIOLENCE

Presented at the Anrmal Meeting 

of the
Academy of Criminal Justice Sciences 

Cincinnati, Ohio 

March, 1979

James J. Fyfe

The American University 
The Police Foundation



-1-

The disproportionate representation of blacks among the clientele

of the criminal justice system is a recurrent theme in the literature

(Wolfgang, 1964:51). Research into police firearms use (e.g., Clark,

1974; Milton, et al, 1977) reports that minority disproportion is also

an explosive issue of these encounters. Goldkamp (1976:183), however,

accurately notes that the present paucity of analysis of this phenomenon

leaves criminal justice agencies free to adopt either of two empirically

unsubstantiated "belief perspectives."

Goldkamp briefly defines those "belief perspectives" as follows:

Some writers suggest that the disproportionately high death 
rate of minorities at the hands of the police can best be 
explained by the disproportionately high arrest rate for crimes 
of violence, or by assumptions concerning the suspects’ re­
sponsibility for his/her own death in violent police-suspect 
interactions. Others see disproportionate minority deaths as 
resulting from both irresponsible use of deadly force by a 
small minority of police officers and differential adminis­
tration of law enforcement toward minority citizenry (which 
in effect produces disproportionately high arrest and death 
rates for minorities in general). Kobler and Knoohuizen,
Fahey and Palmer stress the possibility that police miscon­
duct may play a considerable role in generating civilian deaths. 
Takagi ascribes disproportionality to the simple fact that 
"police have, one trigger finger for whites and another for 
blacks." (1976:169)

Harding and Fahey, whose position seems to straddle both perspec­

tives, suggest that police misconduct is involved in shooting deaths. 
Conversely, they also state that the use of deadly force by police "is



-2-

not an independent aspect of the race problem." They write that:

Police conduct is a dependent aspect of general patterns 
of criminal behavior, patterns that are significantly in­
fluenced by broader considerations of, for example, age, 
class, and affluence. (The) (v)ictimization of those shot 
is directly related to contacts of the sort in which fire­
arms are most frequently used by criminals. (1973:310)

Race and Extreme Police-Citizen Violence: Two Hypotheses

Harding and Fahey suggest, therefore, that racial disproportionality

among police shooting victims may be"related"to raciai variations among 

other indices of violence. To some degree, this assertion is supported 

by the work of Kania and Mackey (1977), who found that such variations 

among fatal police shooting rates across the fifty states were closely 

related to variations in reported violent crime and criminal homicide 

rates. Because of our access to data on extreme police-citizen violence 
in New York City, these prior efforts also suggested two hypotheses 

which became the focus of this paper.
First, on the theory that police shootings are a corollary of the 

frequency of contacts which present the opportunity for such violence 

we postulated that:
HI: Blac3cs and Hispanics would be overrepresented among

police shooting opponents in relation to their repre­
sentation in the New York City population but there 
would be less disporportion by race taking into account 
the racial representation of arrests for violent crime.

Second, to test the assumption that police shootings are related



-3-

to other indices of extreme violence among the races:

H2: Blacks and Hispanic overrepres«ntation among police
shooting opponents in New York City would be reduced 
by taking into accotmt the racial representation in 
reported murders and non-negligent manslaughters.

Data Sources

Our major data source for these analyses consisted of New York City 

Police Department records of all incidents in which officers reported 

discharging weapons and/or being subject of "serious assault" (e.g., 

assault with deadly weapon and/or which resulted in officer death or 

serious injury) during the years 1971-1975. These data included 

"Pirearms Discharge/Assault Reports" (FDAR's) filed by A904 officers, 

of whom 3827 reported discharging firearms in 2926 separate "shooting 

incidents." Since not all these involved shooting at other human beings, 

we excluded from analysis such events as shootings to destroy animals, 

warning shots and officer suicides. Because of the relatively low num­

ber of female opponents included in our data and because they were often 

involved in non-line of duty shootings (Fyfe, 1978:145-159), we also 

excluded them from analysis. Conversely, because our detailed examina­

tion of the data had convinced us that the frequency of police use of 

firearms as a means of deadly force is best measured in terms of officer 

decisions to point and fire at other human beings, we included for 

analysis all such incidents, without regard to their consequences: with

rare exception, missed shots, woundings, and fatalities are only chance



-4-

variations of equally grave decisions.

While we considered this shooting data base nearly ideal for our 

purposes, we found that available United States Census Bureau figures 

were less informative. New York City base population figures provided 

by the census define only two major racial groups, "White," which includes 

Mexican, Puerto Rican, or a response suggesting Indo-European stock"
(U.S. Dept, of Commerce, 1973:B34, App7-8), and "Negro." This inclu­

sion of Puerto Ricans and other Hispanics into the "White" category is 

not a major limitation insofar as measurement of shooting opponent racial 

disproportion is concerned, however, since New York City also develops 
its own population racial distributions.

