Geographic Correlates of Police Shooting: a Microanalysis

    GEOGRAPHIC CORRELATES OF 

POLICE SHOOTING:"

A MICROANALYSIS

James J. Fyfe 
Associate Professor 

The American University 
School of Justice 

Consultant
The Police Foundation 

Ph.D. (Criminal Justice), 1978, 
State University of New York 

at Albany



-2-

Abstract

This study examines the geographic distribution of all reported 

shootings by New York City police officers during the years 1971-1975 

(n=27A6) and indices of police exposure to violent crime (rates of arrest) 

for felonies against the person) and public safety (rates of murder and 

non-negligent manslaughter). The research finds that both these indices 

are closely related to police shooting rates, and then discusses 

theoretical and administrative implications of these findings.



-3-

One of the most momentous events in the criminal process occurs 

at the instant a police officer fires his gun at another human being. 

Further, because police shootings so frequently take place in already 

volatile inner-city areas (Kobler, 1975; Robin, 1963), their consequences 

often extend far beyond the officer-opponent diad most directly involved. 

During recent years, both the media and the professional literature have 

been replete with accounts of riots and civil disorders apparently pre­

cipitated by police shootings (Breasted, 1974; National Adivsory Commis­

sion on Civil Disorders, 1968).

Meaningful data on police shootings have only rarely been made 

available to social scientists.^ Therefore, the state of knowledge about 

this phenomenon is characterized by conjecture and hunches. Several 

writers, for example, have discussed or examined the geographic distri­

butions of shooting incidents within specific police jurisdictions and 

arrived at varying conclusions. Farrell (1977:72) observes that:

When the Los Angeles police shootings are plotted on 

a map, it becomes clear that the vast majority occur 

in a free-fire zone shared by the Hollywood, Ramparts,

Wilshire and 77th Street divisions: the rest of the city

is what the police call a "sleepy hollow," and when its 

population is subtracted from the total, the incident 

rate in the killing zone flares to alarming proportions.



-4-

Others offer different perspectives. Milton, et a L . (1977:144) 

suggest that police shootings are related to such measures of police 

exposure to potential violence as "the number of police on the street, 

the number of calls for service, the number and nature of dispatches...., 

and the number of arrests." Kania and Mackey (1977) relate variations 

in fatal police shooting rates across the fifty states to surrogate 

measures of "police exposure to threats and stress" and to indices of 

the "societal matrix" within which police function. In the former in­

stance, using Uniform Crime Report data, they report high associations 

between rates of fatal police shooting and rates of police exposure to 

reported violent crime. In the latter instance, they find a strong 

relationship between fatal police shooting rates and public homicide rates.

On the basis of these associations, Kania and Mackey propose (1977:43) 

that police violence is a response to general intracommunity violence. 

Police officers, they assert, observe and become intimately familiar 

with the "real behavior" of the communities in which they serve. They 

subsequently develop operating styles grounded on those usually accurate 

preceptions. Thus:

When he works in a community in which the resort to 

violence is a common, appropriate, or functional 

response to conflict and tension, the police officer 

will be more inclined to use violence in pursuit of 

legitimate ends...In those communities where the



-5-

resort to violence is neither normative nor commonly 

occurrent, the police officer will be restrained by 

his awareness of this fact, and will behave accordingly, 

thus minimizing the utilization of force and violence.

(1977:46)

Despite the weaknesses of the data used in the Kania and Mackey 
2study, their work clearly suggests that the geographic variations in

police shooting noted by Farrell (1977:72) are related to other community

characteristics. Consequently, important research questions pertaining

to those associations are investigated, albeit with limitations imposed

by the nature of the data available.

First, we seek to determine the degree to which the police shootings

in our data set are related to police exposure to "threats and stress."

Because we believe that arrests for violent crimes more directly measure
3police exposure to potential violence and are more accurately reported 

than are the Uniform Crime Report data analyzed by Kania and Mackey, we 

postulate that:
HI: The higher the violent crime arrest rate of a

neighborhood (zone), the higher the police shooting 

rate.
Second, to determine whether the association Kania and Mackey found 

between police shootings and the "public safety domain" of "the societal 

matrix" (1977:41) is true in the jurisdiction we studied, we hypothesized



-6-

that:
H2: The higher the reported public homicide rate of

a neighborhood (zone), the higher the police shooting 

rate.

