Understanding the Use of Force By and Against the Police.
Series: NIJ Research in Brief
Published: November 1996
15 pages
30,224 bytes

U.S. Department of Justice
Office of Justice Programs
National Institute of Justice

National Institute of Justice
Research in Brief
Jeremy Travis, Director
November 1996

============================================= 

Figures, charts, forms, and tables are not included
in this ASCII plain-text file. To view this
document in its entirety, download the Adobe
Acrobat graphic file available from this web site
or order a print copy from NCJRS at 800-851-3420.

============================================= 

Understanding the Use of Force By and Against the
Police

Issues and Findings

Discussed in this Brief: An NIJ-sponsored study of
1,585 adult custody arrests in Phoenix, Arizona,
the eighth largest city in the United States, to
determine the use of force both by and against the
police.

Key issues: The research conducted by Rutgers
University and Arizona State University was
prompted by a perceived need on the part of the
Phoenix Police Department and NIJ to gather
information about arrest situations, tactics,
policy, training, and practices. The study surveyed
police officers over a 2-week period in June 1994
and interviewed 185 suspects.

Key findings: Force was used infrequently by the
police and even less frequently by suspects.
Suspects interviewed reported levels of police
force similar to those obtained from officer self-
reports.

Among the other findings: 

o Police used some physical force in about one of
every five arrests. Suspects used some physical
force in about one of every six arrests.

o Phoenix police officers are required to restrain
only felony or belligerent suspects. In 20 percent
of all adult custody arrests studied, officers
opted to use no restraints.

o When force was used by the police or suspects, it
was typically at the low end of the severity scale.

o Weapons were used by the police in 2 percent of
all arrests. The weapon most frequently used by the
police was a flashlight (12 times in 1,585
arrests).

o Of 41 factors examined, only 9 consistently
contributed to the prediction of police use of
force.

o The single best predictor of police use of force
was suspect use of force.

o Two-thirds of the variation in the amount of
force used by police remains unexplained.

Target audience: Law enforcement policymakers,
practitioners, and police trainers

============================================= 

Understanding the Use of Force By and Against the
Police

by Joel Garner, John Buchanan, Tom Schade, and John
Hepburn

The Phoenix Police Department, in conjunction with
Rutgers University and Arizona State University,
designed and implemented a study of the use of
force by and against Phoenix police officers.
Conducted during a 2-week period in June 1994, the
research was prompted by the need to generate
systematic information on the role of force in
arrest tactics policy, training, and practice. This
study, sponsored by the National Institute of
Justice (NIJ), was concerned with determining the
characteristics of arrest situations, suspects, and
officers associated with the use of force, as well
as the amount of force actually used. 

Research results showed that, during the study
period, the Phoenix police employed some degree of
physical force in 22 percent of adult custody
arrests and that suspect use of physical force was
even less frequent. Further, in situations where
members of either group did employ force, the
levels typically were low. Despite a policy that
requires the routine use of restraints in arrests
of felons and belligerent suspects, officers did
not restrain suspects in one-fifth of all custody
arrests involving adults. Finally, a number of
factors--most notably, suspect use of force--were
found to predict police use of force (see exhibit
1).

This Research in Brief explains the methodology
used to measure the use of force, discusses study
results, and presents the implications of these
findings for future policy and research.

Study method

The research was designed to address some of the
more important limitations of prior research on use
of force. First, measures were developed to capture
the range of force used within the Department's
standard force continuum. Measures of the simple
dichotomy between force and no force (or excessive
and reasonable force) neither capture the full
range of police or suspect behavior nor recognize
the great variety of force-related behavior in law
enforcement. Second, rather than focusing
exclusively on situations that involved severe uses
of force such as firearm discharges, serious
injuries, or deaths, data were collected from a
representative sample of arrests to permit
examination of the range of instances where some
degree of force could be used.