More surprisingly, we also found— with one exception— that the New

York City Police Department does not compile racial statistics of arres- 
2tees. In the absence of this ideal data source, we decided to employ 

two surrogate scales of comparison. For our analysis of HI, we elected 
to use Burnham's sample of the races of 700 persons arrested for murder, 

non-negligent manslaughter, robbery, forcible rape and felonious assault
3in New York City (1973:50). Because we were similarly precluded from 

using homicide arrestee data in our examination of E2, we decided to 

employ information on the race of homicide victims. Here, since many 

studies show that homicides tend to be intra-racial crimes, it might not 
be unreasonable to consider the victim index as a crude proxy for the
perpetration index.



RACIAL DISTRIBUl'lON OF NEW YORK CITY POLICE SHOOTING 
OPPONENTS, AND NEW YORK CITY POPULATION 
JANUARY 1, 1971-DECZMBER 31, 1975

-5-

TABLE 1

Shooting
Opponents

New York City 
Population

White 17.5Z (549) 64.12 (5076022)
Black 60.22 (1889) 20.52 (1621583)
Hispanic 22.32 (701) 15.42 (1216557)

TOTALS 100.02 (3139) 100.02 (7914162)

chi-square * 1488.AO 
p * .001
V « .49

iixcludes "Other" racial categories because of low statistical signi­
ficance. Ten shooting opponents were identified as members of "Other" 
racial groups.

Calculated from: New York City Police Department, Chief of Field
Services, Summary of Precinct Populations. 1973.



—6—

HI: Analysis

Given these limitations, our data do indicate that Blacks and 

Hispanics are disporportxmately represented among those New York City 

police shooting opponents whose race is recorded. Table 1 demonstrates 

that Whites, who comprise 64.1 percent of New York City’s population total, 

represent only 17.5 percent of its shooting opponents. Blacks, converse­

ly, are overrepresented among shooting opponents in almost exactly the 

reverse ratio: 60.2 percent of shooting opponents and 20.5 percent of

total City population are Black. Hispanics, who number 15.4 percent of 

City population,- represent 22.3 percent of shooting opponents.

Using the racial composition of the general population of the City 

as expected frequencies, we derived a chi-square for these distributions 

which is significant at the .001 level. Similarly, our obtained 
Cramer’s v (.49) suggests a rather high association between race and 

the likelihood of being shot at by New York City police. The first part 
of this hypothesis is therefore confirmed.

Turning now to a model (Table 2) which utilized the ethnic distri­

bution of felony arrests for violent crimes (Burnham’s data) to generate 

expected numbers of shooting opponents by ethnicity, it may be seen that 

there is a fairly close fit. Although the chi-square is still signifi­

cant, the V value is now only.09, or much smaller than the v value of 

.49 reported for Table 1. While Whites remain slightly underrepresented 

among shooting opponents. Blacks are also underrepresented. Only



-7-

RACIAL DISTRIBUTION OF NEW YORK CITY POLICE 
SHOOTING OPPONENTS, AND PERSONS ARRESTED 

FOR FELONIES AGAINST THE PERSON 
JANUARY 1. 1971 - DECEMBER 31, 1975

TABLE 2

Shooting
Opponents

Felony A a- Arrests

White 17.5Z 22.2%
Black 60.2Z 62.4%
Hispanic 22.3Z 15.4%

TOTALS 100.0% 100.0%

chi-square “ 12.82
p “ .01
V - .09

Calculated from a sample of 700 persons arrested for murder, non- 
negligent manslaughter, robbery, felonious assault and forcible rape 
in New York City, 1971. Source: David Burnham, "3 of 5 Slain by
Police Here are Black, Same as Arrest Rate," The New York Times. 
August 26, 1973, 50.



-8-

Hispanics are overrepresented. Some caution must be observed about the 

finer distinctions in view of the limitations of data on ethnic classi­
fications .

H2; Analysis

Using our murder and non-negligent manslaughter victim racial dis­
tributions to generate expected frequencies of shooting opponents by 
race (Table 3), we find chat there is a fairly close fit. Again Cramer’s 
V is only .09, which suggests that there is a close parallel between the 

racial distributions of homicide victims and police shooting opponents.

Further Comments

Even if both the relationships shown in Tables 2 and 3 were demon­
strated by more comprehensive arrest and victim data, they would not 

prove that the disproportionate involvement of Blacks and Hispanics in 

shooting incidents is related to their disproprotionate involvement as 
violent crime arrestees and homicide victims.