Finally, because the police personnel data upon which we based our 

analyses affect only the presence of uniformed patrol officers and 

because many shootings involve on-duty plainclothes police or off-duty 

officers, we also investigate the following hypothesis:
I

H3: Both these relationships will be highest for

shootings involving on-duty police officers and 

lowest for shootings involving off-duty police.

In other words, we hypothesize that when analyses of HI and H2 are 

conducted separately for on-duty officers and for off-duty officers, 

the association will be stronger for on-duty officers.

THE RESEARCH SETTING

The Data

Police Shootings. The primary data source employed in the analysis 

of this hypothesis consists of New York City Police Department records of 

all reported shootings by its'officers between January 1, 1971 and 

December 31, 1975 (n=2926). New York City police are required to report 

all instances in which they "discharge firearms at other than authorized
4firearms ranges." Many such shootings (e.g., destruction of injured



-7-

animals, accidental discharges while handling weapons) are not relevant 

to the present questions and were excluded from analysis.

Because not all police shootings cause deaths (Milton, et al., 1977: 

16), we included woundings and off-target shots in our analyses in 

addition to cases where death occurred. Stated simply, "deadly force" 

is physical force capable of killing or likely to kill; it does not 

always result in death. Therefore, the true frequency of police use of 

firearms as a means of deadly force can best be determined by consider­

ing woundings, off-target shots, and fatal shootings as varying results 

of equally grave decisions. Consequently, our study includes a five 

year total of 2746 "shooting incidents"— confrontations with other human 

beings which involved shooting by one or more police officers.

Police Personnel Data. The New York City Police Department allocates 

patrol personnel to each of its precincts (the basic geographic entities) 

by using an equation, "Plan C," consisting of several measures which 

gauge the need for uniformed officers. The department's "allocation 

model" includes precinct area size and miles of street, number of radio 

calls for service, amount of time spent on calls per tour of duty 

(allowing for radio cars to be free of assignment and available for 

"preventive patrol" 30 percent of each tour): the model then distributes

10,000 officers among the precincts in direct proportion to the precincts' 

relative standing. "The Public Safety Pool"— all police officers avail­

able for assignment to precinct patrol beyond the 10,000 officers— are



-8-

then distributed by use of a second equation. This equation includes 

"reported outdoor crimes of violence" (murder, non-negligent manslaughter, 

forcible rape, robbery, felonious assault), which are weighted at .5, 

"reported indoor crimes'.' (murder, non—negligent manslaughter, robbery, 

felonious assault, burglary, and grand larceny auto) which are weighted 

at .3, and "all other complaints," weighted at .2 (NYPD, Plan C, 1971).

As noted earlier, the "Plan C" allocation model does not account for 

all department personnel (e.g., it does not affect deployment to units 

with citywide jurisdictions such as the narcotics division). It is, 

however, the most reasonable measure of police presence available. We 

elected to use the personnel distributions which result from it, there­

fore, as the basis for the construction of our police shooting rates.

To simplify this process, we obtained manpower figures effective as of 

July 1, 1973, the midpoint of the period we studied. These rosters 

were perceived as the most representative distribution of uniformed 

personnel assigned, since they fall between a two and one—half year 

"hiring freeze" and the 1975 layoffs of nearly 3,000 officers.

Level of Analysis. A logical unit of analysis for our research is 

the New York City police precinct (n=73). We found, however, that the 

physical boundaries and, thus, the personnel allotments of several pre­

cincts, had been changed during the period studies. For that reason, 

and because tables which used individual precincts as observed classes 

would be unwieldy and inconvenient for interpretation (Freeman, 195^.31),



-9-

we decided not to employ them as a basis of analysis. As an alternative, 

we used the department's more comprehensive command level, the "zone,"

(n=20) which varies in size from three to five neighboring precincts.

