Bad Cops: A Study of Career-Ending Misconduct Among New York City Police Officers

The author(s) shown below used Federal funds provided by the U.S.

Department of Justice and prepared the following final report:

Document Title: Bad Cops: A Study of Career-Ending

Misconduct Among New York City Police

Officers

Author(s): James J. Fyfe ; Robert Kane

Document No.: 215795

Date Received: September 2006

Award Number: 96-IJ-CX-0053

This report has not been published by the U.S. Department of Justice.

To provide better customer service, NCJRS has made this Federally-

funded grant final report available electronically in addition to

traditional paper copies.

Opinions or points of view expressed are those

of the author(s) and do not necessarily reflect

the official position or policies of the U.S.

Department of Justice.

Bad Cops:

A Study of Career-Ending Misconduct

Among

New York City Police Officers

James J. Fyfe

John Jay College of Criminal Justice

and

New York City Police Department

Robert Kane

American University

Final Version

Submitted to the United States Department of Justice, National Institute of Justice

February 2005

This project was supported by Grant No. 1996-IJ-CX-0053 awarded by the National

Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of

views in this document are those of the authors and do not necessarily represent the

official position or policies of the U.S. Department of Justice.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Abstract

This is a study of New York City police officers whose careers ended in

dismissal, termination, or forced resignations or retirements for reasons of misconduct.

The research compares the personal and career histories of all 1,543 officers who were

involuntarily separated from the New York City Police Department (NYPD) for cause

during 1975-96 with a randomly selected sample of their Police Academy classmates

who have served honorably.

The study uses confidential NYPD files as its major data sources. It employs

bivariate techniques to test 37 hypotheses and sub-hypotheses suggested by the literature,

two expert advisory committees, and several focus groups of NYPD personnel. In

addition, the research employed appropriate multivariate techniques (Principal

Component Analysis; Logistic Regression Analysis) which, with some exceptions,

generally supported bivariate findings.

Key findings of the research include:

- Traditional definitions of police misconduct, especially Apolice

corruption,@ are imprecise. In the past, police scholars have classified

acts of police misconduct as Apolice corruption,@ Apolice brutality,@ and

Adrug-related misconduct.@ We found, however, that these classifications

are not mutually exclusive, and that determining whether profit-motivated

criminality by police officers involved job-related police corruption

frequently is impossible.

- Pre-employment history matters. Officers whose life histories include

records of arrest, traffic violations, and failure in other jobs are more

likely than other officers to be involuntarily separated from the NYPD.

- Education and training matter. Officers who hold associate or higher

degrees are less likely than those who do not to be involuntarily separated.

Those who do well in the Police Academy=s recruit training program are

less likely than marginal recruits to be separated as unsatisfactory

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



probationers. Subsequently, they also are less likely to be involuntarily

separated for cause after successful completion of their probationary

periods.

- Diversity matters. As the NYPD has become more diverse, it has become

better behaved. We found a very strong inverse correlation (r=-.71; r

.50)

between the percentage of white male NYPD officers and the

department=s annual rate of involuntary separations.

- Race still matters, but apparently only for black officers: As the

representation of Hispanic and Asian officers in the NYPD has increased,

their involuntary separation rates have decreased and become virtually

indistinguishable from those of white officers. Black officers=

representation in the NYPD remained relatively flat during the years

studied. Black officers= involuntary separation rates have also decreased,

but remain higher than those for other racial groups.

The study concludes with recommendations for policy formulation and further

research.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



iii

New York City Police Department Staff

Beverly L Anderson

George Ayala, Jr.

Kettly Barthelemy

Claudia Bellavia

Diane A. Biondo

Betty Braxton-Omaro

Yolanda Caraballo

Karen A. Carter

Jean W. Chen

Patricia A. Clements

Mildred E. Collins

Marilyn N. Coston

Carol A. Ficarella

Patricia Ann Francis

Awilda Fraguada

Barbara Freeman

Robert T. Gasperetti

Dawn L. Gonzales

Clara Hemmingway

Bettie Hickson

Chandra Jairam

Gene R. Johnson

Gerard J. Joyce

Troy Kirschner

Robert F. Larke

Kristine T. Lawless

Terril V. Lesane

Gloria Lynch

John Marsico

Catherine Montaruli

Michelle Moultrie

Gerald L. Neidick

Audrey Phillips

Sharece S. Phillips

Lillian M. Santamaria

Vilma L. Santiago

Todd E. Singleton

Ernest S. Takacs

John Totaro

Dorothy Wallace

Wendy E. Watson

Trita B. Williams

Police Advisory Committee

Walter P. Connery Michael Julian

Attorney at Law Madison Square Garden

Centerport NY Corporation

Henry DeGeneste Jerome H. Skolnick

Prudential-Bache New York University

School of Law

Methodological Advisory Committee

Michael Maxfield Joan McCord

Rutgers University Temple University

School of Criminal Justice Department of Criminal Justice

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Table of Contents

Page

I. Introduction 1

II. Defining and Identifying Police Misconduct 8

III. Prior Research on Police Misconduct 21

IV. The Research Setting 53

V. The NYPD, 1975-1996 78

VI. Research Methods 97

VII. Research Questions 112

VIII. Hypothesis Testing 127

IX. Multivariate Analysis 272

X. Discussion 288

Sources 295

Cases Cited 309

Appendix I

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



List of Tables

Page

VIII-1 Principal Component Analysis of New York City

Crime Rates 142

VIII-2: Initial Hierarchical Poisson Model Estimates of

Within- and Between-Precinct Police Misconduct in

New York City 144

VIII-3 Hierarchical Poisson Estimates of the Influence of

Homicide, per Capita Income, and Percent Minority

Population on Within- Precinct Police Misconduct in

New York City (No Control Variables Included) 147

VIII-4 Hierarchical Poisson Estimates of the Influence of

Index Crime, per Capita Income, and Percent Minority

Population on Within- Precinct Police Misconduct in

New York City (Control Variables Included) 148

IX-1 Comparisons Between Study and Comparison Officers

Along Predictor Variables of Interest 273

IX-2 Principal Component Analysis of Criminal History

Variables 277

IX-3 Principal Component Analysis of Work History

Variables 278

IX-4 Principal Component Analysis of Social Condition

Variables 279

IX-5 Logistic Regression Estimates and Odds Ratios

Predicting Police Misconduct 282

List of Figures

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Figure Title Page

II-1 NYPD Misconduct Codes 9

IV-1 NYPD Organization Chart 58

IV-2 Characteristics of NYPD Precincts 62

IV-3 Integrity Test Failures, 1999-2001 70

IV-4 EDIT Program Operations, 1999-2001 72

V-1 Police Officer Appointments to the NYPD, 1946-1996 88

V-2 Precinct Officers and Supervisors, 1975-1996 89

VIII-1 Primary Charges against Study Officers 128

VIII-2 All Specifications against Study Officers 130

VIII-3 Profit-Motivated Charges Against Study Officers 121

VIII-4 On-Duty Abuse Charges Against Study Officers by Year 151

VIII-5 Mean Annual Rate per 1,000 Officers of Separations

for On-Duty Abuse, by Mayor 153

VIII-6 Mean Annual Fatal Shooting Rates per 1,000 Officers

in Ten Largest U.S. Cities, 1990-2000 155

VIII-7 Mean Annual Civil Rights Complaint Rates per 1,000

Officers in Ten Largest U.S. Cities, 1985-1990 157

VIII-8 Officers Hired 1965-1995 and Class Separation Rates

During 1975-1995 162

VIII-9 Years in Service Prior to Separation, Study Officers

Appointed 1979-1996 164

List of Figures (continued)

Figure Title Page

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



vii

VIII-10 1979-95 Classes' Cumulative Percent Involuntary

Separations, Career Years 1-18 166

VIII-11 Annual Class Size (in Thousands) and Cumulative

Involuntary Percent Separations, 1979-1995 Classes 168

VIII-12 Annual Percentages of Non-White Officers and

Involuntary Separation Rates per 1,000 Officers,

1986-96 170

VIII-13 Anual Involuntary Separation Rates by Race, 1986-96 173

VIII-14 Annual Involuntary Separation Rates by Gender, 1986-96 176

VIII-15 Probationary Termination Rates by Gender, 1987-95 178

VIII-16 Rates of Separation for Profit-Motivated Offenses per

1000 Officers by Gender, 1986-96 180

VIII-17 Ten Most Frequent Primary Charges against Officers

Separated for Profit-Motivated Misconduct, 1975-96 182

VIII-18 Separations for Profit-Motivated Offense by Gender,

1975-96 183

VIII-19 Rates of Separations on Charges Including On-Duty

Abuse per 1,000 Officers by Gender, 1986-96 187

VIII-20 Rate of Separations for Drug Test Failures or Refusals

per 1,000 Officers by Gender, 1986-96 189

VIII-21 Rates of Separation for Off-Duty Violent and Public

Order Offenses per 1,000 Officers by Gender, 1986-96 191

VIII-22 Rates of Separation for Administrative Rule Violations

per 1,000 Officers by Gender, 1986-96 192

VIII-23 Probationary Separation Rates per 100 Probationary

Officers by Race, 1987-95 196

List of Figures (continued)

Figure Title Page

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



viii

VIII-24 Race of Control Officers and Officers Separated for

Profit-Motivated Offenses, 1975-96 198

VIII-25 Separations for Bribery and Profit-Motivated Offenses,

1975-96 201

VIII-26 Rates of Separation for Profit-Motivated Offenses

per 1,000 Officers by Race, 1990-96 202

VIII-27 Percentage of Officers in Supervisory Ranks by

Race, 1990-96 205

VIII-28 Mean Annual Separation Rates for Profit-Motivated

Misconduct by Rank and Race, 1990-96 207

VIII-29 Commands of Officers Separated for Profit-Motivated

Misconduct and Matching Control Officers by Race,

1975-96 208

VIII-30 Separations for On-Duty Abuse by Race, 1975-96 212

VIII-31 Rates of Separation per 1,000 Officers in Which

Primary Charge was On-Duty Abuse,1986-96, by Race 215

VIII-32 Separations for Non-Line of Duty Offenses by Race,

1975-96 216

VIII-33 Separation Rates per 1,000 Officers for Non-Line of

Duty Offenses by Race, 1986-96 217

VIII-34 Separation Rates per 1,000 Officers for Drug Test

Failures or Refusals by Race, 1986-96 219

VIII-35 Separations For Administrative Rules Violations

by Race, 1975-96 221

VIII-36 Separation Rates per 1,000 Officers for Administrative

Rules Violations by Race, 1986-96 223

List of Figures (continued)

Figure Title Page

VIII-37 Mean Annual Separation Rates per 1,000

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Police Officers/Detectives for Administrative Rules

Violations by Race, 1990-96 225

VIII-38 Commands of Officers Separated for Administrative

Rule-Breaking and Matched Control Officers by Race,

1975-96 227

VIII-39 Separation Rate per 10,000 Officers and

Percentage White Male Officers, 1986-96 232

VIII-40 Separated and Control Officers' Age at Appointment,

1975-96 235

VIII-41 Separated and Control Officers' Age at Appointment,

Profit-Motivated Offenses, 1975-96 236

VIII-42 Separated and Control Officers' Age at Appointment,

On-Duty Abuse, 1975-96 237

VIII-43 Separated and Control Officers' Age at Appointment,

Drug Sales, 1975-96 239

VIII-44 Separated and Control Officers' Age at Appointment,

Non-Line of Duty Offenses, 1975-96 240

VIII-45 Separated and Control Officers' Age at Appointment,

Drug Test Failures and Refusals, 1975-96 241

VIII-46 Separated and Control Officers' Age at Appointment,

Administrative Rules Violations, 1975-96 242

VIII-47 Separated and Control Officers' Prior Employment

and Pre-Entry Experiences, 1975-96 244

VIII-48 Separated and Control Officers' Military Experience,

1975-96 245

List of Figures (continued)

Figure Title Page

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



VIII-49 Percentage of Study and Control Officers with

Military Experience, by Study Officer SeparationYear 249

VIII-50 Percentage of Study and Control Officers with Military

Experience, by Year Officers Appointed to NYPD 251

VIII-51 Year in Which Study and Control Veterans Entered

the Military 252

VIII-52 Military Experiences of Study and Control Officers 254

VIII-53 Separated and Control Officers' Educational Levels

at Entry to the NYPD 258

VIII-54 Percentage of Study and Control Officers with Five+

Years Service in Supervisory or Command Ranks, by

Educational Level at Entry into NYPD 260

VIII-55 Study and Control Officers' Academic and Physical Scores

in Police Academy (Excludes Separated Probationers) 262

VIII-56 Study and Control Officers' Sick, Late, and Disciplinary

Records in Police Academy (Excludes Separated

Probationers) 264

VIII-57 Frequency of Study and Control Officers' Sick and Late

Reports, and Disciplinary Records in Police Academy

(Excludes Separated Probationers) 265

VIII-58 Study and Control Officers' Mean Scores on Police

Academy Instructional Staff Ratings (Excludes Separated

Probationers) 267

VIII-59 Annual Rates of Complaints against Study and Control

Officers Prior to Study Officers=s Separation 270

X-1 Percent Change in Representation in NYPD Ranks 291

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Appendices

Appendix Title

I Handbook for Data Collection Staff

I-A Study Officer Data Collection Instrument

I-B Comparison Officer Data Collection Instrument

I-C Staff Coder Identification Numbers

I-D Command Codes

I-E Military Ranks

I-F Offense Classifications

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xii

EXECUTIVE SUMMARY

This is a study of the differences between New York City police officers whose careers

ended in involuntary separations for cause and their colleagues who have served honorably. The

study addresses many issues and reaches many conclusions. Key findings include:

- Traditional definitions of police misconduct, especially Apolice corruption,@ are

imprecise. In the past, police scholars have classified acts of police misconduct

as Apolice corruption,@ Apolice brutality,@ and Adrug-related misconduct.@ We

found, however, that these classifications are not mutually exclusive, and that

determining whether profit-motivated criminality by police officers involved job-

related police corruption frequently is impossible.