A third aim of the study was to determine the
correlates of force. Although helpful, descriptive
data on arrest situations or force levels do not
provide, by themselves, an appropriate basis for
developing arrest tactics training. Therefore,
researchers sought to understand the specific
circumstances and situations where more or less
force is used to accomplish an arrest and to use
that information to improve training and practice
in arrest tactics.

Obtaining the information. The primary source of
data was a one-page (front and back) survey
completed by Phoenix police officers after arrests
were made during a 2-week period in June 1994 (see
exhibit 2). This form was used to record specific
behavioral aspects of the arrests and gather
information on police mobilization, the nature of
the offense, and officer and suspect
characteristics.

From these surveys, detailed information was
compiled about the frequency of specific officer
and arrestee behaviors along five dimensions:

o Voice (officer/suspect spoke in a conversational
tone, articulated threats, issued commands,
shouted, or cursed).

o Motion (mode of officer pursuit/suspect flight).

o Restraints (mechanical devices used by police to
control arrestee, e.g., handcuffs, body cuffs,
hobble).

o Weaponless tactics (officer/suspect grabbing,
twisting, pushing, shoving, punching, kicking,
biting, scratching, wrestling, scuffling, and use
of control holds or pressure points).

o Weapons (suspect possession or officer/suspect
threatened use, or actual use of martial arts or a
baton, flashlight, knife, blunt instrument,
chemical agent, canine, handgun, rifle, or shotgun;
includes use of a motor vehicle as a weapon).

The detailed survey responses were used to
construct three measures of force--physical force,
continuum of force, and maximum force. These
measures were designed to capture the full range of
force used, including low levels of force not
traditionally examined by police use-of-force
research. Multivariate statistical models also were
developed to evaluate the extent to which officer,
suspect, and arrest situation characteristics
consistently predicted the amount of force used.

The reliability of the police survey was assessed
by interviewing a sample of suspects booked in the
Maricopa County Jail; 185 interviews were matched
to officer surveys. Both the officer surveys and
suspect interviews were voluntary and anonymous.

Overview of survey results

Of the 1,777 surveys obtained during the 2-week
period covered by the study, 1,585 pertained to
adults who were booked by Phoenix police officers
at the Maricopa County Jail (noncustody and
juvenile arrests were excluded). The Department's
automated information system (PACE) recorded 1,826
arrests in which an adult suspect was booked. Thus,
surveys were obtained for more than 85 percent of
arrests involving detained adults.

The surveys revealed that, in making these 1,585
adult custody arrests, officers:

o Used threats or shouts less than 4 percent of the
time.

o Pursued a fleeing suspect 7 percent of the time.

o Placed cuffs or restraints on 77 percent of the
suspects.

o Used a weaponless tactic (holding, hitting, etc.)
in 17 percent of the arrests.

o Threatened to, but did not, use a weapon 3.7
percent of the time.

o Used a weapon in 2 percent of the arrests.

Surveyed officers also reported that the weapon
most frequently used by them against a suspect was
a flashlight (12 arrests). They also noted that no
restraints were used in 20 percent of the surveyed
cases (3 percent of completed surveys did not
indicate whether restraints were used). According
to survey data, use of force was reported rarely by
either police officers or suspects; when some form
of force was used, it was typically at the low end
of the study's measures.

The three measures of force

Physical force. "Physical force" was defined as
officer or suspect use of any weaponless tactic
(such as kicking or shoving) or the threatened or
actual use of any weapon. In addition, police use
of physical force included the application of
severe restraints, including cuffing suspects while
they were prone, hobbling (hog-tying a suspect's
legs), leg cuffing, and body restraint (e.g., a
straitjacket). The possibility that suspects would
use restraints on police officers was excluded from
consideration. Suspect possession of a weapon, even
if not used or threatened, was included as an
example of physical force by suspects. Officer
possession of a weapon was not considered use of
physical force by officers.

In 349 of the 1,585 surveyed arrests (22 percent),
the police reported using some form of physical
force. In nearly four out of every five adult
custody arrests, police officers used no physical
force at all. The surveys indicated that suspects
used physical force in 228 (14.4 percent) of the
1,585 incidents. In roughly five out of every six
adult arrests, suspects reportedly used no physical
force.