Even here, however, a slight modification was necessary. In order to 

assure constancy, the data reported for "Queens Zone 3" included shootings 

which occurred in the 113th precinct: the 113, a recently created juris­

diction, is currently a part of Queens Zone 4. In 1971 and 1972, however, 

the area it encompassed was policed by precincts included in Queens Zone 

3.

DEFINITION OF VARIABLES

Arrest and Homicide Rates. Our analysis involved the computation 

of several variables. First, to measure violent crime arrest rates, we 
collected zone arrest frequencies for murder and non-negligent manslaughter, 

forcible rape, robbery, and felonious assault from department reports 

(NYPD, Crime Analysis Section, 1971-1975). We then converted these to 

rates per 1,000 residents by employing the following equation:

Number of Violent Felony Arrests, 1971-75 ^
Arrest Rate - Resident Population, 1970 Census

The Crime Analysis Section's reports also supplied us with the base 

homicide data, which we converted to rates per 10,000 residents, with the 

following equation:

Reported Murder and Non-Negligent Manslaughter, ^971-75. ^ 10 000 
Homicxde Rate Resident Population, 1970 Census



These calculations resulted in the arrest and homicide distributions 

presented in Table 1- As one might expect, there is a significant rela­

tionship between our arrest and homicide rates over the City's 20 police
2 'zones (r = +.71, r =» .50). In addition, however, the table illustrates 

the hazards of using resident population figures as a "population at 

risk." Manhattan South Zone 3's arrest rate (86.46), for example, is 

nearly 50 percent higher than that of any other zone, but its homicide 

rate (19.21) is exceeded by three other zones.

(Insert Table 1)

The seeming disparity between these two rates may be ascribed, in 

large measure, to the atypicality of Manhattan South Zone 3 and the fact 

that its resident population (the Table's lowest, at 133,260) describes 

only a fraction of its "population at risk." Within this zone are located 

the Times Square area, Broadway and the theater district, the garment 

district, the diamond exchange, Rockefeller Center, Madison Avenue and 

the advertising district, the headquarters of the United Nations, New 

York City's two major railroad stations, its major passenger liner piers, 

its major bus terminal, Madison Square Garden, The New York Coliseum, and 

literally scores of hotels and thousands of commercial buildings. In 

light of the great numbers of non-residents in this area daily (the New 

York City Police Department estimate is 5.5 million), it is evident that 

its resident population is faulty basis for the computation of arrest or 

homicide rates. Consequently, we excluded this zone from further analyses 

in our research.

Shooting Rates. Before calculating our shooting rates, we examined 

the degree to which the duty status of the "primary officers"^ involved

-10-



-11-

in shootings varied across the zones. Table 2 suggests that they differ 

significantly (p = .0001). Further, the obtained Cramer’s v = .17 also 

suggests a moderate degree of independence among the zones in regard to
g

shooter duty status.

(Insert Table 2)

Possible reasons for differential distributions include patterns 

of assignment of non-uniformed officers. For example, civilian clothed 

narcotics and "decoy" officers are probably assigned most often to the 

areas where narcotics traffic and street crime are most prevalent. Simi­

larly, the potential for involvement in off-duty shooting incidents would 

be, at least in part, a function of levels of community violence and the 

numbers of officers residing in various areas of the city. A most obvious 

example is Staten Island (a relatively isolated "suburb within the city" 

consisting largely of private homes), where more than half (53.3 percent) 

of the reported shooting incidents involve off-duty officers. In light 

of Staten Island's 1973 Plan C allotment of 504 patrol officers and its 

April 1976 resident population (the only figure available to us) of 2691 

officers (NYPD, Personnel Data Section, 1976) this statistic is not 

extremely surprising. In the much more "active" borough of Manhattan 

(which included Manhattan South and North Field Services Areas), where 

635 officers resided in 1976 (NYPD, Personnel Data Section, 1976), only 

15.7 percent of the shooting incidents involve off-duty officers.