- Involuntary separations are rare. Separations for cause from the New York City

Police Department (through dismissals, terminations, and forced resignations and

retirements) are infrequent events. Only two percent of the officers employed by

the NYPD during the 22 years (1975-96) we studied were involuntarily separated

from the department.

- Pre-employment history matters. Officers whose life histories include records of

arrest, traffic violations, and failure in other jobs are more likely than other

officers to be involuntarily separated from the NYPD.

- Education and training matter. Officers who hold associate or higher degrees are

less likely than those who do not to be involuntarily separated. Those who do

well in the Police Academy=s recruit training program are less likely than

marginal recruits to be separated as unsatisfactory probationers. Subsequently,

they also are less likely to be involuntarily separated for cause after successful

completion of their probationary periods.

- Diversity matters. As the NYPD has become more diverse, it has become better

behaved. We found a very strong inverse correlation (r=-.71; r

.50) between the

percentage of white male NYPD officers and the department=s annual rate of

involuntary separations.

- Race still matters, but apparently only for black officers: As the representation

of Hispanic and Asian officers in the NYPD has increased, their involuntary

separation rates have decreased and become virtually indistinguishable from those

of white officers. Black officers= representation in the NYPD remained

relatively flat during the years studied. Black officers= involuntary separation

rates have also decreased, but remain higher than those for other racial groups.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xiii

The study reaches these conclusions after examining the life and career histories of all

1,543 officers who were dismissed or forced to leave the NYPD for reasons of misconduct

during 1975-1996 (excluding, e.g., those who failed their recruit training program) and compares

them to a stratified sample of their Police Academy classmates who were selected randomly on

the basis of their NYPD employee, or tax registry, numbers. During this period, the NYPD

averaged well over 30,000 uniformed officers, and its personnel strength ranged between 21,500

in the late 1970s to 38,000 at the end of the study. In all, the NYPD employed about 78,000

different individuals as police officers during the years we studied. On June 30,1975, the

department employed more than 32,000 officers. Between then and the end of 1996, it hired (or

absorbed from the former Housing and Transit Authority police agencies more than 45,000

additional officers. Thus, our population of 1,543 officers separated for cause represents about

two percent of all officers employed by the NYPD during 1975-96.

The data for our study and control officers consist of information compiled in NYPD

personnel folders and disciplinary records. These include each officer=s original application, the

PA-15, a very detailed life history that is prepared by police officer candidates themselves, and

that is then subjected to extensive pre-employment investigation by the NYPD=s Applicant

Processing Division. As officers= careers proceed, notable events are recorded in their

personnel folder, including their recruit school performance; disciplinary histories; recognition

of outstanding performance and commendatory letters from both NYPD officials and the public;

vehicle accidents; injuries and sickness; changes in their social status, address, and educational

achievement; transfers; promotions; temporary assignments; and supervisors= performance

evaluations. We had access to the Central Personnel Index, an automated data base that

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xiv

supplements the personnel file, and that summarizes noteworthy career events (commendations;

complaints; line of duty injuries; sick leave absences; designation as chronically; vehicle

accidents; internal investigations and their dispositions. We checked the Personnel Orders

Section=s records, which contain the official history of each officer=s assignments and

addresses; and also had access to the Police Academy=s records of officers= performance in

recruit training.

The study was intended to describe the circumstances that led to involuntary separations

from the NYPD and to identify differences between the involuntary separated officers (the

Astudy officers@) and an equal, randomly selected number of officers who had entered the

NYPD in the same Police Academy classes as the study officers (the Acontrol officers@). We

chose this method as the best available way to determine whether characteristics and experiences

of involuntarily separated officers were different from those of their colleagues. To do this, we

generated and tested a lengthy series of hypotheses and conducted multivariate analyses

designed to identify factors that distinguished between the study and control officers.

This process led to a discovery about traditional definitions of police misconduct. We

found that it was impossible to define a typology of police misconduct that included Apolice

corruption@ as a classification that could clearly be distinguished from other categories of

wrongdoing. To be sure, many of the officers we studied were corrupt, but the conduct for

which they were separated from the NYPD included a wide variety of money-making

misconduct that had connections of varying strength to their employment as police officers.

When officers accept bribes to refrain from enforcing the law, they unambiguously engage in

what most of us would regard as police corruption. But it is less clear whether officers who

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



perform robberies or burglaries, shoplift, sell drugs, or engage in welfare or insurance fraud

during their off-duty time are engaging in a variety of police corruption. Thus, we created a

category of wrongdoing called profit-motivated misconduct, and recommend that future scholars

and administrators rethink the notion of Apolice corruption.@ Often, in these changing times,

police corruption is not as easy to define as we formerly may have been believed.

Those who have read early drafts of this report have generally expressed surprise at the

small number of officers separated for charges that included brutality and other abusive conduct.

Over 22 years, 119 officers were separated on charges that included some form of on-duty

abuse. Only 37 of these officers B 1.7 per year B were separated in matters in which the primary,

or only charge, against them was on-duty abuse. We believe that there are two explanations for

this. First, despite some spectacular and widely publicized acts of brutality, the NYPD has long

been one of the most restrained police agencies in the country. NYPD officers are less likely

than officers in virtually all big police departments to fire their weapons at citizens, and the data

we reviewed showed that they were less likely than most to be subjects of civil rights complaints

to the U.S. Justice Department.

This, we believe, is because the NYPD historically has held its officers to an extremely

high degree of accountability. Its reviews of police shootings, use of force, and citizens=

complaints are extensive and objective. In the course of our work, we also found that the NYPD

engaged in a wide variety of pro-active strategies designed to deter and detect wrongdoing.

Like many other agencies, the NYPD runs an early warning system, and regularly reviews and

monitors officers who seem to experience their work differently than do their colleagues.

Officers whose histories are marked by repeated complaints, vehicle accidents, line of duty

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xvi

injuries, sick leave, and arrests for such contempt of cop offenses as resisting arrest and

obstructing governmental administration are routinely assigned to close supervision, retraining in

a special and individually tailored program, and monitoring by the department=s administration.

But the NYPD does other things that are not so common. The NYPD=s Internal Affairs

Bureau encourages officers to call anonymously to report apparent wrongdoing by their

colleagues. IAB uses this and other information to conduct hundreds of Integrity Tests B stings

B on personnel who are suspected of wrongdoing. Some of these tests replicate the

circumstances in which it has been alleged that officers have mistreated people by, for example,

presenting a suspect officer with a staged, on-street, opportunity to mistreat a person of color or

member of some protected class; by giving an officer an opportunity to improperly pocket cash.

Integrity tests also are conducted to determine whether ranking officers properly accept and

process complaints about the officers who work for them. On occasion, the tests are very

elaborate, involving lengthy operations designed to arrest criminals (gamblers; drug dealers) who

are believed to be in corrupt relationships with officers for the purpose of turning them in order

to gain evidence to prosecute crooked officers. These proactive investigative steps are unusual B

perhaps even unique to the NYPD B and are worthy of study by both scholars and police

administrators. They should also be taken into account in attempts to generalize our findings to

other settings.

The second reason that the number of separations for brutality is lower than what one

might expect is the difficulty of proving these cases. Unless evidence of excessive force is

unambiguous B as in the taped beating of Rodney King and the injuries sustained by Abner

Louima when police sodomized him with a stick B it is very difficult to show that the force used

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xvii

by a police officer was more than was reasonably necessary to subdue a resisting person. For

this reason, the proactive work of the NYPD B its early warning system and its system of tests

in which suspect officers are presented with opportunities to be abusive B are such an important

part of its work.

We also found that the NYPD=s program of drug testing its officers is more extensive

than most. Every year, on previously unannounced days, a random sample of 20 percent of the

members of every NYPD unit appears at work and is directed to proceed immediately to the

Medical Unit in order to undergo a Dole Test. Test failure or refusal to take such a test results

in immediate suspension and, almost invariably leads to dismissal. This program, combined

with an extensive for cause testing program, in which personnel may be ordered to undergo

testing on the suspicion that they are abusing controlled substances, may result in a higher

percentage of drug-related terminations than is true of most agencies.

We tested most of our hypotheses by comparing the percentage of study officers who

shared a trait or experience with the comparable percentage among the control officers. Our

results on each were as follows:

HYPOTHESIS RESULTS

H1: More involuntary separations are attributable to CONFIRMED

profit-motivated corruption than to brutality and other

non-profit abuses of citizens.

H2: MODIFIED TO: Variations in community structure CONFIRMED

(i.e., per capita income, percentage minority population)

and public crime (homicide, FBI index crimes)

will predict variations in police misconduct within police

precincts over time.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xviii

H3: The rate of involuntarily separations for brutality REJECTED: KOCH

and other non-profit abuse of citizens was greater during ADMINISTRATION

the administration of David Dinkins than during either the HAD HIGHEST

Koch or Giuliani administrations. RATES

H4: Rates of involuntary separation for reasons other REJECTED: KOCH

than brutality and non-profit abuse were greater during ADMINISTRATION

the administration of Rudolph Giuliani than during either HAD HIGHEST

the Koch or Dinkins administrations. RATES

H5: Rates of involuntary separation for corruption, REJECTED

brutality, and other forms of misconduct are positively

associated with the size of Police Academy recruit training

cohorts.

H5a: The rate at which new officers are REJECTED

involuntarily separated as unsatisfactory

probationers is inversely correlated with the

size of Police Academy recruit training cohorts.

H6: Rates of involuntary separation for UNTESTABLE:

corruption, brutality, and other forms of misconduct INSUFFICIENT

are inversely associated with the rigor and intensity DATA

of recruit training, as measured by:

H6a: numbers of hours of training; UNTESTABLE

H6b: rates of involuntary separation for UNTESTABLE

academic and physical failure, and for

disciplinary reasons.

H7: Rates of involuntary separation for corruption, CONFIRMED

brutality, and other police misconduct are inversely

associated over time with the percentage of non-white

officers in the department.

H8: Rates of involuntary separation for corruption, UNTESTABLE:

brutality, and other police misconduct are positively INSUFFICIENT

Our test of this hypothesis was less rigorous than we would have liked.

Because measurement issues made it impossible to include in the analysis all the study

officers, we limited analysis to officers who were both hired and fired during 1979-1995.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xix

associated over time with the percentage of officers DATA

who reside outside New York City.

H9: Female officers' involuntary separation rate is REJECTED

positively associated over time with the percentage of

females in the department.

H10: The rate of involuntary separations of CONFIRMED

probationers is greater among female officers than

among males.

H11: The rate of involuntary separations for CONFIRMED AS TO

corruption is greater among male officers than among BRIBERY; REJECTED

female officers. ASTOALLPROFIT-

MOTIVATED

MISCONDUCT

H12: The rate of involuntary separations for brutality CONFIRMED

and other non-profit abuses is greater among male officers

than among female officers.

H13: The rate of involuntary separations for drug test CONFIRMED

failures and refusals is greater among female officers than

among male officers.

H14: The rate of involuntary separations for non-line of REJECTED: FEMALE

duty criminal conduct (e.g., off-duty thefts and fraud; drug RATE HIGHER

crimes) is greater among male officers than among female

officers.