Most specifically, we would still be left with the possibility that 
both relationships are spurious and merely reflections of varying degrees 
of risk due to differential age distributions and/or differential en­

forcement and police deployment practices. It is still possible that 

Che races are differentially represented among shooting opponents because 

more Blacks and Hispanics fall into the age groups most frequently



RACIAL DISTR.ISUTION OF NEW YORK CITY POLICE SHOOTING 
OPPONENTS, JANUARY 1, 1971-DECEMBER 31, 1975 AND 

VICTIMS OF MURDER AND NON-NEGLIGENT MANSLAUGHTER,^ 
JANUARY 1, 1973-DECEMBER 31, 1975

-9-
TABLZ 3

Shooting
Opponents

Homicide
Victims

White 17.5Z 22.5%
(5A9) (1069)

Black 60.2Z 51.0%
(1889) (2A19)

Hispanic 22.35; 26.5%
(701) (1259)

Totals 100.0% 100.0%
(3139) (A7A7)

chi-square =« 5A.68
p * .001
V - .09

^Source; New York City Police Department, Homicide Analysis Unit, 
Annual Report. 1976.



-10-

involved in these incidents. Alternatively, it is still possible that 

Blacks and Hispanics are disproportionately represented among shooting 

opponents because police do have "one trigger finger for Whites arirf 

another for minorities."

Table 4, which summarized the age characteristics of New York City’s 

male Whites, Blacks and Puerto Ricans (the city’s major Hispanic sub­

population) , confirms the existence of differential age distributions 

among these groups. As the table indicates. New York City’s white males

are generally considerably older (median age = 33.3 years) than either 

its male blacks (median age * 23.1 years) or its male Puerto Ricans 
(median age ■ 19.4 years).

To da:ermine ;rtiether similar age discrepancies existed among the 

shooting opponents included in our data, we crosstabulated opponent race 
and age. The results of this analysis are presented in Table 5, and 

demonstrate that the age distributions of shooting opponents vary little 
among the races, a finding that runs counter to those of Jenkins and 

Faison (1974) and Kobler (1975). Indeed, our obtained levels of p (.99) 

and V (.03) indicate that these three age distributions are so close as 
to be nearly indistinguishable.

The Table also suggests that confrontation with armed police is 

largely an activity of the young.^ More than half (1093) of the 2149 

opponents whose race and age are included in our data set are less than 

24 years old. Older opponents are not entirely excluded, however, since



- 11-  

TABLE 4

AGE CHARACTERISTICS OF NEW YORK CITY 
MALE POPULATION BY ETHNIC GROUP

Puerto
Black R1nan White

Median Male Age 23.1 19.4 33.3
Percentage of 
Male Population 
Under 18

42.0 45.6 27.6

Source: U. S. Department of Conmerce. Characteristics of the
Population, Part 34 NY Section 1, March 1973 U. S. 
Government Printing Office, pp. 34-108, 34-432.



TABF.E 5

NEW YORK CITY POLICE SHOOTING INCIDENT OPPONENT RACE 
BY AGE, JANUARY 1, 1971-DECEMBER 31, 1975

Opponent
•

Opponent Age

TotalsUnder
16 16,17 18,19 20,21 22,23 24,25 26,27 28,29 30f

White 3.2%
(13)

10.6%
(A3)

15.0)
(62)

14.5%
(59)

10.6%
(43)

11.3%
(46)

4.9%
(20)

7.2%
(29)

22.7%
(92)

18.9%
(407)

Black 3.6% 8.7%
(107)

12.8%
(185)

13.7%
(169)

12.5%
(155)

12.7%
(158)

7.5%
(93)

5.8%
(72)

22.7%
(280)

57.6%
(1236)

Hispanic 2.0%
(10)

11.1%
(56)

10.5%
(53)

13.1%
(66)

10.7%
(54)

11.1%
(56)

7.9%
(40)

5.9%
(30)

27.7%
(140)

23.5%
(505)

Totals 3.1%
(67)

9.6%
(206)

12.7%
(273)

13.7%
(294)

11.7%
(252)

12.1%
(260)

7.1%
(153)

6.1%
(131)

23.9%
(512)

100.0%
(2148)

I

c;

chi-square = 3.866 
p = .99 
V = .03



-13-

almost one-quarter (23.9 percent) of the group are thirty or more years 
old.®

Table 6 presents the frequency of shooting opponents per 10,000 

population aggregated by Census Bureau age and race subgroups. Examin­

ation of Table 6 reveals that White/Hispanic and Black age distributions 

are very similar. For both race groupings, those 20 to 24 years old 

are most frequently involved in shootings; subsequently the rates 

steadily decline for each age subgroup, reaching the lowest levels among 

those 50 years or moreTear^oldr-The-only major discrepancy in age 

trends occurs for the 15-19 year age group; the White/Hispanic male rate 

is only slightly less than that of the 20—24 year group, while, for

Blacks, the rate of 15-19 year olds is substantially lower than for 20- 
24 year olds.