After noting this variation, we calculated our shooting rates (while 

holding constant the influence of variations in uniformed police presence) 

as follows;



-12-

Total Shooting Rata Number of Police Shootings, 1971-75 
Number of Uniformed Police Assigned ^

On-Duty Uniformed 
Shooting Rate

Number of Police Shootings Involving On-
_____ Duty Uniformed Police, 1971-75_____
Number of Uniformed Police Assigned ^

Number of Police Shootings Involving On-
On-Duty Plainclothes ^ Duty Plainclothes Police, 1971-75______  X 100

Shooting Rata Number of Uniformed Police Assigned

Off-Duty Shooting 
Rate

Number of Police Shootings Involving
_______ Off-Dutv Police, 1971-75_____
Number of Uniformed Police Assigned X 100

HI, H2, H3: ANALYSIS

Violent Felony Arrests and Police Shootings. To test HI, we cal­

culated Pearson's r's to describe the relationship between arrest rates 

and the four police shooting rates over the nineteen zones included for 

analysis. These correlation coefficients and their associated levels of 

significance and regression lines are presented in Figure 1. The data 

reveal that HI is confirmed; the relationship between arrest rates and 

total shooting rates (r = +.62) is significant at the .005 level. Further, 

H3 is confirmed insofar as it-applies to the relationship between shootings 

and arrest rates. The highest obtained r (+.72)_ described the correlation 

between arrest and on-duty uniformed shooting rates, while the least sig­

nificant relationship exists between arrest and off-duty shooting rates 

(r =« -.21). In addition, the figure's negative off-duty shooting rates



-13-

regression line reinforces the earlier observation that "quiet " areas 

generally are characterized by relatively high percentages of off-duty 

shootings.
(Insert Figure 1)

Homicide Rates and Police Shootings. H2 and the remainder of H3 

are also confirmed by the obtained correlation coefficients and their 

siginficance levels presented in Figure 2. Here, the Pearson's r's for 

total (+.78) and on-duty uniformed (+.89) shooting rates strongly suggest 

that police shootings are closely associated with this index of public 

safety. Further, inasmuch as only one or two officers were arrested each 

year for homicide during the period we studied, it is clear that police 

shootings themselves are not responsible for this variation. In addi­

tion, we find again that "busier" areas— in terms of reported murders 
and non-negligent manslaughters— are marked by relatively high percentages 

of shooting by on-duty officers, and that off-duty shootings are usually 

relatively greatest in "quiet" areas, where many officers reside and 

socialize.
(Insert Figure 2)

CONCLUSIONS

Theoretical Implications. Although the measures employed in our 

study differ from those used by Kania and Mackey, we have also found 

high correlations between extreme police—citizen violence and indices of



-14-

threats to police and of general public safety. Further, as the scatter 

diagrams presented in Figures 3 and 4 (which include Manhattan South 

Zone 3) illustrate. New York City, like Los Angeles, is composed of a 

few "free fire zones" and several "sleepy hollows." To the degree that 

our findings may be generalized to Los Angeles and other jurisdictions, 

therefore, it is likely that the geographic distributions of their police 

shootings also parallel variations in these indices.

(Insert Figures 3, 4)

Like Kania and Mackey, we acknowledge the difficulty of imputing 

causal order to these relationships. Further, we find ourselves in 

agreement with their conclusion (1977:46) that police violence varies 

as "the police officer (reacts) to the community as he perceives it, a 

perception which is usually correct." Like Kania and Mackey, we arrive 

at this conclusion only after considering and rejecting the two alternate 

interpretations of our findings.
Most specifically, we would discount the converse proposition that 

arrest and homicide rates are a product of variations in the frequency 

of police use of deadly force. The literature is replete with attempts 

to explain arrest and general violence levels, but neither previous 

research nor speculation attributes these levels to variations in the 

frequency with which police use their gxins. Except for the previously 

noted civil disorders of the 1960s, we doubt that such a comparatively 

rare phenomenon as police shooting (the highest annual rate of shootings



-15-

per thousand residents during the five years we studied was .21, or one 

shooting per 5,000 residents, in Manhattan North Zone 3) exerts signi­

ficant influence over arrest rates (which varied nearly tenfold) or 

homicide rates (which varied more than 25-fold). Our data set does con­

tain isolated exceptions (e.g., officers who apparently "expected trouble" 

in situations in volatile areas and subsequently became involved in self- 

fulling prophecies), but we would argue that the police perception of the 

community is "usually correct."