H15: The rate of involuntary separations for CONFIRMED

administrative rule breaking is greater among female

officers than among male officers.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



H16: The rate of involuntary separations of CONFIRMED AS TO

probationers is greater among non-white officers than BLACKS; REJECTED

among white officers. AS TO HISPANICS

AND OTHERS

H17: The rate of involuntary separations for corruption CONFIRMED AS TO

is greater among non-white officers than among white BLACKS; REJECTED

officers. ASTOHISPANICS

AND OTHERS

With the exception of separations for brutality (where we confirmed H19, that

white officers= rate would be highest), the pattern in all our analyses involving officers=

race was consistent. Over time, the separation rates of Hispanic officers have decreased

so that they have become near indistinguishable from the rates for Whites. Because of

their very low representation in the NYPD, the rates for AOthers@ (mostly Asian officers)

were statistically meaningless in the early years covered by our study. As the

representation of Asians in the NYPD has grown, their separation rates have recently

been much like those for Whites and Hispanics. The rates for Black officers, which

started out much higher than those of other groups, also have decreased, but remain

considerably higher than those for the other three major racial groups.

H18: The discrepancy between white and non-white REJECTED AS TO

officers' rates of involuntary separation for corruption is RANK; UNTESTABLE

accounted for by differential patterns of assignment and rank. AS TO ASSIGNMENT

H19: The rate of involuntary separations for brutality and REJECTED; WHITE

other non-profit abuses is greater among white officers than RATES LOWER THAN

among non-white officers. OR EQUAL TO BLACK

AND HISPANIC RATES

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxi

H20: The rate of involuntary separations for non-line of CONFIRMED, WITH

duty criminal conduct (e.g., off-duty thefts and fraud; drug BLACK RATES

crimes) is greater among non-white officers than among HIGHER THAN ALL

white officers. OTHERS

H21: The rate of involuntary separations for drug test CONFIRMED AS TO

failures and refusals is greater among non-white officers BLACKS; REJECTED

than among white officers. AS TO HISPANICS AND

OTHERS

H22: The rate of involuntary separations for CONFIRMED , WITH

administrative rule breaking is greater among non-white BLACK RATES FAR

officers than among white officers. HIGHER THAN ALL

OTHERS

H24: The rate of involuntary separations of CONFIRMED

probationers is greater among officers who were less than

22 years old when appointed than among officers who were

22 or more years old when appointed.

H25: The rate of involuntary separations for corruption REJECTED

is greater among officers who were less than 22 years old

when appointed is greater than among officers who were

22 or more years old when appointed.

H26: The rate of involuntary separations for brutality REJECTED

and other non-profit abuses is greater among officers who

were less than 22 years old when appointed than among

officers who were 22 or more years old when appointed.

H27: The rate of involuntary separations for non-line of REJECTED

duty criminal conduct (e.g., off-duty thefts and fraud; drug

crimes) is greater among officers who were less than 22

years old when appointed than among officers who were 22

or more years old when appointed.

H28: The rate of involuntary separations for drug test REJECTED

failures and refusals is greater among officers who were

less than 22 years old when appointed than among officers

who were 22 or more years old when appointed.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxii

H29: The rate of involuntary separations for administrative REJECTED

rule breaking is greater among officers who were less than

22 years old when appointed than among officers who were

22 or more years old when appointed.

H30: Rates of involuntary separation for all types of police

misconduct are higher for the following categories of

officers than for other officers:

H30a: officers who were dismissed by previous CONFIRMED

employers.

H30b: officers whose prior employers gave police CONFIRMED

investigators derogatory information about them.

Readers may wonder how we confirmed H24, that officers hired at ages 20-21

were more likely than others to be separated, when the data did not confirm any of the

offense-specific hypotheses (H25-H29) to the same effect. The answer is that we found a

consistent pattern in which younger recruits were somewhat more likely than others to

end their careers in separations for profit-motivated offenses; on-duty abuse; non-line of

duty offenses; and administrative rule violations. Although these were not statistically

significant, they did make a difference in the aggregate.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxiii

H31: Officers with military records that include no REJECTED; MILITARY

discipline are less likely than officers without military VETERANS MORE

experience or with military disciplinary histories to be LIKELY THAN

nvoluntarily separated. NON-VETERANS TO BE

DISCIPLINED

H32: Rates of involuntary separation for all types of police CONFIRMED

misconduct are inversely associated with officers' years of

education at entry into the NYPD.

H33: Officers= years of education at entry into the NYPD CONFIRMED

is positively associated with movement into supervisory

and management ranks.

H34: Officers= educational attainment after entry into the UNTESTABLE:

NYPD is positively associated with movement into supervisory INSUFFICIENT

and management ranks. DATA

. There was considerable within variation in separation rates among military

veterans. Former Marines were more likely that veterans of other services to be

separated. So were veterans who served more than a four-year military enlistment; those

who were disciplined in the military; those who joined the military immediately after the

1974 abolition of the draft; and those who had not advanced above the ranks of corporal

or seaman. Navy and Air Force veterans and former military officers had very low

separation rates. In any event, much of the association between military service and

separation washed out during our multivariate analysis, suggesting that it was

confounded by other variables (e.g., age at appointment; race; level of education,

employment history).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxiv

H35: Rates of involuntary separation for all types of police CONFIRMED

misconduct are inversely associated with officers' Police

Academy academic averages.

H36: Rates of involuntary separation for all types of

misconduct are positively associated with the following

variables (all standardized by rates per year of service):

H36a: prior citizens' complaints: UNTESTABLE:

INSUFFICIENT

DATA

H36b: prior supervisory disciplinary actions; UNTESTABLE:

INSUFFICIENT

DATA

H36c: prior Abelow standards@ evaluations on UNTESTABLE:

performance evaluations; INSUFFICIENT

DATA

H36d: prior line of duty civil suits; UNTESTABLE:

INSUFFICIENT

DATA

For purposes of this analysis, we excluded officers who were separated while on

probation. We did this on grounds that, while it was not misconduct, their poor Academy

performance might have played a role in their separations. Thus, this analysis included

only officers who were already tenured and who were separated as the result of decisions

made by officials who would not take into account their recruit school performance.

Study officers also performed more poorly than control officers on Police Academy

physical examinations. During their recruit training, they also were more often sick;

injured, late, disciplined, and held back from graduating with their classes than were

study officers. They received fewer Aexcellent@ performance ratings; more Apoor@

performance ratings, and were deemed to have performed more poorly than study officers

in Law, Police Science, Social Science, and Physical Training. Despite all this, they

received higher Aoverall@evaluations (mean=2.80 on a 0-5 scale) than did study officers

(mean=2.58). Clearly, there is a need to bring the Police Academy=s subjective ratings

of recruits into line with objective measures of recruits= performance.

Because the origin of complaints is not always clear (e.g., did it originate with a

citizen or with a police official?), we were unable to test H36a and H36b as stated. We

were able to test and confirm the hypothesis that study officers were more likely than

control officers to have histories of prior complaints.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxv

H36e: prior line of duty injuries; UNTESTABLE:

INSUFFICIENT

DATA

H36f: prior line of duty vehicle accidents; UNTESTABLE:

INSUFFICIENT

DATA

H36g: prior designations as chronically sick. UNTESTABLE:

INSUFFICIENT

DATA

H37: Rates of involuntary separation for brutality UNTESTABLE:

and other abuses are positively associated with INSUFFICIENT

rates of departmental commendations. DATA

We also employed Principal Component Analysis (PCA) to extract three dimensions of

officers= histories:

- Criminal History, including violent and property crime arrests, juvenile

delinquency findings, misdemeanor arrests and convictions, public order arrests,

moving and parking violations, and driver=s license suspensions.

- Work History, including numbers of jobs, workman compensation claims, 30

day (or longer) periods of unemployment, jobs from which fired, work-related

disciplinary actions, and derogatory comments by prior employers.

- Social Condition, including officers= social circumstances at the time of their

appointment, including marital status (coded as a series of binary variables);

number of children; and whether officer was actively enrolled in school at time of

appointment.

The PCA identified subsets of these three dimensions comprised of closely related

variables. These were then entered into a Logistic Regression Analysis designed to find the

factors that most significantly distinguished between study and control officers. While

controlling for all other covariates, the strongest risk factors for termination were whether an

officer was black (the derived odds ratio showed that black officers were 3.27 times more likely

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxvi

than whites to be involuntarily separated; had an average of one or more complaints per year of

service (3.03 times as likely as officers with fewer complaints to be separated); and worked in

inspector precincts at the time of the incident leading to separation (2.48 times more likely than

officers assigned to staff units to be dismissed).

Other significant risk factors for police misconduct included Latino (B=.69), being

assigned to DI precincts (B=.81), captain precincts (B=.71), police academy/field training units

(B=.61), having criminal histories (B=.20), public order offense histories (B=.58), prior

employment disciplinary problems (B=.32), and prior employee reliability problems (B=.15). In

sum, minority-group officers had higher probabilities than white officers of incurring an

organizational response to known misconduct; officers assigned to posts that placed them in

regular contact with the public under relatively unsupervised conditions (i.e., precincts), or to

posts at times in their careers when they did not enjoy civil service protection (i.e., police

academy/training units) were at greater risk of being separated for misconduct; officers who

were officially recognized disciplinary problems during both their pre-police and NYPD

occupational tenures were at greater risk than others of being dismissed for police misconduct;

and officers who had an officially sanctioned history of deviance had higher odds than others of

being dismissed for misconduct. These multivariate findings largely support our bivariate results

(see Table IX-1).

The logistic regression model also found that length of service; holding an Associate or

Bachelor=s degree at appointment, and increased age at appointment weighed against

To assure that we did not confound prior departmental complaint history and the

events that may have led to officers= separation in this analysis, we treated officers who

had received complaints during careers that lasted less than one year as missing cases.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxvii

involuntary separation. In addition, officers whose fathers had served as NYPD officers were

significantly less likely (albeit, not substantially) than other officers to be dismissed for

misconduct, and officers who achieved a supervisory rank were less likely than line officers to

be dismissed for misconduct. These findings suggest that overall, academically well prepared

officers, those who were ambitious, and those with parental links to the NYPD were less likely

than other officers to either engage in, or be sanctioned for, occupational misconduct.

The logistic regression model also identified factors statistically not associated with

occupational deviance. Among these was military service. Our bivariate findings showed that

military service was significantly related to police misconduct. The multivariate findings did not

support this earlier result, however, suggesting that the bivariate relationship may have been

confounded by other factors that were not controlled. It is possible, for example, that while

considering the effects of age at appointment, level of education, and employment history,

military service may be an unimportant independent factor in the prediction of police

misconduct. In addition, officer sex, prior police service, and background investigator

recommendation were non-significant in the prediction of misconduct. It should be noted that

although the mayor at time of officer=s separation was included in the model as a control for the

effects of social and political climates in New York City, it was a non-significant predictor of

police misconduct.

Contrary to what we had expected and hypothesized, our bivariate analyses showed that,

except for cases involving on-duty abuse, female officers generally had higher separation rates

than their male colleagues. Like military service, however, gender washed out of our

multivariate analyses, suggesting that the bivariate relationships we found had been affected by

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxviii

other factors. Here, we also suspect that low absolute frequencies may have distorted the results

of our bivariate analyses to some degree.

In all, therefore, our work confirms the conventional wisdom regarding the police: young

officers who entered the police service with minimal educations, and records of prior

delinquency, criminality, and poor prior employment; who did not advance in the NYPD; who

worked in busy patrol assignments; and who accumulated histories of complaints were more

likely than others to have ended their careers in involuntary separation. Conversely, well-

educated officers with clean histories, perhaps including a family history in the NYPD, and who

worked their way to advanced rank were least likely to be involuntarily separated from the

agency.

Overriding all this is the race issue. We found that, despite the many years in which

African-American officers they have been a major presence in the department, and despite their

many contributions to it and to New York City, where discipline is concerned, they remain an

outgroup in the NYPD. They are far more likely than other officers to be involuntarily separated

and, we found in earlier work (Fyfe, et al., 1998), to be subjects of less severe discipline, as well.

It is hard to determine the extent to which this may be the result of discrimination, but our

analyses suggest that the disparity among black officers has resulted largely from separated

officers= involvement in situations (e.g., criminal or drug related behavior) in which

administrators= paths of action are clearly defined and leave little or no room for arbitrariness.

This finding is consistent with other work, both in an out of policing, that has found

strong associations between race and deviance, and official responses to the latter. We

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxix

recommend further study of this issue, and suggest research that examines the status of African-

American police officers in the NYPD and elsewhere.

We also reached far more heartening findings regarding gender, race, and officers=

misconduct:

- the greater the percentage of women in the department, the less likely women are

to be fired;

- the greater the percentage of men in the department, the more likely men are to be

fired;

- the greater the percentage of whites in the department, the more likely whites are

to be fired;

- the greater the percentage of Hispanics, Asians, and other non-white groups in the

department, the less likely they are to be fired.