Equally as striking as the general similarities in age patterns 
are the great numeric discrepancies in rates. Specifically, we find 
that Black males are six times' more likely to have been involved in 

police shooting incidents (24.20 per 10,000 population) than are male 

White/Hispanics (3.95 per 10,000). Indeed, the overall Black male rate 

(24.20) is more than twice as large as the highest White/Hispanic male 

rate (11.40). The general discrepancy ratio holds in each and every 
age group, and is strongest within the 20-29 year range.



table 6

NEW YORK CITY MALE SHOOTING OPPONENT RACE 
PER 10,000 POPULATION, BY AGE AND RACE,^ 

JANUARY 1, 19.71 - DECEMBER 31, 1975

-14-

Age White/Hispanic Black

15-19 10.04 38.82

20-24 11.40 67.38

25-29 7.62 43.06

30-34 6.02 21.92

35-39 3.88 15.02

40—44 1.77 9.24

45-49 .98 4.54

50+ .16 1.86

Totals 3.95 24.20

^Excludes opponents under fifteen years of age (White/Hispanic n“9; 
Black n*17).



-15-

Racial Disproportion

To this point, our investigation has shown that the variable of 

race is linked to the likelihood of being a police shooting opponentj 

in a similar fashion, this risk factor is apparently linked to arrest 

rates for violent felonies. The possibility remains that the great 

numeric disproportion of minorities among both arrestees and shooting 

opponents is a function of differential police enforcement or deployment 
practices. ‘

2̂ -̂ ce most prior literature which addresses racial disproportion 
among opponents (e.g., Goldkamp, 1976; Harding and Fahey, 1973; Robin, 

1963; Jenkins and Faison, 1974) examines only fatal shootings, we com­

menced our investigation with an analysis of incident consequences in 
terms of opponent injury. Table 7 provides a crosstabulation of reported 

shooting opponent race by injury, and demonstrates that these consequences

little among the races. Regardless of race, approximately three 

in five opponents suffer no injury, one in five is wounded, one in ten is 

killed and one in ten escapes after the police have shot at him with 

unknown effect. Once an officer decides to employ his "trigger finger," 

the race of his opponent apparently matters little in terms of the effect 

of police shots. Blacks escape with unknown injuries approximately twice 

as often (16.6 percent) as White (7.9 percent) or Hispanics (9.8 percent), 
but as our obtained chi-square (significant only at the .50 level) indi-



TABLE 7

NEW YORK CITY POLICE SHOOTING OPPONENT 
RACE BY INJURY

JANUARY 1, 1971-DECEMBER 31, 1975^

-16-

Opponent
Race

Opponent Injury

None Wounded Killed Unknown^ Totals

White 61.8% 21.0% 9.2% 7.9% 17.6%
(335) (114) (50) (43) (542)

Black 53.4% 21.0% 9.1% 16.6% 60.0%(983) (386) (168) (305) (1342)

Hispanic 56.0% 22.2% 12.1% 9.8% 22.3%
(384) (152) (83) (67) (686)

Totals 55.4% 21.2% 9.8% 13.5% 100.0%^^(1702) (652) (301) (415) (3072)

chi-square » 5.45 
p - .50 
V » .03

^Excludes cases in which opponent race not reported; excludes 4 suicides 
(1 White, 2 Black, 1 Hispanic).

^Not apprehended opponents at whom shots were fired with unknown effect. 
'Tercentage subcells may_not total 100.0 due to rounding.



-17-

cates, even this variance is likely to be a result of chance.

This lack of variance "within" confrontations, obviously, does not 

address numeric disproportionality. Stated most simply, we can observe 

little difference among the races once they become involved in conflint 

situations: we have not, however, touched on the issue of why so many

minority opponents become involved in these incidents in the first in­
stance.

To examine this question, we attempted to determine whether there 

existed significant variation among the t3rpes of shooting incidents in 

which different races became involved. We hypothesized that if we found 

high frequencies of incidents in which police shot at unarmed Blacks or 

Hispanics, it would suggest that police did, indeed, have "different 

trigger fingers" for minorities. Conversely, we felt that significant 

differences among the races in the precipitating event types and the 

degree of danger confronting police might help to explain the dispro­
portion of minority opponents.

To simplify this process and clarify its results, we decided to em­

ploy the race of each incident's "primary opponent".^ This resulted in 

very little loss of accuracy since incidents involving mxiltiple opponents
gare overwhelmingly intraracial events. Further, since our operative 

definition of "primary opponent" translated into either the only opponent 

or the one posing the greatest threat to police (e.g., the most combative, 

most heavily armed), we concluded that it was this person's conduct upon



-18-

which police reaction (or overreaction) would be principally based.

Thus, a shooting precipitated by a robbery involving one Black suspect 

armed with a gun and a White one armed with a knife becomes a "Black 
opponent" incident.