We also reject the third alternative which, as Kania and Mackey 

note (1977:43), involves the possibility "that the entire societal matrix 

(including police shooting) is produced by a third factor, or group of 

factors, as yet unknown." However, we would not concur with them that 

"there is nothing in the literature to suggest even in which direction 

to proceed in seeking this hypothetical factor."

It seems reasonable to suggest that there are two major sources of 

spuriousness vis-a-vis the presently investigated relationships between 

(1) threats to police and police shootings, and (2) general public safety 

and police shootings. With respect to (1), the major source of spur­

iousness would be the intervening effects of variation in perception of 

threats across the zones. Our measure of threat was a measure of actual 

threat, viz., violent felony arrests, yet it could be that differences 

exist among officers so that those who are more likely to perceive a 

given situation as a threat are assigned to the zones having greater rates

/



-16-

of police shootings. In fact, Kania and Mackey's argument (pp 4-5 above) 

suggests that police learn to respond to violence with use of deadly force, 

a fact that is itself indicative of a perceptual effect. On the other 

hand, a spurious perceptual effect would be constituted by perceptual 

differences that existed before exposure to a high threat zone. For the 

current study, such measures are not available, but one would assume that 

indicants of such a "tendency of the zone to foster stress-laden percep­

tions" could be developed from measures of rate of calls answered per 

officer (a general workload measure) and rate of sick calls (a general 

disability or stress avoidance measure).

With respect to (2), the major source of spuriousness would be the 

known correlates of general public safety. Kania and Mackey either rule 

out the effects of a number of social correlates or indicate their high 

multicollinearity, thereby supporting the general social matrix argument 

(1977:41-46). On the other hand, given the broad difference in size 

between units of analysis in their study and this study (states vs. zones 

of a city), it might be possible for such social correlates to have a 

different elaboration or specification of the relationship between general 

public safety and police shootings in zones of a city. Although we have 

not investigated these social""correlates directly, there is evidence that 

offenses other than homicide (e.g., robbery and burglary), which have 

different criminogenic correlates (and hence measure different dimensions 

of public safety), are related in the same way as homicide rates to police



-17-

shootings- Consequently, we would argue that it is unlikely that social 

forces directly productive of general public safety from crime in 

general affect police shootings independently of the public safety 

construct.
Thus, these analyses suggest no spuriousness in the relationships 

investigated when the major likely sources of spuriousness are considered.

Administrative Implications. Our findings regarding the relation­

ships that exist between police shooting and violent felony arrest and 

homicide frequencies support the measure of police shooting restraint 

and departmental firearms policy suggested by Milton, et al. (1977:140). 

Even though the New York City shootings apparently vary in terms of pre­

cipitating events from those which occur elsewhere, we would anticipate 

finding similar arrest/shooting and homicide/shooting relationships in 

other jurisdictions.^ If so, comparisons of ratios of violent felony 

arrests to shootings among police agencies might provide useful information 

about the effects of varying firearms policies and other organizational 

and environmental influences. Although a rigid empirical analysis of 

variations among these ratios might not be possible (or practical), we 

would argue that jurisdictions-whose police used their guns relatively 

often (in terms of the numbers of reported homicides or violent felony 

arrests affected) might profit from the experiences of a police depart­

ment whose members did not. Similarly, these same ratios might be employed 

within police agencies to examine longitudinal shooting patterns or to 

assess the effects of changes in policy or operational strategies.



-18- 

TABLE 1

,a,bNEW YORK CITY VIOLENT FELONY ARREST ’ AND REPORTED MURDER AND
NON-NEGLIGENT MANSLAUGHTER^ RATES BY ZONE, 

JANUARY 1, 1971-DECEMBER 31, 1975

Zone Resident
Population

Violent Felony 
Arrests

Number Rate per 
1000 pop

Reported Murders and Non- 
Negligent Manslaughters

Number Rate per 
10,000 pop.