Taken together, these findings produce what is for us a very important conclusion: as the

NYPD has become more diverse by gender and race, it has become significantly better

behaved.

Probably because the percentage of black officers in the NYPD has remained

relatively flat over the years studied, the relationship between it and black separation rate

is non-significant. The percentage of black officers in the NYPD ranged between 10.7%

in 1986 (n=2,799) to 11.6% in 1994 (4,293), then increased to 13.9% by 1996 (5,155).

This comparative jump over the last two years resulted largely from the merger of the

more diverse New York City Transit and Housing police departments into the NYPD.

During 1986-96, NYPD=s percentage Hispanic increased from 9.5% (2,505) to 16.7%

(6,205); percentage Asian/Other went from 0.6% (154) To 1.3% (478). Percentage

female increased from 9.5% (2,504) to 15% (5,684). Percentage white decreased from

79.2% (20,816) to 68.1% (25,240).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxx

There may be many reasons for this finding. It has long been claimed that domination of

policing by a single racial group has produced a solidarity that includes such unhealthy aspects

as a willingness among officers to tolerate misconduct by their colleagues. To the extent that

this may have existed in the NYPD, it may be breaking down as the NYPD has become more

diverse. Alternatively, some believe that, as the numbers and, presumably, the influence of

minority officers on the organization increase, their vulnerability to disciplinary arbitrariness has

decreased. The decreases in career ending misconduct may have to do with the Internal Affairs

Bureau=s greatly increased vigor, or with the increasing rigor of drug testing over the years

studied. The department also grew significantly so that, independent of their gender and race,

the newcomers may have wrought great changes in its culture.

These are subjects for future research, but our data seem to provide the best evidence to

date that diversity produces a healthier brand of policing. Thus, this finding gives researchers

what they want B an avenue for further study. More important, we think, it also gives

administrators reason to believe that their efforts to enhance diversity in the ranks have a highly

desirable product: an organization in which the percentage of officers of all genders and races

who disgrace themselves and their agency is significantly decreased. This finding may have

great significance not only for police administrators, but for all those who charged with running

organizations, in and out of the public sector.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxxi

Acknowledgments

The debts owed for this study are enormous. We are extremely grateful to former

National Institute of Justice Director Jeremy Travis, who funded our work and who, more

importantly, encouraged us and understood the difficulties we encountered while doing

it. Fyfe, especially, wants Travis to know how grateful we are, and how delighted he is

that Travis has just been named to be his boss when Fyfe returns to John Jay College,

where Travis has become president. Jeremy=s encouragement and patience have been

continued by NIJ=s current director, Sarah Hart, and we are very grateful to her, as well.

Dr. Sally Hillsman, the former Director of NIJ=s Office of Research and Evaluation,

championed this research from the time of its inception and through a major mid-course

shift in direction. Bryan Vila, the Chief of NIJ=s Crime Control and Prevention

Research Division, deserves thanks for his encouragement. Dr. Steven Edwards, our

original Project Monitor, was a model of support and encouragement. Steve was always

there when we needed him, making suggestions, but never pressing us. Maggie Heisler,

who succeeded Steve in these duties, followed his working style, for which we are

greatly indebted to her, as well. Paul McCauley of Indiana University of Pennsylvania

planted the seed that grew into this study. Robert Tillman of St. Johns University

supervised data collection, which was an enormous task done very well. Peter Jones of

Temple cleaned the data and made them suitable for analysis. We could not have done

the work without him.

We also are grateful to Michael Farrell, NYPD=s Deputy Commissioner for

Policy and Planning when we started this project and its Deputy Commissioner for

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxxii

Strategic Initiatives when we finished it, and to Chief of Internal Affairs Charles

Campisi, Chief of Housing Joanne Jaffe, former Deputy Commissioner for Trials Rae

Koshetz, and Deputy Chief Diana Pizzuti who reviewed drafts of our work, assuring that

it was accurate. Mike Farrell also was our point of entree to the NYPD. Mike and his

support for the work were keys to the virtually unlimited access we subsequently were

granted. His staff, especially Deputy Inspector Joseph Lovelock, Sergeant Vincent

Henry, and Carol Frazier, helped us to get the work off the ground, and we owe them

thanks. Assistant Commissioner Philip McGuire, an old friend who has been helping

Fyfe with research in the NYPD since the days of his doctoral dissertation, was a great

resource to us. There is a vast amount of information stashed here and there in the

hundreds of NYPD units. Invariably, Phil knows precisely where it is, what it means,

and how to get it. For more than a quarter-century, Phil Has been a tremendous asset to

the department, to the City, and to scholars. He is always eager to assist, to make

suggestions that facilitate research and assure its quality, and has played a central role in

scholarly and administrative analyses of the NYPD and its problems.

Mike Farrell enthusiastically endorsed our proposal, and passed it on to then-

Police Commissioner William Bratton, who agreed immediately that it was long overdue,

and who made sure that the New York City Police Department opened its doors to us.

His successor, Howard Safir, made the study a priority, and assured that we had

unlimited access to all the records, people, and resources we needed to do our work.

Raymond Kelly, the incumbent police commissioner, has continued this pattern.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxxiii

On a day-to-day level, we worked most closely with two people and their NYPD

units. Howard Safir called First Deputy Police Commissioner George Grasso a yes

lawyer, an attorney who finds ways to get things done when others might simply say no.

Commissioner Grasso and his staff certainly did that for us, and without them, this

project would not have been what we hoped. We are particularly indebted to him, to

Inspector Patrick Conry, to former Sergeant (now Firefighter) Michael Ansbro, and to

retired Detective John Totaro.

Most of our work was quartered in the Employee Management Division of the

Personnel Bureau. EMD=s Director, Arnold Wechsler, was a terrific asset. He helped

us organize the immense data collection and coding processes involved in our study,

helped us to identify and recruit project staff, resolved differences among the people who

worked with and for us, and navigated us through what would otherwise have been

nightmarish logistics. He, retired Deputy Inspector Francis Smith, and retired

Administrative Manager Elaine Glass acted as liaison between us and former Chief of

Personnel Michael Markman who, before dying too young, also helped us greatly. We

are grateful to them all.

We consulted with NYPD=s former Deputy Commissioner for Training James

O=Keefe and with his staff, especially Lt. Paul Kennedy of the Academy=s Research

Unit. Both were very helpful in pointing us to data related to our study officers=

performance in training which, as our research subsequently showed, certainly was

worthy of close analysis.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxxiv

The people who got this work done, who located, pulled, and pored through the

thousands of files we studied are the project=s New York City Police Department staff,

and are identified at the front of this report. We thank them one and all for their hard

work and good humor under pressure. We are especially grateful for their many

suggestions, which helped to both facilitate data collection and to increase the quality of

the information we compiled.

During the course of this project, we also worked with three focus groups, who

must necessarily remain anonymous. One was a group of patrol officers, selected from

across the city. A second was a group of police patrol sergeants, and the third were

patrol precinct commanders at the captains and deputy inspector ranks. We tapped their

expertise and perceptions to determine what they saw as the characteristics, pre-

employment and in-service experiences, and general conditions associated with police

disciplinary problems. Their insights were invaluable, and helped us to focus our

research and to develop and refine the data collection instruments and techniques we

employed.

We also assembled two advisory committees to work with us. The first were

people who possess great substantive knowledge of police and New York City and,

without compensation, they devoted considerable amounts of their very valuable time

and resources to help us focus and refine the study. They include former NYPD

Department Advocate Walter Connery, who also headed the Immigration and

Naturalization Service=s anti-corruption efforts; Henry DeGeneste of Prudential-Bache,

formerly Chief of the Port of New York and New Jersey Authority Police and a member

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxxv

of the Commission on Accreditation for Law Enforcement Agencies; former NYPD

Chief of Personnel Michael Julian of the Madison Square Garden Corporation; and

Professor Jerome Skolnick of New York University Law School, an internationally

renowned scholar of the police. Richard Koehler of John Jay College of Criminal

Justice, another former NYPD Chief of Personnel and ex-New York City Corrections

Commissioner, also served with us briefly, but had to leave the advisory committee

because of a conflict of interest. All five are good friends, whose combined expertise on

the matters we studied is without equal. We owe them all big time.

Our methodological advisory committee included Professor Michael Maxfield of

Rutgers University=s School of Criminal Justice. Mike is the co-author of

Maxfield/Babbie (1997), the standard criminal justice research text, and needs no

introduction to anyone who has studied human behavior over the last generation.

Professor Joan McCord was Fyfe=s colleague at Temple, and has built an international

reputation conducting longitudinal studies of deviant behavior. Both Mike and Joan

carefully reviewed the first draft of this report and made comments and suggestions that

have improved it immensely. Joan=s death earlier this year after a very brief illness, was

particularly painful. We with her loss, we lost a great friend and colleague, and the world

of social science lost a major contributor.

We owe special debts to two people. Dr. Carl Silver, professor emeritus at Drexel

University and Joan McCord=s husband, was with us from the beginning of this project.

His expertise in research design is reflected in the proposal for this project and, we are

both well aware, was a major factor in convincing NIJ=s reviewers, staff, and director

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



xxxvi

that this was a feasible and worthwhile endeavor. We hoped to work with Carl until this

study was completed and beyond, although we all knew that Carl was battling an

inevitably terminal illness long before this project got off the ground. Carl=s valor in his

long fight against death was an inspiration to us all and to Joan, as was his unremitting

intellectual vigor and enthusiasm for this, the last major research endeavor of his life.

We hope that our product is worthy of Carl.

Much of the data we studied were stored in Brooklyn in the Old Records facilities

of the Central Records Division. CRD=s Director Linda Scotti was of great assistance in

arranging for access to that facility. We are indebted to Linda, and were extremely

saddened by her premature death. We have all suffered a great loss, and we extend our

condolences to Linda=s family.

This study needs one last preface. Although the subjects of our research include

policing=s worst, we want to make plain that we do not view the bad cops we studied as

in any way representative of the NYPD or of American policing in general. Far more

representative, we believe, are the actions of New York cops on September 11, 2001. On

that terrible day and on every day since, George Grasso observed, New York cops

showed everybody that they were the kind of people those of us who have worked with

them always knew them to be.

Thus, we dedicate this study of bad cops and their good cop colleagues to Joan

McCord, Carl Silver, Linda Scotti, Mike Markman, and to the thousands upon thousands

of good cops who make the NYPD the world=s best police agency.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



I. INTRODUCTION

This is a study of police officers whose careers have ended in disgrace. Such bad

cops are a concern, most obviously because the most visible of them so severely hurt the

people they are paid to protect. Regardless of whatever commendable actions they may

have performed during their police careers, the Los Angeles officers who beat Rodney

King and the New York officers who beat and sodomized Abner Louima caused terrible

damage to both of these individuals. They changed the lives of their victims forever, and

will themselves forever be regarded as both symbols and causes of the gap of rage and

distrust that has too often characterized relationships between our police and our

communities of color.

FINANCIAL COSTS OF FAILED POLICE CAREERS

In addition to the damage they sometimes inflict on others and on the relationship

between the police and the community, there are other reasons to study police officers

whose careers end in disgrace. New police officers represent a significant commitment

on the part of the governmental entities that employ them. Typically, new officers enter

their departments only after long and elaborate screening processes designed to select the

best possible candidates in the fairest possible ways. Over the last generation, this

process has been the subject of litigation and legislation designed to enhance its validity

and to remove from it all traces of discrimination and arbitrariness. Thus, bad cops are

worth studying because they may tell us something about whether these ideals have been

achieved and whether the processes used to select officers do, in fact, predict satisfactory

police performance.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Further, the process of finding, selecting, training, and certifying new officers is

very costly. In most places, it involves the administration of a written examination, as

well as psychological, medical, physical performance and agility tests, and a background

and character investigation. Once this first set of steps is completed, new officers are

placed on the agency payroll, but they still are not working cops, who make a direct and

immediate contribution to the public welfare. Instead, they are likely to spend six months

or more in formal classroom training, followed by several additional months in a field

training experience in which they work under the tutelage of senior officers who grade

their performance in actual street situation. Thus, from the moment individuals apply to

become police officers until the completion of recruit training, they cost their employers

the expense of the screening process; the cost of either running and staffing an agency

training facility or contributing to the cost of sharing in a regional or state police

academy; and, usually, at least a year=s pay and benefits. During this period, new

officers= employers get little return on their investment, save perhaps the rookies=

occasional turn directing traffic or working at parades and other special events.

Once new officers= formal entry level training is completed, they may, as in New

York City, the jurisdiction we studied, continue on probationary status for as much as an

additional year.