Our first measure of variance in shooting incident types among the 

races involved an investigation of the events which precipitated police 

shooting. As Table 8 and its chi—sqtaare significance level (.00001) 

reveal, there is considerable variance here. Perhaps most striking is 

the great frequency with which police confront Blacks at robberies. 

Indeed, nearly half (45.8 percent) of the incidents involving Blacks 

were reportedly initiated by robberies. This is a rate nearly twice that 

of Whites and Hispanics (23.4 and 26.3 percent, respectively) and repre­

sents a raw frequency (495 incidents) greater than the total of all 

incidents involving either Whites (354) or Hispanics (429). We see, in 

fact, that robberies involving Black primary opponents comprise 26.6 

percent of all the incidents included in this Table.

The Table also demonstrates that Whites are disproportionately more 

frequently counted among those who confront police at burglaries (12.4 

percent versus 6.4 percent and 9.1 percent for Blacks and Hispanics, 

respectively). In addition, perhaps because of the general relation­

ship between race and social status, the percentages of "Car Stop" 

incidents involving Blacks (11.3 percent) and Hispanics (8.4 percent) 
are far smaller than those of Whites (19.5 percent). Similarly, the



NEW YORK CITY rOLlCB SDOOTINC PRIMARY 
OPPONENT RACE BY PRECIPITATINC EVENT, 

JANUARY 1, 1971 - nECEMBEK 31, 1975

TAm.ii: B

Primary
Opponent

Race

Precipitating Event
Reupond

to
Olatur-
bunce

Burglary Rubbery
Attempt
other

Arreata
Handling
Prlaoner

Inveati-
gallng

SLiaplclout
Perauno

Anibuali Mentally
Deranged

Auto
Purault/
Stop

Aasault
on

Officer
Otlier TOTALS

Willie 10.2T 
(35)

12.«X
(«M

23.AX 
(83)

A.2X
(15) 0.8X

(3)
7.3X
(26) 0.3Z

(1)
2.5X
(9)

19.5X 
(69) 8.8X

(31)
10.5X 
(37) 19.OX 

(35A)
Black 9. OX 

(97) 6.AX 
(69)

A5.8X
(495)

3. IX
(3A)

1.6X
(17)

11.OX 
(119)

0.9X
(10)

1.5X
(16)

11.3X 
(122)

5.AX 
(58)

A.OX 
(43)

58. OX 
(1080)

lllapurilc 15.6X 
(67) 9.IX

(39) 26.3X 
(U3)

A.9X
(21) I.2X

(5)
16.6X 
(71)

0.9X
(A)

2.IX
(9)

a.AX 
(36)

S.6X
(2A)

9.3X
(40) 23. OX 

(429)
Tutuls 10. 7X 

(200) 8.2X
(152) 37.IX 

(691)
3.ex 
(70) 1.3X

(25) 11.6X 
(216) 0.8X

(15)
i.ex
(3A) 12.2X 

(227) 6.IX 
(113)

6.AX 
(120)

100.OX 
(1863)

U3I

nut aucurtalned •• 15 
chl-sqiiiire " 151.88078

p “ .00001
V - .20



-20-

percentage of "Other" incidents, which often include offduty disputes 

and the like, is greater for Whites (10.5 percent) and Hispanics (9.3 

percent) than for Blacks (4.0 percent). Conversely, Whites are less 

frequently involved in generally pro-active "Investigative Suspicious 

Person"— or "Stop and Question"— incidents than are Blacks and Hispanics 
(7.3 versus 11.0 and 16.6 percent, respectively). Finally, the Table 

reveals considerable difference in the frequencies of "Respond to 

Disburbance' incidents; 15.6 percent for Hispanics, 10.2 percent for 
Whites and 9.0 percent for Blacks.

A measure of the threat to officer safety at these incidents is 

provided by Table 9,,which crosstabulates primary opponent race with 

weapon type. Here again, it can be seen that striking differences exist 

among the races, with the chi-square proving significant to .0001. 

Approximately half of Black and Hispanic opponents (52.9 and 48.0 per­

cent) were armed with handguns. Nearly half (47.2 percent) of all inci­
dents included in the Table were police confrontations with Hispanic or 
Blacks armed with handguns, rifles, machine guns or shotguns; White 

handgun, rifle machine gun or shotgun incidents account for 6.9 percent 

incidents. Conversely, we find that Whites are more frequently 

involved in incidents involving no weapon or no assault on police (15.5 

percent) than are Blacks (7.8 percent) or Hispanics (5.1 percent). Whites 

are also overrepresented in incidents involving the use of vehicles 

(16.1 percent versus 6.1 percent for Blacks, 6.3 percent for Hispanics)