Manhattan So. 1 138198 5964 43.16 220 15.92
Manhattan So. 2 264566 9497 35.90 414 15.65
Manhattan So. 3 133260 11521 86.46 256 19.21
Manhattan No. 1 359819 8358 23.23 333 9.25
Manhattan No. 2 417285 9538 22.86 564 15.67
Manhattan No. 3 249270 12473' 50.04 1271 50.99
Bronx 1 374065 22610 60.44 1000 26.73
Bronx 2 612740 10406 16.98 579 9.45
Bronx 3 481835 ; 7876 16,34 246 5.10
Brooklyn So. 1 604658 5377 8.89 177 2.93
Brooklyn So. 2 543817 4455 8.50 137 2.62
Brooklyn So. 3 427256 10229 23.94 411 9.62
Brooklyn So. 4 305341 8417 27.57 242 7.92
Brooklyn No. 1 389823 14375 36.88 842 21.60
Brooklyn No. 2 451138 9649 21.39 533 11.81
Queens 1 505084 4318 8.49 155 3.07
Queens 2 500827 4133 8.25 • 138 2.76
Queens 3*̂ 752753 10154 ' 13.49 338 4.50

0Queens 4 255231 3349 13.12 113 4.43
Staten Island 355690 2429 6.83 67 1.88

Totals 8,102,476 -175,128 21.61 8036 9.92

r = +. 71 r  ̂=.50 p = . 01

Includes murder, non—negligent manslaughter, forcible rape, robbery, felonious 
assault-
Calculated from: New York City Police Department, Crime Analysis Unit, Monthly
Arrest Report, December, 1971, 1972, 1973, 1974, 1975.
'^Calculated from; New York City Police Department, Chief of Field Services, 
Snmniary of Precinct Population, 1973. 

d.Includes 113 Precinct.
'Excludes 113 Precinct.



TA3LE 2

-19-

DUTY STATUS OF PRIMARY OFFICER IM NEW YORK CITY POLICE 
SHOOTING INCIDENTS INVOLVING OPPONENTS BY ZONE, 

JANUARY 1, 1971-DECEMBER 31, 1975

Zone

Primary Officer Duty Status
On Duty 
Uniform

% n

On Duty 
Civilian 
Clothes 

% n

Off

%

Duty

n

Totals

Manhattan So. 1 50.0 (38) 31.6 (24) 18.4 (14) 2.8 (76)
Manhattan So. 2 51.5 (68) 30.3 (40) 18.2 (24) 4.8 (132)
Manhattan So. 3 58.6 (58) 29.3 (29) 12.1 (12) 3.6 (99)
Manhattan No. 1 47.8 (55) 37.4 (43) 14.8 (17) 4.2 (115)
Manhattan No. 2 51.6 (79) 32.0 (49) 16.3 (25) 5.6 (153)
Manhattan No. 3 63.4 (168) 21.5 (57) 15.1 (40) 9.7 (265)
Bronx 1 57.9 (154) 27.8 (74) 14.3 (38) 9.7 (266)
Bronx 2 44.8 (100) 33.6 (75) 21.5 (48) 8.1 (223)
Bronx 3 44.7 (59) 26.5 (35) 28.8 (38) 4.8 (132)
Brooklyn So. 1 46.7 (28) 26.7 (16) 26.7 (16) 2.2. (60)
Brooklyn So. 2 54.9 (39) 22.5 (16) 22.5 (16) 2.6 (71)
Brooklyn So. 3 49.6 (66) 34.6 (46) 15.8 (21) 4.9 (133)
Brooklyn So. 4 55.6 (50) 21.1 (19) 23.3 (21) 3.3 (90)
Brooklyn No. 1 60.6 (134) 17.6 (39) 21.7 (48) 8.1 (221)
Brooklyn No. 2 58.8 (94) 27.5 (44) 13.8 (22) 5.8 (160)
Queens 1 43.6 (48) 25.5 (28) 30.9 (34) 4.0 (110)
Queens 2 33.8 (27) 23.8 (19) 42.5 (34) 2,9 (80)
Queens 3 49.5 (110) 20.7 (46) 29.7 (66) 8.1 (222)
Queens 4 43.8 (32) 26.0 (19) 30.1 (22) 2.7 (73)
Staten Island 36.7 (22) 10.0 (6) 53.3 (32) 2.2 (60)