Continuum of force. The study's second measure of
force was based on two six-step rankings used by
the Phoenix Police Department (see exhibit 3),
which are similar to those used in many other
jurisdictions to locate degrees of force in a
progression. The rankings are independent; a "3" on
the police scale (control and restraint) is not
necessarily equivalent to a "3" on the suspect
scale (passive resistance). The measurement of this
"continuum of force" was intended not only to
reflect the official policies of the Phoenix Police
Department but also to incorporate the widely held
notion that the force/no force dichotomy is
inadequate to capture all important variations in
the ways police handle encounters with the public.

Exhibits 4 and 5 show the highest level of force
used by police and suspects in terms of the force
continuum. In 918 (57.9 percent) arrests, the
highest level of force used by the police was some
form of restraint; in another 350 (22.1 percent),
no restraints were used. Chemical weapons were the
highest level of force employed by the police in 2
(.01 percent) arrests; although not discharged,
firearms were threatened or somehow used in 54 (3.4
percent) arrests. Other weapons and weaponless
tactics were used in 261 (16.5 percent) arrests.

In 977 (61.6 percent) arrests, suspects offered no
resistance to officers. In another 196 (12.4
percent), the forms of resistance were either
psychological or verbal. In 136 (almost 9 percent)
arrests, suspects used or threatened to use a
physical tactic or a weapon; in 11 (0.7 percent) of
those arrests, the weapon was a firearm (see
exhibit 5).

Maximum force. A measure of maximum force was
constructed by identifying the single most severe
use of force employed by the police, based on a
ranking of 80 different types of police statements,
physical restraints, tactics, and weapon use. These
rankings ranged on a scale of 0 to 100 (with 100
denoting maximum force). The severity of behaviors
was weighed according to the judgment of 11
experienced and currently active Phoenix patrol
officers, whose opinions were solicited for this
purpose. In a similar exercise, a measure of the
maximum amount of force used by suspects was
developed (see exhibits 6 and 7).

Suspects' views on use of force. In order to assess
the reliability of the study's force measures,
results of interviews with suspects from a sample
of 185 adults taken into custody during the 2-week
period were matched to corresponding police
surveys. Interviewers from Arizona State University
asked detainees about the force used by police in
the course of their arrests, as well as the force
that they themselves exhibited. The interviews
resulted in findings that were comparable to those
of the police surveys (see exhibit 8).

Influences on use of force

In addition to developing three measures of force,
the study team compiled detailed information on how
the police were mobilized, the nature of each
offense, the location of each offense, the personal
characteristics of officers, the personal
characteristics of suspects, and any associations
that could be made between police and suspect
characteristics. Information about each arrest was
collected on 41 specific items thought to influence
the use of force.

Analyses of these variables in relation to each
measure of force identified 16 consistent
nonpredictors, 16 inconsistent predictors, and 9
consistent predictors of force.

Consistent nonpredictors. More than one-third (16)
of the 41 factors examined did not predict any of
the three measures of police or suspect use of
force. These factors were:

MOBILIZATION OF THE POLICE
Custody status of arrestee
Dispatch or on-view arrest
Night time
Weekends
Weekend nights

NATURE OF THE OFFENSE SITUATION
Number of suspects at initial contact
Number of suspects at arrest completion
Victim and suspect friends or same family
Bystander and suspect friends or same family

NATURE OF THE ARREST LOCATION
Inside a residence
Location known for criminal activity

PERSONAL CHARACTERISTICS OF THE POLICE
Number of arrests in past 30 days
Years since last training

PERSONAL CHARACTERISTICS OF SUSPECTS
Known to have criminal record

ASSOCIATED SUSPECT AND OFFICER CHARACTERISTICS
Height
Weight

Given the size and representative nature of the
sample of adult arrests from which various measures
of potential predictors were obtained and the
strength of the multivariate analyses employed,
these findings of "no effect" cannot be attributed
to either the research design or its
implementation. Although generalization of these
findings awaits replication, future discussions of
police use-of-force policies and practices (and
designs for additional research on police use of
force) should consider this study's evidence that
these 16 factors consistently failed to predict
police or suspect use of force.