During this period, they are expected to learn policing by doing it, and

to demonstrate that they are worthy of jumping from probation to tenure. Those who do

The NYPD=s probationary period for new officers is two years. New sergeants,

lieutenants, and captains hold their ranks on a probationary basis for one year.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



not make this move and are instead terminated as unsatisfactory probationers represent a

lost investment to the taxpayers.

PERSONAL COSTS OF FAILED POLICE CAREERS

Such officers= failures also cost themselves dearly. Typically, they have left

some other line of work to enter policing, often to much fanfare and attention from

friends and family. Their absence from their old jobs may mean that they have been

replaced by their former employers, and cannot simply resume life as it had been before.

In seeking new work, they are left with the difficult problem of explaining their

prematurely terminated sojourns into police work: for most private employers, a fired cop

may not be the most attractive job candidate. Personally, officers who wash out at this

stage of their police careers may be humiliated before those who have so recently wished

them well on their new endeavor, and may suffer losses of reputation and personal

confidence, as well as a stigma that lasts a lifetime. Fired cops have much in common

with disbarred lawyers, defrocked priests, and others who have violated special trusts.

POLICE TENURE: A ONE-SIDED MARRIAGE

Officers who successfully complete probationary status become, in effect, half of

one-sided marriages in which all the commitment rests with their partners, their

employers. The step from probation to civil service tenure carries with it many

guarantees, the most significant of which is that incumbents cannot be removed, or

divorced, by their employers except for cause. In most places, this means that dismissal

can follow only upon formal due process in which it is demonstrated that one has

committed egregious acts or omissions that violate criminal law or critical police rules

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



and regulations. In some of these places, police administrators= authority to dismiss for

even the most serious misconduct is subject to review by arbitrators who frequently

overturn or reduce severe disciplinary penalties such as dismissal (Fyfe, 1998). In all

places, simple mediocrity rarely is the cause for dismissal of police officers or any others

who have earned civil service tenure.

Thus, serious misconduct aside, a police agency

grant of tenure is a guarantee of career-long employment and a generous lifetime pension

thereafter.

The other side of this marriage faces no such commitment. At any time, police

officers are free to walk out, taking with them the benefits of all the testing and training

given them by their employers. Indeed, many New York City officers do precisely this,

leaving the police department for employment in either the city=s fire service or for more

lucrative employment in nearby suburban police agencies.

Thus, the process of screening, training, and socialization by which citizens

proceed from police applicants to consideration for tenure as officers is important

In making this observation, we recognize that we may be accused of criticizing a

term of employment in another discipline when we ourselves enjoy much the same

benefit: life in a tenure system. But we would also be the first to agree that, while both

police and university tenure systems serve the valuable function of insulating incumbents

from arbitrary dismissal because of unpopular actions, both systems may also serve to

insulate marginal performers from accountability. Further, the analogy between the two

systems is inexact. Universities commit to marry professors only after they have come to

know them far better than is true of police departments and officers. University processes

typically require that candidates for tenure demonstrate compatibility and satisfactory

performance for seven years before the award of tenure. Police departments typically

make their commitments only after a year or two. We believe that, in policing and the

academy as in the rest of life, the probability of marital success is positively associated

with length of courtship.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



because, when it ends in rejection by one partner or the other, it results in great costs that

cannot be recovered. When this process results in the award of tenure, it may begin an

expensive lifelong relationship that, while never overtly hostile, is nonetheless

unrewarding for both partners.

As suggested above, however, the worst consequences of bad marriages between

officers and police departments are those flowing from the serious misconduct that may

precede dismissals for cause or other, similar, involuntary separations. When the

misconduct precipitating these involuntary separations includes brutality or other abuses

of citizens, it hurts not only victims such as King and Louima, but also the notion of trust

in the police, and the credibility and reputation of a police agency and the political entity

of which it is a part. We have seen this repeatedly in recent decades. Their immediate

victims aside, the consequences of the King and Louima atrocities have been immense

and continuous. In both cases, the public was outraged. Faith and trust between the

police and the public, especially its most vulnerable inner-city communities of color, was

damaged or destroyed. This, in turn, has hindered police ability to work with citizens to

prevent crime and to gather the information necessary to solve crimes. In Los Angeles,

there even followed a riot that took 40 lives and caused millions in property losses. Both

incidents have tarnished the image of the United States itself: how, newspapers, critics,

and politicians abroad speculate, can the United States be all that it claims when, instead

of protecting its people, its police beat and sodomize them, and then lie about what they

have done?

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Serious, career-ending, police misconduct also has consequences for police

officers and their families. Deviant or otherwise, the fact that an individual is a police

officer is a central part B perhaps the central part B of his or her self-identification. Even

in non-criminal cases that do not present the risk of imprisonment, the loss of what New

York officers have long called the job is a major blow that, in our long exposure to police

work, frequently precedes future unemployability B who hires bad cops? We know of

cases in which officers= deterioration has even included suicide. Fired cops bring their

own troubles upon themselves and, it can be argued, get no more than they deserve. Still,

what they do get is severe, and should be counted as a cost to be avoided.

The fired cop=s family also suffers. Dismissal brings with it an immediate loss of

income and of the secure future promised by a civil service salary and subsequent

lifetime pension. When the conduct leading to dismissal is sufficiently egregious to

generate press coverage, humiliation and shame follow. Even when this does not occur,

neighbors and friends invariably learn that an officer has been dismissed, and their

relationships with cops= families change, almost invariably for the worse.

These bad marriages B involving police officers who are fired or otherwise forced

to leave police employment B are the focus of this study. This work examines the pre-

employment and career histories of the 1,543 New York City police officers who, during

1975-1996, were fired or forced to resign or retire because of their involvement in serious

misconduct. Our study compares these officers to a random sample of their colleagues,

stratified by the date on which these officers became probationary officers for the

purpose of identifying factors that may distinguish disgraced officers from the great

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



number of their colleagues who did not deviate from what is expected or, as one of our

advisors suggested, who at least were not identified as deviants. We do this by testing a

series of hypotheses drawn from both the literature of the police and the collective

experiences of project staff and those with whom we consulted. We also present

multivariate analyses drawn from our hypothesis testing and explorations of the data, and

close with a discussion of our work=s implications for police practice and scholarship.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



II. DEFINING AND IDENTIFYING POLICE MISCONDUCT

As originally proposed, this research was to be an examination of duty-related

misconduct that led to officers= involuntary separations from the New York City Police

Department (ANYPD@).

This conception implied a distinction between the venal

activities in which police participated at work, such as beating suspects and taking bribes,

and that which was not directly related to their status as police officers, such as beating

their spouses, driving drunk, or engaging in insurance fraud.

WHAT IS POLICE MISCONDUCT?

This distinction between line of duty misconduct and that which was unrelated to

officers= police status seemed reasonable to us, to our advisors, to the NYPD, and to NIJ

and its peer reviewers. When we began our examination of the data, however, it became

apparent that this distinction was not nearly as clearcut as we all had believed. Figure II-

1 presents the NYPD=s coding schema for disciplinary charges, and gives some

indication of the difficulty of trying to draw a bright line between police deviance and

deviance committed by people who happen to be police officers.

Both NIJ and the NYPD were enthusiastic supporters of this work and, under

four NYPD commissioners, we were given virtually unlimited access to the NYPD files

and resources we needed to compete our work. We began our negotiations for access to

NYPD during the administration of Police Commissioner William Bratton. The great

cooperation he extended us was continued by his two predecessors, Howard Safir,

Bernard Kerik, and Raymond Kelly.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



FIGURE II-1

NYPD MISCONDUCT CODES

001 FOOD STAMP FRAUD

002 FRAUD TO OBTAIN

GOVERNMENT

BENEFITS

003 WELFARE FRAUD

004 ALL UNLISTED

MISDEMEANORS

005 ALL UNLISTED

FELONIES

006 ASSAULT 3 OFF DUTY

007 ASSAULT 3 ON DUTY

008 BRIBE TAKING

009 BURGLARY

010 CONSPIRACY

011 FELONIOUS ASSAULT

OFF DUTY

012 FELONIOUS ASSAULT

ON DUTY

013 GAMBLING

014 GRAND LARCENY

015 DISORDERLY

CONDUCT/

HARASSMENT OFF

DUTY

016 INSURANCE FRAUD

017 DWI

018 LEAVING SCENE -

PERSONAL INJURY

019 LEAVING SCENE -

PROPERTY DAMAGE

020 MENACING

021 MURDER

022 OFFICIAL

MISCONDUCT

023 PERJURY

024 PETIT LARCENY

025 NARCOTICS

POSSESSION

026 POSSESSION OF AN

ILLEGAL WEAPON

027 RECEIVING UNLAWFUL

GRATUITIES

028 RECKLESS

ENDANGERMENT

029 NARCOTICS SALE

030 SCOFFLAW

031 SEXUAL MISCONDUCT

- FELONY

032 SEXUAL MISCONDUCT

- MISDEMEANOR

033 REFUSE RANDOM

DRUG TEST

034 FAIL RANDOM DRUG

TEST

035 USE OF NARCOTICS

036 REFUSE TO TAKE A

AFOR CAUSE@ DRUG

TEST

037 FAIL AFOR CAUSE@

DRUG TEST

038 OTHER CRIME

101 USE OF FORCE ON

DUTY

102 USE OF FORCE OFF

DUTY

103 RACIAL/ETHNIC/

GENDER SLURS

104 RACIAL/ETHNIC/

GENDER

DISCRIMINATION

105 ABUSE OF AUTHORITY

- ARREST

106 ABUSE OF AUTHORITY

- STOP AND FRISK

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



FIGURE II-1 (Continued)

107 ABUSE OF AUTHORITY

- SUMMONS

108 DISCOURTESY

109 VERBAL ALTERCATION

ON DUTY

110 MAKING HARASSING

TELEPHONE CALLS

111 SEXUAL HARASSMENT

112 FAMILY DISPUTES

113 NEIGHBOR DISPUTES

114 OFF DUTY VERBAL

ALTERCATION

201 FAIL TO REPORT

ILLEGAL

ACTIVITIES/DRUGS

202 FAIL TO SUPPLY

PHONE NUMBER

203 FAIL TO PERFORM

ASSIGNED DUTIES

204 FAIL TO REPORT LOST

PROPERTY/FIREARMS

205 FAIL TO RENDER AID

TO INJURED PERSONS

206 FAIL TO PREPARE

REPORTS

207 FAIL TO NOTIFY

COMMANDER OF

SUSPENDED OR

REVOKED LICENSE

208 FAIL TO CONDUCT AN

INVESTIGATION

209 DISOBEY A DIRECT

ORDER

210 FAIL TO REMAIN

ALERT

211 FAIL TO REPORT

MISCONDUCT

212 FAIL TO SAFEGUARD

PROPERTY

213 FAIL TO TAKE POLICE

ACTION

214 FAIL TO VOUCHER

PROPERTY

215 FAIL TO COMPLY WITH

AN ORDER

216 INSUBORDINATION

217 LOSS OF

HOSPITALIZED

PRISONER

218 MITIGATED LOSS OF

PRISONER

219 NEGLIGENT LOSS OF

PRISONER

220 MALINGERING

221 FAIL TO PROCESS

EMERGENCY CALLS

301 BOGUS SHIELD

302 CARRYING

UNAUTHORIZED

FIREARM

303 STORE

UNAUTHORIZED

FIREARMS IN

DEPARTMENT

FACILITY

304 USE OF [FIELD

COMPUTER

TERMINAL FOR

PERSONAL REASONS

305 FAIL TO PROPERLY

SAFEGUARD GUN

306 FAIL TO SAFEGUARD

PROPERTY

307 IMPROPER USE OF

FIREARM

FIGURE II-1 (Continued)