Totals 52.1 (1429) - 26.4 (724) 21.5 (588) 100.0 (2741)

not ascertained = 5 
chi-square = 155.75984 
p = .0001 V = .17

Includes 113 
^Excludes 113

Precinct,
Precinct,



FIGURE 1

LINEAR REGRESSIONS FOR NEW YORK CITY VIOLENT FELONY 
ARREST RATES AND POLICE SHOOTING RATES BY ZONE, 

JANUARY 1, 197I-DECEMBER 31, 1975^

36

IK)01

Arrest Rate
(per 1,000 population)

Excludes Manhattan South Zone 3



FIGURE 2

LINEAR REGRESSIONS FOR NEW YORK CITY REPORTED 
MURDER AND NON-NEGLIGENT MANSLAUGHTER RATES 

AND POLICE SHOOTING RATES BY XONE 
JANUARY 1, 1971-DECEMBER 31, 1975^

Total Shooting Rate

I
N3

Homicide Rate
(per 10,000 population)

Excludes Manhattan South Zone 3



SCATTER DIAGRAM OF NEW YORK CITY VIOLENT FELONY 
ARREST RATE AND POLICE SHOOTING RATE BY ZONE 

JANUARY 1, 1971-DECEMBER 31, 1975

FIGURE 3

36

IN3
NJ
I

Arrest Rate
(per 1,000 population)

Manhattan South Zone 3



FIGURE 4

SCATTER DIAGRAM OF NEW YORK CITY REPORTED 
MURDER AND NON-NEGLIGENT MANSLAUGHTER 
RATE AND POLICE SHOOTING RATE BY ZONE,

IN)
Oj
I

Manhattan South Zone 3



-24-

FOOTNOTES

1- Many, if not most, empirical studies of police shooting have been 

conducted in the absence of cooperation from the police agencies in­

volved. See, for example, Harding and Fahey (1973); Kobler (1975). For 

the assistance extended us in this effort, we are indebted' to former 

Chief of Personnel Neil Behan, former Assistant Chief Patrick S. 

Fitzsimons, and Lieutenant Frank McGee of the New York City Police 

Department.

2. Kania and Mackey attribute their use of the states as communities 

to the absence of complete municipal, county, or metropolitan area 

statistics (1977:47). Further, the questionable accuracy of the Uni­

form Crime Report data upon which they base one of their major findings 

is be now well documented (see, for example, Skogan, 1974).

3. This is especially true in New York, where officers are permitted 

to employ deadly force to effect arrests for felonies against the 

person and in defense of life.



-25-

4. This reporting requirement is included in a New York City Police 

Department directive, S.O.P. 9 (1969). This mandate does not exempt 

our study from the criticism that the validity of research examining 

reported phenomena may be questioned because of the possibility of a 

"dark figure" of unreported incidents. While a precise assessment of 

the number of such omissions from our data is not possible, it is likely 

that they are minimal. It is probable, for example, that unreported 

shootings would most often involve shots which violated department 

shooting regulations and which missed their targets and which were 

fired in thinly populated or deserted areas. This is so for three 

reasons. First, an officer whose actions were justifiable would have 

little to fear from a review of his shooting but would be disciplined 

if his failure to report it were discovered. Second, an officer who 

missed one or more questionably justifiable shots at a civilian and 

who wished to avoid an investigation into the use of his firearm would 

be far less likely to report the incident than would an officer who had 

(or thought he might have) hif his target. Finally, the possibility of 

unreported incidents would be greatest in thinly populated or deserted 

areas, where the chances of the presence of uninvolved bystanders (or 

other officers) who might later call the shootings to official attention



-26-

would be least. Because of the population density of New York City 

(where police usually do not work alone), it would appear unlikely 

that more than a few incidents occur beyond the earshot of third parties. 

We concluded, therefore, that the impact of unreported shootings on our 

data and findings would be relatively minimal. Similarly, because our 

study involves variables reasonably immune to reporting'bias (e.g., 

shooting location, officer duty status), we concluded that reporting 
inaccuracies were of little or no import.