Inconsistent predictors. Sixteen of the 41 factors
considered in the study predicted only one or two
of the three measures of force used by and against
police. These factors retain their candidacy as
predictors of force and warrant inclusion in
subsequent research. Each predicted some measure of
force but failed to meet this study's conservative
standard of predicting all three measures.

In the following list of inconsistent predictors,
the minus (-) signs indicate a negative
relationship. Thus, as a predictor (such as
visibility) increases in value, the use of force
decreases. The negative sign on the race variable
is based on the finding that Hispanic suspects on
average use less force against the police than non-
Hispanic suspects. 

MOBILIZATION OF THE POLICE
Patrol division (-)
Early, middle, or late phase of shift
Number of police at initial contact

NATURE OF THE OFFENSE SITUATION
Traffic offense
Property offense
Vice offense
Domestic call
Bystander's demeanor

NATURE OF THE ARREST LOCATION
Visibility (-)
Inside a nonresidential building (-)
Location known to be hazardous

PERSONAL POLICE CHARACTERISTICS
Length of service
Past injury

PERSONAL SUSPECT CHARACTERISTICS
Drug impaired

ASSOCIATED SUSPECT AND OFFICER CHARACTERISTICS
Age
Race (-)

Because suspect use of force influences police use
of force, this list of inconsistent predictors
includes any factor that predicted one measure of
suspect use of force. For instance, prior injury to
an officer did not predict police use of force
directly, but it did predict one of the three
measures of suspect force. Similarly, the race of
officers and suspects played no role in predicting
police use of force, but because Hispanic suspects
used less force than other suspects on one measure,
race was listed as an inconsistent predictor of
force.

Age is one of the most consistent predictors of
participation in and frequency of criminal
behavior, and we were surprised that age was not a
substantial and consistent predictor of the use of
force. In fact, at least one of the various
components of "age"--officer age, suspect age, and
the interaction of the two--was found to be
associated with all three measures of force
employed in the study, but the direction of the
effects changed with different measures and
components.

Consistent predictors. The analyses identified nine
characteristics that consistently predicted police
use of force:

MOBILIZATION OF THE POLICE
Use of contact and cover tactics
Increased number of police

NATURE OF THE OFFENSE
Arrest for a violent offense

NATURE OF THE LOCATION
Presence of bystanders

PERSONAL POLICE CHARACTERISTICS
None

PERSONAL SUSPECT CHARACTERISTICS
Use of force
Gang involvement
Alcohol impairment
Known to be resistive, assaultive, or to carry a
weapon

ASSOCIATED SUSPECT AND OFFICER CHARACTERISTICS
Both male

Some predictors affected police use of force
directly; others affected it through their
influence on suspect use of force. Some
characteristics predicted both.

Among the predictors of police use of force,
suspect use of force had the largest impact on each
of the three measures of police use of force. This
remained true when controlling for the possibility
that some suspect use of force could be a reaction
to police use of force. This finding supports the
perspective that underlies use-of-force policies
and arrest tactics training in the Phoenix Police
Department and in many other departments around the
country: Police use force (and are authorized to do
so) in response to suspects' levels of resistance.

However, suspect use of force does not explain all
or even a large proportion of the variation in the
amount of force used by the police. This finding
supports the perspective that response to suspect
force, although significant, is not the only
situation in which the police use force (see
exhibit 1).

Implications of study findings

Policy implications. This research has specific
implications for police policy, training, and
practice in the use of force. First, it provides
systematic evidence that the use of force in
Phoenix is infrequent; when used, force was
typically at the lower end of the measures of
force. In addition, no evidence was found to show
that force was applied unevenly or in
discriminatory ways against racial minorities. 