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



308 MITIGATED LOSS OF

GUN

309 NEGLIGENT LOSS OF

GUN

310 NEGLIGENT LOSS OF

RADIO

311 LOSS OF RADIO

TAKING POLICE

ACTION

312 MITIGATED LOSS OF

SHIELD

313 NEGLIGENT LOSS OF

SHIELD

314 CAUSE FALSE ENTRY

TO BE MADE IN

DEPARTMENT

RECORDS

315 DESTROY SUMMONS

316 SUBMIT

FALSE/FORGED

MEDICAL DOCUMENT

317 FALSE STATEMENT

318 PREPARE FALSE

REPORTS

319 FALSE/IMPROPER

ENTRIES IN

DEPARTMENT

RECORDS

320 FALSE/IMPROPER

ACTIVITY LOG

ENTRIES

321 ATTEMPT TO PREVENT

ARREST

322 BRING ALCOHOL INTO

DEPARTMENT VEHICLE

323 CONDUCT PERSONNEL

BUSINESS ON DUTY

324 CONSUME ALCOHOL IN

UNIFORM

325 DESTROY SUMMONS

326 IMPEDE AN

INVESTIGATION

327 FAIL TO PAY FOR

GOODS OR SERVICES

328 AWOL 5 DAYS OR

329 AWOL LESS THAN 5

DAYS

330 ALL MINOR PATROL

GUIDE VIOLATIONS

331 APPROPRIATE

PROPERTY FOR OWN

USE

332 ASSOCIATE WITH

KNOWN CRIMINALS

333 ASSOCIATE WITH

PROSTITUTES

334 AUTHORIZED LEAVE

ABUSE

335 CONDUCT

PREJUDICIAL TO

ORDER OR

DEPARTMENT

336 COOPING

337 DISCLOSE OFFICIAL

DEPARTMENT

BUSINESS

338 DISCOURTESY TO A

SUPERIOR

339 FEIGN ILLNESS

340 IMPROPER PATROL

341 IMPROPER

SUPERVISION

342 IMPROPER UNIFORM

343 LATENESS

344 MILITARY LEAVE

ABUSE

345 OFF POST

346 OTHER SICK LEAVE

VIOLATION

347 OUT OF RESIDENCE ON

SICK REPORT

348 OVERTIME ABUSE

FIGURE II-1 (Continued)

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



349 RESIDENCY

VIOLATIONS

350 TRAFFIC VIOLATION

351 INTOXICATED ON

DUTY

352 UNFIT FOR DUTY - OFF-

DUTY

353 WORK ILLEGAL

OCCUPATION

354 WORK UNAUTHORIZED

OCCUPATION

355 WORK WHILE ON SICK

REPORT

356 WORKING WITHOUT

PERMISSION

357 FAIL TO CONTACT

SURGEON

358 FAIL TO SUBMIT

MEDICAL

DOCUMENTATION

359 VIOLATE

DISCIPLINARY

PROBATION

369 FREQUENT ILLEGAL

LOCATION

370 MISCELLANEOUS

ADMINISTRATIVE

Failed Probation NOT INVOLVING SPECIFIC MISCONDUCT DESCRIBED

ABOVE (Excluded from Analysis)

401 ACADEMIC FAILURE

402 PHYSICAL SCHOOL FAILURE

403 FIREARMS AND TACTICS FAILURE

404 DISCIPLINARY FAILURE

405 PSYCHOLOGICAL FAILURE

406 FIELD TRAINING FAILURE

407 END OF PROBATION RECOMMENDATION BY SUPERVISOR OR

COMMANDER

408 NOT SPECIFIED

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



As we quickly discovered during data collection, a considerable number of the

cases we examined were dismissals caused by officers= failure or refusal to take part

in the NYPD=s extensive drug-testing program. Other officers were caught possessing

or trafficking drugs. Such cases presented the insoluble problem of determining

whether and to what extent these offenses were related to officers= membership in the

NYPD. Was Officer A already a drug abuser when he became a police officer? If so,

did his abuse become worse B or did he himself become a trafficker through the

connections he made on duty? If he was not a drug abuser when he joined the

department, did he become a drug abuser because of on-duty contacts or experiences?

In either case, did he use drugs or sell drugs on-duty as well as off-duty? Did Officer B

engage in insurance fraud independent of her role in the NYPD, or was she given

opportunities, or an education in how to do it, by people or experiences associated with

her police work? Did Officer C drive drunk because he was self-medicating with

alcohol to deal with the stresses of the job, and because he believed that his police

status would help him talk his way out of any contact with the police and cover up any

damage he did?

In short, we found that the line between duty-related misconduct and

officers= private business was not nearly as bright as we had anticipated. We also

found that our examinations of NYPD files did not resolve these questions and, even

had there been some practical way of locating and interviewing the officers involved,

A recent police scandal in Philadelphia involves precisely this issue. See

Fazlollah, 2001.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



we had no reasons to anticipate that they would be forthcoming about the details of

their problems.

There was another reason to expand our research beyond the original line-of-

duty misconduct classification. Regardless of whether it can be neatly classified as

duty-related, a police firing is a firing, that brings with it consequences that redound

negatively to officers involved, to those they may have victimized, to the reputation

and good order of the department, to the municipal treasury, to officers themselves,

and to their families. Thus, with additional funding from NIJ, we expanded our

research to include all classifications of career-ending misconduct by police officers,

and it became a study that compared failed police careers and those which, at least by

December 31, 1996, had either ended honorably or were still in progress.

COUNTING THE FIRED

A second issue in studies such as this is determining whose career has ended

because of misconduct. The NYPD maintains no central file that would provide the

answer to this question. Instead, usually two to four times a week, it publishes

Personnel Orders and disseminates them to every departmental unit. These orders

report every appointment, promotion, transfer, change in designation, resignation,

retirement, vesting, dismissal, termination, or death of both uniformed and civilian

During Fyfe=s first tenure in the Police Academy, staff attempted several times

to gain the participation of dismissed officers in training programs and videos. Only in

the early 1990s, did one officer finally agree to do so.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



NYPD personnel. For several reasons, however, these are not a ready source of

information for a study such as ours.

Involuntary Separations not Related to Misconduct

First, not all involuntary separations are related to misconduct. In New York

City, dismissal is the term used to describe firing of tenured civil servants, including

the police. Since dismissal is always a penalty for misconduct, all officers dismissed

by the NYPD between 1975 and 1996 were included in our analysis. Prior to the

award of tenure, however, the department typically does not dismiss officers, because

this category of involuntary separation requires some due process, beginning with the

specification of charges against officers. Instead, probationary officers typically are

terminated, a designation that does not require specification or proof of charges, but

that instead requires only a statement of the Police Commissioner=s determination that

an individual has proven to be an Aunsatisfactory probationer.@ Much more often

than not, such terminations are based on candidates= failures to satisfactorily meet the

Police Academy=s standards for performance in the academic, physical, or firearms

and tactics training programs. Since these failures involve inadequacies rather than

misconduct, we excluded them from analysis.

The number of these terminations is substantial and, over the period of our

study, may be equal to or larger than the number of separations we included. We

began assembling our data by checking 1996 records and proceeding backward to

1975. By the time we had worked back from 1996 to 1987, we had identified 1,591

officers who had been involuntarily separated. Only 741 of these officers eventually

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



were included in our study group; the great majority of the 850 excluded officers were

non-behavioral probationary failures.

We did retain for analysis probationary officers whose terminations obviously

were rooted in misconduct, including failure to abide by the Police Academy=s

disciplinary rules or, quite often, failure to take or pass a drug test. Determining

which of the terminated probationers had been separated for behavioral reasons, rather

than for simple training failures, required us to use the Personnel Orders to identify

every terminated officer, and thence, to proceed to each officer=s personnel history to

determine the cause for termination. In doing so, we reviewed at least 1,000 cases that

eventually were deleted from our analyses.

In this process, we also encountered a

An historical note on this point may be instructive. The principal investigator of

this study is both an alumnus (1963) and a former and present staff member (1973-79;

2002-present) of the NYPD Police Academy. In 1963 and, indeed, until 1973, when the

NYPD began hiring large numbers of recruit officers after a hiatus forced by a minor

fiscal crisis, officers rarely were terminated during their probationary periods for reasons

unrelated to serious misconduct. Written, physical, and strength and agility exams, and

background and character investigations at that time were sufficiently stringent so that

they were regarded as the agency=s major screeners; that they also had not been

validated as job relevant and that they had discriminatory effects against women and

other protected groups was not yet on any administrators= radar screens. At that time,

anybody who passed through the pre-employment process was deemed qualified to

become a tenured officer, so that the probationary period was a mere formality. Indeed,

new officers during those years were instructed to make certain that any supervisors who

might catch them in wrongdoing understood that they were probationers because

department norms demanded that ranking officers refrain from disciplinary actions

against probationers in order to avoid ending their careers for youthful mistakes. In

effect, the award of an officer=s shield at one=s probationary appointment to the NYPD

was a de facto lifetime appointment that one might lose during probation only for

conduct that would also have resulted in severe discipline or dismissal of officers whose

probationary periods had been completed. This changed in the early 1970s, when the

NYPD responded to equal opportunity legislation and litigation by modifying its

standards for probationary appointments so that their discriminatory effects against

women and members of racial and ethnic minorities were reduced. However

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



small number of cases in which probationary officers had been Adecertified@ when it

was discovered that they had concealed pre-employment histories of criminal behavior

or mental illness. These cases, too, were excluded on the reasoning that, although

these officers were effectively living a lie after they were hired, they would have been

screened out in a more thorough pre-employment investigatory process.

Apparently Voluntary Separations Related to Misconduct

Another complication in our efforts to identify involuntarily separated officers

was the NYPD=s practice of forcing some officers to retire or resign under honorable

conditions in return for their cooperation in investigating and prosecuting wrongdoing

of which they had been a part. Perhaps the best known example of this mode of

leaving the agency was Detective Robert Leuci, the protagonist in the 1970s Prince of

the City scandals involving the Special Investigations Unit, the NYPD=s elite narcotics

squad (Daley, 1978). Although Leuci admittedly was involved in chargeable offenses,

his cooperation and testimony in prosecutions of other corrupt officers was part of an

agreement that allowed him to remain in service until he became eligible to retire on

the twentieth anniversary of his appointment.

commendable, this change meant that very few candidates were screened out by the

entrance examinations and background investigations. Consequently, the presumably

more job relevant training and probationary periods generally became the agency=s

major screening devices. For the first time, then, significant numbers of new officers

were terminated during these periods, both for reasons of inadequate performance, and

because of the end of tolerance of improper conduct by probationers.

Leuci was one of the first Aturned@ officers whose cooperation and testimony

were rewarded in this manner. Prior to the early-1970s corruption scandals in New York,

officers implicated in wrongdoing were offered no deals, and B as Arogue cops@ and

Abad apples@ B were instead prosecuted severely in both the criminal courts and the

NYPD=s internal disciplinary mechanisms. However intuitively appealing it may be to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Not all such cases are so readily identifiable. Consequently, we attempted to

locate as many others as possible by checking into the circumstances of all apparently

premature departures from the NYPD. We also sought out such information from

members of departmental units most likely to know of them (e.g., Internal Affairs; the

Department Advocate, its prosecutor in administrative disciplinary cases; the Legal

Bureau; and the Personnel Bureau).

The Study and Control Officers

Identifying the study and control officers were extremely labor intensive

processes. Our review of the Personnel Orders produced an original pool of about

3,000 officers who were deemed worthy of further investigation for inclusion in the

data set we would eventually analyze.

More than 1,000 turned out to be recruits

terminated for substandard performance that did not involve specific acts of

misconduct, and several hundred others were determined to be officers who left in

good standing. This left us with 1,543 officers who fit our definition of involuntarily

separated. Almost certainly, we have missed some additional cases, but it is fair to say

that we did everything reasonably possible to capture them all, and that the few that

may have slipped through our net do not affect the direction or strength of our

punish deviant officers harshly in this manner, this policy made it impossible to use such

officers to develop evidence against either their corrupt colleagues or the members of the

public who were parties to their corrupt arrangements. Only when this policy was ended

by application to police misconduct cases of the more traditional practice of using little

ones to get big ones did the NYPD and other investigators begin to make real inroads into

organized corruption.

It is impossible to state this figure with precision because our pool of potential

study officers was constantly changing. We attempted to screen cases out of the data

simultaneous with data collection. Thus, some officers were excluded on the same day

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Because of the random manner in which they were selected, we presume

them to be

findings. To serve as controls to these study officers, we also selected a random

sample of their police academy classmates. We did this the old fashioned way: by

running through an alphabetical list of the officers appointed in each class and using a

list of random numbers to count down the list to select the appropriate number of

control officers. This was a labor intensive process. In some cases, we selected and

coded controls, only to find that their corresponding study officers did not meet our

criteria for involuntary separation. In other cases, we found that control officers had

resigned from the NYPD within the first few days after their appointments, so that

their files included insufficient information for comparisons of any kind. In still other

cases, we found that designated control officers had themselves left the NYPD so long

ago that their files had been destroyed in accord with the agency=s 21-year document

retention schedule. In instances in which our original randomly selected control

officers turned out to be unusable for analysis, we included in our control group the

next officers on the class rosters. In the end, we derived a sample of 1,542 control

officers.

that they had been identified, and other were screened out only some time later.