TADLB 9

HEH VnUK CITY fOUCE SIIOOTIHC TRIHARY Oi’rONEHT jRACB 

BY UEAFOH. JANUARY i, 1971-DECEHDER 31 , 1975

Primary Type of Weapon
- -  ̂- -■ .( '

Opponent
Riice Hone Handgun

Rifle/
Hoclilne
Gun

Shotgun
Knife/
'Cutting

Instrunent
Vehlc)

Physical
Force Other Totala

Ulilte
t

IS.SX
(56)

32.IX 
(U6)

1.4X
(5)

1.9X
(7)

13.3X
(48)

16.IX
(58)

9.4X
(34)

10.2X
(37)

19.2X
(361)

Black 7.8X
(85)

52.9X 
(574)

1.3X
(14)

6.3X
(68)

14.IX 
(153)

6.IX
(66)

4.2X
(46)

7.4X
(80)

57.8X
(1086)

Hispanic 5.IX 
(22)

48.OX 
(207)

2.6X
(11)

3.IX 
(14)

21.3X
(92)

6.3X
(27)

5.IX
(22)

8.4X
(36)

22.9X
(431)

Totals 8.7X
(163)

47.8X 
(897)

1.6X
(30)

4.7X
(89)

1S.6X
(293)

8.OX 
(151)

S.4X
(102)

8.IX 
(153)

10 0.ox“
(1878)

Iro

chi-square r 131.62032 
P - .0001 
V - .19

Subcell pcccentagea nay not total 100.0 due to rounding.



-22-

or physical force (9.4 percent versus 4.2 and 5.1 percent) as means of 

assaulting police. Hispanics use knives against police considerably 

more often (21.3 percent) than do Whites (13.3 percent) or Blacks (14.1 

percent). Given that most police are killed or seriously injured by guns 

or knife wounds, therefore, we would tentatively conclude that Blacks and 

Hispanics are more often involved— both proportionately and in terms of 
sheer numbers— in incidents that present greater potential danger to 
police than are Whites.

--- That—po-tentia±—danger does not necessarily translate into real

negative consequences in terms of officer injury is indicated by Table 

10, which provides a cross tabulation of FDAR incident primary opponent 

race by degree of officer injury (excluding non line of duty injuries, 

which are not relevant to this analysis). Here although the nature and 

seriousness of non fatal injuries are not specified and in fact, vary 

considerably, it can be seen that proportionally more officers are in­

jured in encounters with Whites (22.8 percent) or Hispanics (18.0 per­

cent) . Proportionally more officers are killed in the line of duty by

Blacks (1.2 percent) than by Whites (.8 percent) or by Hispanics (.7 
9percent). Although this Table’s chi-square indicates that the dif­

ferences among these distributions are not significant (p**.80), it is 

also important to note that 55 percent of line of duty officer injuries 
and deaths occur with Black opponents.



TABLE 10

-23-

NEW YORK CITY POLICE FIREARMS DISCHARGE/ASSAIJLT INCIDENT 
PRIMARY OPPONENT RACE BY OFFICER INJURY,^

JANUARY 1, 1971-DECEMBER 31, 1975

Primary
Opponent

Race

Officer Iniury

TotalsNone Injured Killed

White 76.4S
(402)

22.8%
(120)

.8%
(4)

17.5%
(526)

Black 82.2Z
(1471)

16.6%
(297)

1.2%
(21)

59.6%
(1789)

Hispanic 81.3%
(557)

18.0%
(123)

.7%
(5)

22.8%
(685)

Totals 81.0%
(2430)

18.0%
(540)

1.0%
(30)

100.0%^
(3000)

—

chi-square » 1.84
p « .80
V = .02

^Includes only officers woulded or killed in the line of duty.

Subcell percentages may not total 100.0 due to rounding.



-24-

Some Conclusions

In sunmiarizing this research within the contest of prior literature 
and the liniits of our data, we are led to two major conclusions. First, 

Harding and Fahey's assertion that minority disproportion among police 

shooting opponents is related to differential age distributions among 

the races is, in New York City at least, inaccurate. Our data demon­

strate that, while police shooting opponents are generally young and a 

greater proportion of the Black population is young. Black males in  ̂
age groups are considerably more liable to become police shooting oppon­

ents than are their White/Hispanic contemporaries.