5. The term "primary officers" refers to the first (or only) officer 

at a police shooting and/or the most aggressive (in terms of having 
fired the most, or only, police shots).

6. Cramer's v is a normed measure of association for nominal variables 

whose value is not affected by case number or distribution. It varies 

from zero to 1.0 and its value increases with the degree of association. 
See Loether and McTavish (1974:197,198).

7. Our data reveal that "robberies" are the modal precipitant of New 

York City police shootings (31 percent), while "family disputes" and 

other "respond to disturbance" incidents precede only 11.7 percent of the 

shooting incidents studied (see Fyfe, 1978:500-504). This finding is

at variance with those of Kobler (1975), Milton, el al., (1977) and 

Robin, (1963), who generally report that "respond to disturbance" calls 

or "vehicle stops" account for a considerable larger percentage of 

shootings than they do in New York City. One might postulate several



-27-

reasons for this variance (e.g., the comparatively high frequency of 

robberies in densely populated urban areas such as New York, the rela­

tively small geographic areas patrolled by New York City police and the 

consequent short response time). Its major significance to the present 

study, however, involves the possibility that the strong relationship 

between New York City police shootings and arrests for violent felonies 

is a result of the fact that robberies (one of the offenses included in 

our violent felony index) precipitate an inordinate amount of the city's 

police shootings. In other jurisdictions, however, we would anticipate 

finding that the relationship between violent felony arrest rates and 

shootings is more heavily dependent on "respond to disturbance" calls 

which presvimably also result in arrests for "aggravated assault" on 

the officers involved.



-28-

REFERENCES

BREASTED, M.

1974 "Police Use Cars and Clubs to Quell Brownsville Riot."

The New York Times (September 20):20.

FARRELL, B.

1977 "The Deadly Sin of Police Panic." New West 

(September 26):72-70.

FREEMAN, L. C. , ^

1965 Elementary Applied Statistics for Students in Behavioral 

Science. New York:Wiley.

FYFE, J. J.

1978 "Shots Fired: An Examination of New York City Police

Firearms Discharges." Doctoral dissertation. School of 

Criminal Justice, State University of New York at Albany.

Ann Arbor, Michigan: University Microfilms.

HARDING, R. W. , and R. P.: FAHEY

1973 "Killings by Chicago Police, 1969-70: An Empirical Study."

Southern California Law Review 4:284-315.

KANIA, R. R. E., and W. C. MACKEY

1977 "Police Violence a‘s a Function of Community Characteristics." 

Criminology, 15:43-47.



-29-

KOBLER, A. L.

1975 "Figures (and Perhaps some Facts) on Police Killing of 

Civilians in the United States, 1965-1969." Journal of 

Social Issues 31:185-191.

LOETHER, H. J., and D. G. McTAVISH

1974 Descriptive Statistics for Sociologists. BostonrAllyn 

and Bacon.

MILTON, C. H., et al

1977 Police Use of Deadly Force. Washington, D.C.:Police Foundation 

NATIONAL ADVISORY COMMISSION ON CIVIL DISORDERS

1968 Commission Report. New York:Bantam 

NEW YORK CITY POLICE DEPARTMENT

1969 S.O.P. 9.

NEW YORK CITY POLICE DEPARTMENT, CRIME ANALYSIS SECTION 

1971-1975 Monthly Arrest Report. (December).

NEW YORK CITY POLICE DEPARTMENT, OFFICE OF PROGRAMS AND POLICIES 

1971 Plan C.

NEW YORK CITY POLICE DEPARTMENT, PERSONNEL DATA SECTION

1976 Residence of Sworn Personnel (April).

NIE, N. H., et al.
1975 Statistical Package for the Social Sciences, second edition.

New York:McGraw-Hill



-30-

ROBIN, G. D.

1963 "Justifiable Homicide by Police Officers." Journal of

Criminal Law, Criminolosy and Police Science. 54:225-231. 
SKOGAN, W. G.

174 "The Validity of Official Crime Statistics: An Empirical

Investigation." Social Science Quarterly, 55:25-38