In some cases, the official appointment dates of study and control officer pairs

differ. This typically is an artifact because one or the other has an appointment date

adjusted by the award of credit for prior government service or because one was in

military service, and therefore unavailable for police training on the date upon which he

or she first became eligible for promotion.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



representative of their Police Academy cohorts as a whole. Consequently, differences

between them and the study officers may be presumed to define the distinctions

between involuntarily separated officers and those who served honorably or, at least,

who have not been caught engaging in career-ending misconduct.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



III. PRIOR RESEARCH ON POLICE MISCONDUCT

There exists an extensive body of literature on police misconduct, but we know

of no previously published work that focuses on a range of behavior as broad as that

studied in this research. For this reason, as Figure II-1 suggests, the misconduct

classifications used in prior research are not fully adequate to describe the phenomena

that are the subject of our work. We have examined every incident in which an officer

was involuntarily separated from the NYPD for any behavioral reason. Previous work

typically has concentrated on specific types of occupational deviance, such as

corruption (e.g., taking bribes from vice operators) or brutality (e.g., beating

arrestees), that is clearly linked to the offender=s status as a police officer, and does

not examine wrongful behavior that may be less obviously associated with police

status. Still, it is worth discussing this prior work because it sets a context for our own

and because such a discussion illustrates the inadequacy of existing classifications for

our purposes.

EXISTING CLASSIFICATIONS OF POLICE MISCONDUCT

Occupational deviance by police officers has been variously described by many

policing scholars, leading to a general conceptualization that distinguishes among

Atypes@ of job-related misbehavior. As the following discussion suggests, however,

these discussions are useful primarily for studies that focus on particular cases or

episodes, and are of limited utility in a broad study such as this.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



Police Crime

The first category of police deviance cited in the literature is police crime,

which involves the use of officers= positions of public trust to violate existing criminal

statutes (Kappeler et al., 1994). As Sherman (1978) noted, and as we found, however,

police crime does not describe all crimes committed by police officers, since many

offenses may have nothing to do with officers= employment status. Our data set

includes officers who engaged in off-duty burglaries, domestic assaults, or tax evasion,

all of which certainly are crimes. Absent abuse of their police authority to gaining the

opportunity B or, perhaps, the skills B to commit the crimes, however, these acts of

deviance probably should not be considered police crime. An example makes the

point: in our view, any officer who steals drugs from an evidence locker and sells

them has engaged in police crime because his employment status created access to the

evidence locker and, therefore, made the crime possible. But, absent such an

indication of where an officer charged with dealing drugs may have obtained her

wares, one can only speculate on whether her offense is job-related. In the absence of

such evidence, as well as other information that would allow one to clearly distinguish

between police crime and other offenses crimes by police, it is very difficult in

practice to draw a bright line that clearly delineates police crime.

Police Corruption

The next form of deviance is police corruption, which has been the subject of

varying definitions. The consistent feature of most definitions of police corruption is

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 23

that it involves officers who use their position or authority to engage in misconduct,

through act or omission, to achieve personal benefit. Thus, these definitions suggests

that police corruption may be conceptualized as profit-motivated police misconduct.

But see McCafferty and McCafferty (1998) who define corruption to include

Amooching, chiseling, favoritism, prejudice, shoplifting, extortion, accepting bribes,

shakedown, perjury, premeditated theft. Other forms of corruption include drinking on

the job, having sex with informants and others, carrying unauthorized weapons, sleeping

and doing personal chores while on duty, assault, and others.@

Criminal and Administrative Corruption. The literature=s conception of police

corruption as profit-motivated misconduct means that it is not entirely distinguishable

from police crime, and illustrates a major problem with existing classifications. In

addition, Hale (1989) points out, there are differing views of whether corruption

should be defined to include only illegal behaviors. McMullan (1961) B who did not

restrict his scope to police officers B noted that any public officials are Acorrupt@ if

they accept compensation for not performing regular duties, or for performing duties

normally proscribed by their employment positions. McMullan=s formulation

recognized that both legal and illegal behavior may be considered corrupt. So, too,

does the definition offered by Sherman (1978:30), who wrote simply that an act of

police deviance represents corruption when the act is committed for Apersonal gain.@

Finally, Goldstein (1977:188) articulated a definition similar to Sherman=s, defining

corruption as profit-motivated misconduct, without clearly specifying that its wrongful

nature must be defined in criminal law.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 24

Whether one considers corruption to include only crime or, more broadly, to

also include administrative or ethical violations has important implications for our

attempts to classify and simplify the range of behaviors described in Figure II-1. In

New York and other cities, for example, police agencies have placed administrative

limits on officers= off-duty employment activities. New York officers who violated

such limits by, say, working second jobs for more than the maximum 20 hours allowed

by department regulations or by accepting administratively prohibited private security

positions in the patrol precincts to which they were assigned would be considered

Acorrupt@ under McMullan=s definition. Because such conduct violates only

administrative regulations and is not proscribed by law, it could not be considered

Acorrupt@ if only statutory violations were included in this classification.

Police corruption or employment corruption? Another ambiguity is illustrated

by the first several misconduct classifications defined in Figure II-1 (food stamp fraud;

fraud to obtain government benefits; welfare fraud). These all are crimes, and all are

profit-motivated. Their relationship to offenders= police authority, however, is absent

or, at least, far less clear than in cases involving officers who take bribes to allow

illegal behavior or who sell drugs they have stolen from narcotics traffickers during the

course of arrests and seizures. Instead, in most cases, these offenders are persons who

were receiving food stamps, welfare, or other government benefits at the time they

were appointed to the NYPD and who failed to notify the agencies supporting them

that they had secured paying employment and were therefore no longer eligible for

such benefits. Many such offenders were detected in the course of cross-checks of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 25

New York City welfare roles and employment rosters. Thus, these offenders were

corrupt in that they committed profit-motivated crimes that were related to their

employment; but only because the fact of their employment B no matter what it may

have been B disqualified them for the benefits they were receiving. These offenses

involved no apparent use or abuse of police authority and, as experience showed, could

just as easily have occurred had the offenders worked in the city=s board of education

or in any of a range of other non-police employment situations. In short, while these

activities unquestionably involve corruption and offenders= employment, they do not

involve crime that is unique to the police.

Unambiguous police corruption. When working within the parameters of his

definition of corruption as profit-motivated abuse of police authority, Sherman (1978)

noted two types of police corruption: events and arrangements. Officers who engage in

corrupt events are generally individuals who practice profit motivated misconduct with

varying degrees of repetition, and most frequently with different victims (Sherman,

1978). An example of event corruption is a drug enforcement officer who removes

and sells some of the drugs he seizes in the course of arrests. According to Sherman,

corrupt events are difficult for police administrators to detect since the officer-victim

combinations are different during each transaction. Moreover, officers who engage in

this type of deviance can further minimize their risk of detection by choosing

unsympathetic victims of limited credibility (Kappeler and Potter, 1993).

Corrupt arrangements tend to involve police officers acting in groups,

representing organized corruption involving the same officers and the same victims,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 26

and maintaining a standard degree of repetition (Sherman, 1978). An example of

corrupt arrangements is a group (or squad) of officers who extort or accept money

from gamblers, so the latter may operate without police interference. Corrupt

arrangements were discovered to exist in the New York City Police Department by the

Knapp Commission (1972), which identified pads, or networks of payoffs to officers at

regular, usually monthly, intervals. As Sherman (1978) noted, the vulnerability of

detection is largely a function of predictability. Therefore, officers who establish

corrupt arrangements risk detection at a higher rate than those who participate in

corrupt events. It is relatively easy for officials to detect and sanction corruption

involving regular monthly payoffs because participants must meet or otherwise arrange

to make exchanges. It is more difficult to predict when opportunities for event-based

corruption will arise.

Abuse of Power

The final form of police deviance described in the literature is abuse of power,

which Carter (1985:322) defined as Aany action by a police officer without regard to

motive, intent, or malice that tends to injure, insult, tread on human dignity, manifest

feelings of inferiority, and/or violate an inherent legal right of a member of the

[public].@ This definition is commonsensical, but suffers because it and corruption are

not mutually exclusive. Instead, they often are one and the same. Included among the

profit-motivated misconduct identified by those who have recently investigated

corruption in Los Angeles, New York, and Philadelphia were such abuses of power as

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 27

robbery, kidnaping, and attempted murder (Los Angeles Police Department, 2001;

Mollen Commission, 1994; Philadelphia City Council, 1995).

Setting aside this classification problem, Kappeler, et al. (1994) point out,

Carter=s definition considered three broad areas of police abuse including physical,

psychological, and legal domains.

Physical Abuse. Over the years since the internationally publicized Los

Angeles Police Department beating of fleeing motorist Rodney King, police violence

(or physical abuse) has become perhaps the most widely discussed and debated form of

police occupational deviance committed by police (Skolnick & Fyfe, 1993). While

excessive force is often considered a single construct of police abuse of authority, Fyfe

(1986) distinguished between extralegal and unnecessary police violence. Fyfe argued

that extralegal force, or brutality, represented intentional physical abuse inflicted

maliciously and for no legitimate police purposes against persons whose major offense

were challenges to police authority (see also Van Maanen, 1978). In this context,

brutality is a form of improper punishment. It is designed to convey the message that

such behavior as fleeing from the police or questioning police judgment or officers=

power to take action has a great and immediate cost independent of whatever formal

penalties may subsequently be imposed by the courts (Worden, 1996). Officers who

engage in brutality typically justify it as a deterrent: as a method of assuring that the

next officer who encounters one who has been thus instructed in the cost of

challenging the police will find only compliance rather than resistance.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 28

By contrast, unnecessary force typically results from police incompetence or

carelessness, and is generally not the product of malice. It usually occurs when

officers unnecessarily put themselves in harm=s way by using poor tactics while

approaching potentially violent persons or situations. Then, when potential violence

suddenly becomes real, the officers find that their exposed and vulnerable positions

have left them no options but to resort to force to defend themselves (Fyfe, 1986). A

typical example of this occurs when police respond to a man with a gun call. They

might arrive on the scene to find an agitated man pacing in his front yard with a pistol

in hand. Because officers neglect to find cover or concealment

(i.e., failing to

position themselves behind parked cars or otherwise out of the direct line of fire), they

shoot the man when he makes a sudden movement that they perceive to be threatening.

When, in the manner of criminal prosecutors, one focuses on the Afinal frame@ of the

incident (e.g., the instant immediately preceding the shooting; see Binder & Scharf,

1980) one might conclude that the force used was be justifiable since, at that instant,

the police had no real choice except to use force to protect themselves. However,

when considering the events that led to the shooting, as some courts have done (Zuchel

v.Denver,1993) and as responsible police administrators should (Fyfe, 1975, 1986;

Klockars, 1996), one might find that incompetence and/or carelessness on the part of

officers left them at risk of being shot in the first place. Had they found proper cover,

Like the military, the police define cover as a barrier that will stop a bullet (e.g.,

a brick wall; a vehicle=s engine block; a thick tree), while concealment includes only

barriers capable of hiding an individual=s presence without the capacity to stop a bullet

(e.g, a bush; a sheetrock wall; a typical door).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 29

they might have found it unnecessary to shoot (e.g., thus protected, they might have

been able to wait and persuade the man to drop the gun, which would have averted the

use of violence).

None of the officers in our study were involuntarily separated for using

unnecessary force. This is not surprising because this type of wrongful force typically

indicates an agency-wide training or policy deficiency rather than individual venality

by the officers who engage in it. Thus, in the NYPD, it usually has resulted in less

drastic action (e.g., retraining in law or tactics; occasional suspensions) rather than in

dismissal. The force that served as the basis for involuntary separations of the officers

studied in this research, therefore, consisted of various types of brutality.

Psychological abuse. According to Carter (1985), police psychological abuse

has historically been rooted primarily in police interview practices, and developed

largely in response to the prohibition of the use of physically compelling tactics during

interrogations. As del Carmen (1991) noted, the Supreme Court=s decision in Brown

v. Mississippi (1936), which banned the police use of physical coercion during

interrogations, led officers to begin a reliance on psychological coercion as a means of

obtaining confessions. Though the Court officially forbade Aintense psychological@

coercion of suspects in police custody (see Spano v. New York, 1959), del Carmen

suggests that it is likely that such abuse continues in U.S. policing (del Carmen,

1991).

Legal abuse.

Generally, legal abuse involves police officers who violate

criminal statutes, or the rights of citizens (typically, accused offenders) in order to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 30

achieve some organizational goal (Kappeler, et al., 1994) or to accomplish some

presumably noble cause (Crank and Caldero, 2000; Klockars, 1980). One of its forms

is police perjury that is designed to ensure that an accused offender is adjudicated

guilty. It may also involve officers who set up illegal wire taps in order to

surreptitiously gather incriminating information on suspects. Certainly, these

examples are not exhaustive B one of the more appallingly imaginative episodes

involved six New York State troopers who planted and then pretended to discover

latent fingerprints at crime scenes, purportedly to strengthen cases against suspects

whom they were convinced were guilty (Perez-Pena, 1997).