Our second conclusion deals with whether that greater liability is
)

associated with greater Black participation in activities most likely to 

lead to justifiable extreme police-citizen violence or with "the simple 
fact that 'police have one trigger finger for whites and another for 

blacks.'" Here we are led to choose Goldkamp's "Belief Perspective II": 

our data indicate that Blacks make up a disproportionate share of shooting 
opponents reportedly armed with guns and a disproprotionate share of 

those reportedly engaged in robberies when police intervened. If one 

accepts both the accuracy of these reports and the premise that opponents 

armed with guns generally present the greatest and most immediate danger 
to police, there is little to support the contention that Blacks are 

shot disproportionately in relatively trivial and nonthreatening situa­



-25-

tions. A more conclusive answer to the question would require the cal­

culation of shooting rates for specific arrest situations by race. As was 

indicated earlier, the lack of race information on arrests precludes this 

analysis.
While our research has not conclusively confirmed Goldkamp’s Belief 

Perspective II, therefore, it has reduced to two the assumptions upon 

which one might base acceptance of the "police misconduct and different 

trigger finger" hypotheses implicit in his alternate theory. First, of

course, one might not accept the accuracy^of- the xeports. o£_Black/gun^ ---

incidents which account for most of our data set’s Black opponent dis­

proportion. The sheer nmnber of those shootings (656 are shown on 

Table 9), however, is so large as to suggest that the argument that 

"irresponsible use of deadly force by a small number of police officers 

accounts for disproportionate minority deaths is ill-founded.
Second, one might accept the accuracy of these reports, but properly 

note that we have not demonstrated that New York police do not refrain 

from shooting at Whites in situations comparable to those in which they 

do shoot at Blacks. Since most of the Black opponents in our data set 

were reportedly armed with guns, the assumption based on this observa­

tion requires its proponents to argue that police generally regard Blacks 

with guns as more threatening than Whites with guns. Our own logic and 

experience, however, suggest that police responses to such situations 

are based not upon opponent race, but rather upon opponent weapon.



-26-

Finally, we must qualify our acceptance of Goldkamp's Belief 

Perspective II. There is nothing in these analyses to support the con­

tention that the disproportion of Blacks among New York City police 

shooting opponents is reflective of police misconduct or racial discrim­

ination; but the limitations of our data have prevented us from examining 

the degree to which that disproportion is associated with the generally 

lower socio-economic position of Blacks. Differences among the shooting 

types which characterize the races (e.g., the high incidence of Black 

participation in shootings precipitated by robberies, which are most fre­

quent in blighted inner city areas and the high incidence of shootings 

involving Whites and vehicles, often preceded by car thefts, which are 

most frequent in middle and working class areas), however, suggest that 
this association may be strong.

Were we to conduct further research based upon data which included 

information about opponent socioeconomic status, we would hypothesize 

that Harding and Fahey's assessment of the role of class and affluence 
in shooting opponent racial disproportion would be confirmed. Were we 

successful, our research would strongly indicate that Black shooting 

opponent disproportion is neither a consequence of "overreaction" by 

invididual police officers nor of some racially varying predisposition 

toward violent crime. Conversely, it would point up the continuing 

existence of An American Dilemma described so well by Myrdal (1944) a 

a generation ago; Blacks are the node among New York's police shooting



-27-

opponents because they are also the mode among the lower socioeconomic 

groups which most frequently participate in the types of activity likely 
to precipitate esctreme police-citizen violence.



-28-

acknowledgement

This stiidy would not have been possible had not former Chief of 

Operations James T. Hannon, former Chief of Personnel Neil J. Behan, 

former Assistant Chief Patrick S. Fitzsimons, and Lieutenant Frank McGee 
of the New York City Police department been so generous in providing 
access to the necessary data.



-29-

FOOTNOTES

Oxir reservations here involve the accuracy and degree of detail 

provided in the shooting incident reports filed by individual officers. 

Because we attempted to limit our analyses to variables reasonably im­

mune to reporting bias (and often supported by witness statements), we 

regarded the question of the veracity of the data we did possess as one 

of minor importance. We were more troubled by the imoact on our research 

of the data we did not possess: 1058 (25.1 percent) of the data set's

shooting opponents were not identified by race. We did not find any 

s^idsmce that this information was deliberately withheld from reports in 

order to prevent or avoid any sensitive racial issues. This is so for 

several reasons. First, the "Firearms Discharge/Assault Report." forms 

which served as our primary data source include no caption requesting 
"opponent race," so that it appears only on reports filed by officers who 

volunteered it. These forms, originally designed to collect information 

training purposes, have more recently been supplemented by more com­
plete narrative reports. As a^result, the annual percentage of missing 
opponent race data declined from nearly forty percent in 1971 to 5.4 

percent in 1975. Second, despite this regular decrease in missing data, 

annual known opponent racial distributions have remained relatively con­

stant over the period studied. Third, the percentage of missing opponent 

race data is relatively evenly distributed across the City's police pre­
cincts, regardless of the racial characteristics of their total



-30-

populations (which one might reasonably expect to impact upon the charac­

teristics of their shooting opponents). Fourth, our opponent data suggest 

that many of these opponents were never seen by the officers involved; 

more than nine of ten "unknowns'* (90.6 percent) suffered no injury (72.6 

percent) or escaped their confrontation with unknown injuries (18.0 per­

cent). Further, our opponent arrest data reveal that 397 (37.5 percent) 

of opponents whose race is classified as unknown were not apprehended by 

police: in many cases, these individuals were merely shadowy figures

encountered on diJiily“lit“streets”or-rooftops.
2Except for homicides, the department began compiling these statistics 

only in 1976. The department's Homicide Analysis Section began systemati­
cally recording the race of those arrested for homicide in 1973. We 

elected not to use these data in our analyses because they do not describe 
the experience for the full five years of our study, and because they do 
not identify homicide perpetrators who are not apprehended.