The findings did raise some issues for further
consideration. For instance, the single most
frequent weapon used when arrests were made in
Phoenix was the flashlight. At the time of this
study (June 1994), the Phoenix Police Department's
arrest tactics training program provided limited
guidance regarding the use of a flashlight as a
weapon. Officers are currently receiving enhanced
training in the use of the flashlight as a weapon.
The same rules that apply to the use of batons
apply to flashlights.

A second area of concern is that the widely
promoted "contact-and-cover" tactic was
consistently associated in this study with
increased use of force. Contact and cover is a
tactic whereby one of two officers makes contact
with the suspect(s) or complaining citizen(s),
while the second officer takes a position a short
distance away to maintain a view of the entire
situation and "cover" the contact officer. No
assertion can be made on the basis of available
evidence that the use of this tactic caused the
police in Phoenix to use more force. However, the
research design included controls for some of the
characteristics of the arrest situation that might
lead officers to use contact and cover--suspect use
of force, violent offense, number of suspects and
bystanders, low visibility--but these
characteristics cannot explain the consistent
association between use of contact and cover and
all three measures of force. In addition, the
contact-and-cover tactic is intended to provide
officers with a tactical advantage should a
physical confrontation occur. That advantage, some
might assume, should reduce the amount of force
used by police and suspects, but no evidence was
found to support this assumption. A thorough
examination of all aspects of the use of contact
and cover, beginning with a more indepth review of
the data collected in this study, is recommended.

The third implication for policy stems from the
finding that the gender of the suspect and the
police officer directly affected the amount of
force used by the police. More force was used by
police when both officer and suspect were male than
when both officer and suspect were female or their
gender was unreported. However, male suspects were
not found to use more force against the police than
female suspects when all of the predictors were
considered. For most officers and researchers, this
is counterintuitive. Statistical controls for the
height and weight of officers and suspects, as well
as for the suspect's use of force, were included,
and these controls (or other factors not included
in the research) may account for the absence of an
effect for suspect sex on suspect use of force.
Since most arrests (1,059, or 67 percent) involved
a male officer and a male suspect, we interpreted
these findings to mean that arrests of female
suspects by female officers involved less force
than the typical arrest.

A fourth area of concern is use of restraints and
the possibility that Phoenix police officers are
using too little force when they make custody
arrests. In more than 20 percent of arrests, the
police officers asserted that they had used no
handcuffs or other restraints; in another 3 percent
of the arrests, this item was not filled in on the
officer survey. The available data do not provide a
basis for determining the soundness of the current
policy, which authorizes and encourages officers to
restrain any custody arrestee but requires them to
do so only with felons and belligerent suspects.
The concern is based on the seemingly high
frequency with which the discretion to not use
restraints is exercised.

The last issue raised by this research is the
generic and imprecise quality of some of the 12
categories of suspect resistance and officer
response that are central to the department's use-
of-force policies. Behavioral indicators for each
of these categories were developed, with
difficulty, and it is suspected that officers may
have similar problems determining whether, for
instance, flight in a 2,000-pound automobile
belongs in category 4, "defensive resistance," or
category 5, "active aggression." Similarly, the
policy groups together all weapons except chemical
agents and firearms, which are placed in separate
categories. We do not suggest the use of more than
six categories or the kind of detailed ranking
distinguishing weapon possession, threat, and use
that was helpful for this research. However, the
policy could be more clearly stated and the
relative rankings based more explicitly on the
relative severity of officer and suspect behaviors.
Finally, as new weapons and tactics become
available, the review of the continuum-of-force
categories is inevitable.

Research implications

We used a representative sample of police behavior,
developed a variety of measures of police and
suspect use of force, and employed explicit models
and appropriate multivariate statistical procedures
to assess the strength of individual predictors of
force. Previous research has not used these
standards, but future research could benefit from
using and improving on them.

The sampling was representative but not systematic.
We used one 2-week period in June 1994 and were
unable to discern if seasonal variation played a
role in the amount or distribution of force used.
Other sampling schemes need to be developed and
implemented that will provide a more formal
statistical basis from which to make inferences
about all arrests in a particular jurisdiction.