3Burnham's sample is not random, but consists of 700 consecutive 
arrestees.

4See Loether and McTavish (1974:197,198) who describe Cramer's v 
as follows:

Cramer's v is, so to speak, a properly normed measure of associa­
tion for bivariat distributions of nominal variables, it is 
"margin free" in that the nximber or distribution of cases in 
row or column totals does not influence its value, nor is it 
influenced by the number of categories of either variable...
Cramer's v...can only be thought of as a magnitude on a scale



-31-

between zero and 1.0; the bigger the number, the stronger the 
association. It can not be interpreted, for example, as the 
percentage of variation in one variable explained by the other, 
nor can it be interpreted as the proportion of predictive 
error which may be reduced by prior knowledge of one of the 
variables.

^By aggregating our opponent age data into eight values to conform 

with those reported on New York City Police arrest data, we also found 

that the age distributions of shooting opponents and violent felony arres­
tees closely paralleled each other (v « .09).

^The decision to create this open ended age grouping (which inrltidAg 

individuals up to 79 years old) was made to simplify presentation and 

discussion of opponent age. The frequencies of single year values drops 
off dramatically after this point.

We defined the "primary opponent" as the only opponent, or in 
incidents involving more than one opponent, as the most heavily armed 

and/or most agressive and/or most seriously injured.
g
Intraracial events accounted for 94.4 percent of the multiple 

opponent shootings in which opponent race was reported.
9Although these differences would shrink if four alleged politically 

motivated "Black Liberation Army" assassinations were considered apart 
from other officer deaths perpetrated by Blacks.



-32-

RETERENCES

Breasted, Mary. "Police Use Cars and Clubs to Quell Brownsville Riot."
The New York Times. September 20, 1974, p.20.

Burnham, David. "3 of 5 Slain by Police Here Are Black, Same as the 
Arrest Rate." The New York Times. August 26, 1973. p.20.

Clark, Kenneth. Ooen Letter to Mayor Abraham D. Boame and Police
Commissioner Michael J. Codd. New York, September 17, 1974.

Fyfe, James J. Shots Fired: An Examination of New York City Police
Eirearms Discharges. (Ph.D. dissertation, S.U.N.Y., Albany), 
Arm-ARbor,-Michigan'Unxversrfy'Micf^ofilms International, (1978).

Goldkamp, John S. "Minorities as Victims of Police Shooting: Interpre­
tations of Racial Disproportionality and Police Use of 
Deadly Force." Justice System Journal. 2/2, (Winter, 1976), 
169-183.

Harding, Richard W., and Richard P. Fahey. "Killings by Chicago Police,
1969—70: An Empirical Study." Southern California Law Review.
4/6, (1973), 284-315.

Jenkins, Betty and Adrienne Faison. An Analysis of 248 Persons Killed bv
New York Citv Policemen. New York:Metropolitan Applied Research 
Center, Inc., 1974.

Kania, Richard R.E. and Wade C. Mackey. "Police Violence as a Function of 
Community Characteristics, Criminoloery, 15/1, (1977), 27-48.

Kobler, Arthur L. "Police Homicide in a Democracy," Journal of Social 
Issues, 31, (1975), 163-184.

Milton, Catherine H.; Jeanne Wahl Halleck; James Lardner; and Gary L.
Abrecht, Police Use of Deadly Force, Washington, D. C.:
Police Foundation, 1977.



-33-

Myrdal, Gunnar. An American Dileamia: The Negro Problem and Modern
Democracy. New York:Harper and Row, 1944.

New York City Police Department, Chief of Field Services. Summary of 
Precinct Populations. 1973.

Crime Analysis Unit. Monthly Arrest
EsDort. December, 1971, 1972, 1973, 1974, 1975. 

______________ — ♦ Homicide Analysis Unit. Annual Renort.
1976.

New York State Penal Law. 1967.

Robin, Gerald D. "Justifiable Homicide by Police Officers." Journal 
of Criminal Law, Criminology and Police Science. (May/June, 
1963), 225-231.

United States Department of Commerce. Characteristics of the Population. 
Washington, D. C.:U. S. Government Printing Office, 1973.

Wolfgang, Marvin. Patterns in Criminal Homicide. Philadelphia: 
University of Pennsylvania, 1958.