Data collection was anonymous, and understanding
the behavior of individual officers or suspects
over time was thus precluded. Anonymity also
complicated the matching of officer surveys and
suspect interviews. Federal protections of
research-subject confidentiality are strong, even
for police officers. Future research could attempt
to better integrate survey, interview, and official
records of police and suspect behavior.

The measures of force are improvements over simple
dichotomies of the past, but, as ordinal or
interval measures, they are simply illustrative. At
best, they should be early prototypes of
measurement models that reflect true scales or the
full extent of harm caused by different forceful
actions.

The data collection instrument included many items
thought to be important in the study of police use
of force. The length of the form was a burden to
participating police officers and reflected the
fact that the researchers could not match officer
responses to official records about the arrest, the
suspect's prior record, or the officer's career.
Improved data on weapon possession and use are
essential, and the sequencing of officer and
suspect behaviors should be a high priority in
future research.

The use of multivariate statistical models improved
the rigor of this study, but the available methods
were only a small subset of approaches to assessing
causal influences. Most police professionals and
researchers believe that community context is an
important consideration in how much force is used.
This research does not incorporate contextual
models to account for such influences.

Because this is only one study in one jurisdiction
at one point in time, the results may not
generalize to other jurisdictions and their
relationships between citizens and police. There is
no substitute for replication.

Conclusion

All prior assessments of police use of force that
employed a systematic sample of police behavior as
a foundation report that use of force is infrequent
and that many of the factors commonly thought to
influence it do not. Limitations in prior research
left the validity and reliability of those findings
uncertain. This research implemented a design to
overcome some of these limitations, and the
researchers reached the same conclusion: Using
three alternative measures of force, the survey of
adult custody arrests in Phoenix revealed that no
force, or only low levels of force, was used in a
large proportion of cases.

In addition, the results did not support the notion
that the race of officers or suspects directly or
indirectly affects the amount of force used in
adult custody arrests. The popular focus on racial
factors in use of force seems to be unsupported by
this study and other research evidence. Subsequent
research must be attentive to the low base rate for
use of force by police and the even lower base rate
by suspects.

On the other hand, official policy and training on
the use of force do not indicate an awareness of
the common absence of force. Recognition that force
is rare and, when used, varies along a continuum
has implications for law enforcement policy,
training, and street behavior.

============================================= 

Joel Garner, Research Director, the Joint Centers
for Justice Studies, was the principal investigator
for this study. John Buchanan, Commander, and
Richard Groeneveld, Lieutenant, Phoenix Police
Department, served as project director and deputy
project director, respectively. Co-principal
investigators included Thomas Schade and John
Hepburn of Arizona State University and Jeffrey
Fagan of Rutgers University.

============================================= 

This Research in Brief is based on the study
supported by award 92-IJ-CX-K028 from the National
Institute of Justice, the contributions of the
participating officers of the Phoenix Police
Department, and the assistance of the Maricopa
County Sheriff's Department.

The complete report is available through the
National Criminal Justice Reference Service on
interlibrary loan or, for a fee, photocopy
reproduction. Call 1-800-851-3420. Ask for NCJ
158614.

Dr. Garner discussed his work with an audience of
researchers and criminal justice professionals and
practitioners. A 60-minute videotape, Understanding
Use of Force, is also available for $19 ($24 in
Canada and other foreign countries). Ask for NCJ
159739.

============================================= 

Points of view expressed in this report are those
of the authors and do not necessarily represent the
official position or policies of the Phoenix Police
Department, the Maricopa County Sheriff's
Department, or the U.S. Department of Justice.

The National Institute of Justice is a component of
the Office of Justice Programs, which also includes
the Bureau of Justice Assistance, Bureau of Justice
Statistics, Office of Juvenile Justice and
Delinquency Prevention, and the Office for Victims
of Crime.

NCJ 158614