Police abuse of authority is an interesting form of deviance. Like other forms

of misconduct (i.e., police crime and corruption), it involves abuse of the police

powers of office. However, unlike the other forms of police deviance, police abuse of

authority is often structured to achieve organizational B as opposed to personal B gain.

Often, it is, as Klockars (1980) observed, the use of dirty means to achieve what most

would regard as noble ends: the punishment of people who did, in fact, commit the

crimes of which they are suspected but who, for one reason or another, could not be

convicted on the basis of legally admissible evidence. In other cases, as in the recent

Oakland police ARiders@ scandal involving officers who allegedly planted drugs on

persons whom they

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 31

then arrested, it is an even more perverted method of generating impressive arrest

statistics (Glionna, 2001).

Off-Duty Misconduct

Not all misconduct by the police occurs while they are on-duty. Fyfe (1980a)

reported that, in New York City, about 20 percent of police firearms discharges

involved officers who were off-duty, and that the NYPD found cause for disciplinary

or criminal action in half of these cases. He subsequently reported a similar pattern of

inappropriate off-duty police shooting in Philadelphia (Fyfe, 1987), and found that

Philadelphia officers frequently engaged in lesser degrees of inappropriate force while

off-duty (Fyfe, 1998). In addition, the data examined in these evaluations included

incidents in which off-duty officers abused their police authority to resolve personal

disputes; domestic violence; bar fights; drunk driving and related vehicle accidents,

including hit and run collisions; acts of vandalism; sex offenses; and such property

crimes as larceny and burglary. In short, in addition to misconduct directly related to

their status as police, off-duty officers engage in all the offenses available to the

general public.

On occasion, of course, police abuse of authority serves both organizational and

personal goals: big arrests in highly publicized cases often bring praise to police agencies

and promotions to arresting officers.

When is Off-Duty Not Line of Duty? Drawing the line between police off-duty

conduct that is not job-related and that which is associated with offenders= status as

police officers often is not an easy task. Most obviously, as suggested above, police

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 32

officers= guns, badges, authority, special knowledge, and access to contraband and

criminal opportunities while on-duty may facilitate off-duty criminality or violations of

departmental regulations.

In addition, in New York and other jurisdictions, police officers have law

enforcement authority 24 hours a day, seven days a week, both within the

municipalities and counties that employ them and throughout the state. This

arrangement occasionally causes controversy about whether actions by off-duty

officers B especially when involving weapons issued or authorized by their police

employers B should be considered to have occurred Ain the line of police duty.@

Although the NYPD has long attempted to discourage officers from taking police

action in situations that are not imminently life-threatening,

there remain

controversies about which off-duty actions are legitimate, which are wrongful abuses

of police authority, and which are simply aberrations independent of officers= police

status. In 1987, the NYPD advised officers in a legal notice that it would be unlikely

to defend them in civil suits arising from their attempts to take off-duty police action

while outside the city. This notice also advised officers that the department would

scrutinize off-duty actions within the city on a case-by-case basis to determine whether

they were line-of-duty activities (NYPD, 1987). Regardless of the attempts of the

department and its lawyers

The NYPD=s original temporary order limiting officers= off-duty authority

was issued on March 20, 1981, and became a formal amendment to the Patrol Guide, the

department=s manual, in 1984 (NYPD, 1984).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 33

to divorce itself from off-duty abuses by officers, both the police union and attorneys

representing persons who claim to have been victimized by of-duty police invariably

argue that whatever officers do off-duty is associated with their police work. Often,

but not always, they win these arguments.

Two Illustrations. Two cases from different sides of the continent illustrate the

ambiguity concerning whether and when police off-duty misconduct is job-related.

Perhaps the leading New York City case on this point is Bonsignore v. City of New

York (1982). Blaise Bonsignore was an NYPD officer who was regarded by his

commanders and supervisors as emotionally and psychologically unfit for field duty.

Consequently, even though he was stationed at a patrol precinct, Bonsignore had been

assigned to station house duty (as the attendant, or Abroom@) for more than a decade.

He was, however, permitted to retain possession of his guns.

While off-duty, he used

his gun to shoot his estranged wife, causing permanent brain damage, and to kill

himself. In a suit brought by Bonsignore=s wife, a jury found that his actions were a

predictable result of the department=s failure to keep Bonsignore=s gun from him

while he was off-duty. Consequently, the shooting was found to be job-related, and

the NYPD was found liable for the injuries Bonsignore inflicted upon his wife. The

At the time, the NYPD did not issue guns to officers, but instead required them

to purchase and equip themselves with designated weapons. NYPD presently supplies

officers with duty weapons, leaving them the option of buying one or more of several

designated smaller weapons for use while off-duty or on plainclothes duty.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 34

Second Circuit United States Court of Appeals rejected New York City=s appeal from

this verdict.

In California, however, the federal courts reached a different conclusion in

Huffman v. County of Los Angeles (1998), a similar, if not precisely analogous, case.

There, Thomas Kirsch, an off-duty deputy of the Los Angeles County Sheriff=s

Department (LASD) spent a Sunday afternoon and evening drinking in a neighborhood

bar. Apparently very intoxicated, he became involved in an argument with John

Huffman, a young man, and apparently accepted an invitation to step outside. Once

there, allegedly without identifying himself as a deputy, Kirsch drew his service

weapon from behind a large western-style buckle on his belt, and shot and killed

Huffman. Huffman=s family sued Kirsch and the LASD arguing, in effect, that the

death was the predictable result of, first, a departmental requirement that deputies be

armed all the time and, second, an official policy that allowed deputies to carry

weapons even while intoxicated. As evidence that the shooting was job-related and

consistent with the Sheriff=s policy, Huffman=s survivors pointed out in court papers

filed more than two years after the shooting that the LASD had taken no action to

discipline Kirsch. As evidence that the shooting was predictable, Huffman=s survivors

showed that:

Thomas Kirsch received no training or suggestion that he should

not carry a gun while drinking alcohol. He has never been

provided with any written procedures or training bulletins or

special orders, or any written material on the subject. He stated,

under penalty of perjury, that he was required and encouraged to

carry a gun at all times. He was trained that he was a deputy

sheriff 24 hours a day.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 35

LASD knew of at least 80 incidents, between [December]

1989 and [mid-[ 1994,in which there were allegations that off-

duty deputy sheriffs had brandished or discharged firearms.

Approximately one-half of the incidents disclosed involved

freeway disputes [situations in which deputies in their private

vehicles became involved in disputes with other motorists].

Approximately six involved personal disputes in which officers

fired guns. Approximately fifteen of the incidents that were

disclosed involved situations in which the use of alcohol was

undisputed. There were additional instances in which it appeared

that alcohol was involved but in which LAPD failed to mention

that fact in its investigative reports, e.g., incidents outside of bars

or parties...

It is the express policy of LASD that it does not restrict the use or

carrying of a gun by an off-duty deputy even if he is intoxicated

(Huffman v. County of Los Angeles, Petitioner=s Writ for

Certiorari, 1998).

These arguments convinced a jury, which ruled that Huffman=s death was the

result of the LASD=s inadequate policies and practices. The Ninth Circuit U.S. Court of

Appeals, however, overturned this finding, ruling instead that Huffman=s death was the

result of Aprivate acts@ unrelated to either LASD policies or to Kirsch=s position as a law

enforcement officer. The Circuit=s decision was allowed to stand by the U.S. Supreme

Court, which denied Huffman=s writ for certiorari.

Thus, the Second Circuit treated Bonsignore=s attempt to kill his wife with his

police gun as a job-related act, and held the NYPD liable because its policy and practice

(failing to see that an emotionally unstable officer did not have access to his service

weapon while off-duty) made this shooting predictable. The Ninth Circuit, by contrast,

ruled that Kirsch=s killing of Huffman was a private act that was unrelated to his

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 36

employer=s failure to see that off-duty officers behaved responsibly and did not carry or

use weapons while intoxicated.

Other Ambiguities. Elizabeth Reuss-Ianni studied New York City police in

the1970s, when the first of the officers in our study were separated from the NYPD. She

concluded that there then existed in the department a Astreet cop culture,@ in which

officers placed a very high value on not Agiving up another cop@ (Reuss-Ianni, 1983:14).

Variants of this code of behavior have been reported in virtually every relevant study of

the police (Westley, 1953; Skolnick, 1966; Reiss,1968, 1971; Stoddard,1968; Chambliss

and Seidman, 1971; Rubinstein,1973; Ivkovich and Klockars, 1995), as well as in official

investigations into policing in New York (Knapp Commission,1972; Kelly, 1992; Mollen

Commission, 1994); Boston (St. Clair Commission,1992); Los Angeles (Christopher

Commission, 1991; Los Angeles Police Department, 2000); Los Angeles County(Kolts

Commission,1992); Milwaukee (DiUlio, et al.,1991); New Orleans (New Orleans, 1993);

Philadelphia (Philadelphia Police Study Task Force,1987); and Sydney (Royal

Commission, 1997).

As one examines data such as those analyzed in this report, a question related to

this ambiguity evinces itself in cases less dramatic than the two shootings described above:

To what extent is off-duty misconduct job-related in the sense that offending officers trust

that other officers who may discover it will treat them differently from ordinary citizens

caught in the same circumstances? Can an officer who drives recklessly or while

intoxicated, beats his spouse, or engages in street brawling, for example, do so in some

confidence that, even if caught, he will not be subjected to the same formal processing that

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 37

would befall one who could not flash a badge and ask for Aprofessional courtesy?@

Examples of such extraordinary treatment, mostly in agencies other than the NYPD,

appear in the press or in the courts with sufficient frequency to suggest that at least some

officers may engage in off-duty misconduct only because they believe their police status

will grant them immunity from punishment.

OUR CLASSIFICATIONS OF POLICE MISCONDUCT

The unsatisfactory nature of existing classifications of police misconduct and our

own inability to divine either officers= logic or whether much misconduct was job-related

led us to construct our own classifications of police misconduct, which is employed in

several sections of our research. To do so, we collapsed NYPD=s broad range of

misconduct categories into the following:

1. Profit-motivated crimes: All offenses, other than drug trafficking and

whether on-duty or off-duty, in which the end or apparent goal of officers=

wrongdoing doing was a profit.

2. Off-duty crimes against persons: All assaultive behavior, except for profit-

motivated robberies, by off-duty officers.

Fyfe (1998) reported on a civil rights action involving off-duty Philadelphia

officers who had assaulted two young men after a highway dispute. When witnesses

(who did not realize that the assailants were police) called the police department to

intercede, they were surprised to see that the responding officers allowed the beatings to

continue, and that they arrested the victims rather than the assailants. This incident came

to light only because the off-duty officers subsequently vandalized the young men=s car;

a prior similar incident involving one of the officers had resulted in no formal action by

the department. More recently, Fazlollah (2001) reported on the cover-up of an apparent

hit and run accident by an off-duty Philadelphia police captain who was driving a police

department car while intoxicated .

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 38

3. Off-duty public order crimes: All offenses, other than drug trafficking or

possession, against public order, including driving while intoxicated and

disorderly conduct.

4. Drugs: Possession and sale of drugs, and related conspiracies, as well as

failing or refusing to submit to departmental drug tests.

5. On-duty abuse: All offenses by on-duty officers involving use of excessive

force, psychological abuse, or discrimination based on citizens=

membership in a class (e.g., gender, race, ethnicity, sexual preference).

6. Obstruction of justice: conspiracy, perjury, official misconduct, and all

offenses in which the apparent goal is obstruction or subversion of judicial

proceedings.

7. Administrative/failure to perform: Failure to abide by departmental

regulations concerning attendance, performance, obedience, reporting, and

other conduct not including in other offense types.

8. Conduct-related probationary failures: All misconduct-related terminations

of probationary officers in which misconduct in types1-7 is not specified,

and excluding simple failure in training programs.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 39

CORRELATES OF POLICE MISCONDUCT

Studies attempting to identify factors associated with police misconduct have

focused on community and organizational variables, as well as upon personal

characteristics of police officers. These works serve as guides for our own study.

Community and Organizational Correlates of Police Deviance

In large eastern cities, the literature suggests, the trouble most frequently affecting

police is profit-motivated misconduct (e.g, Mollen Commission, 1994). In more recently

developed western jurisdictions, police deviance may usually involve on-duty abuse (e.g.,

Christopher Commission, 1991). There are some exceptions to this generality,

but the

literature is replete with observations that confirm it (Bobb, et al., 1992, 1993, 1994a,

1994b, 1995, 1996a, 1996b; Chevigny, 1995; Chicago Police Committee, 1931; Cohen,

1980; Daley, 1978; Domanick, 1994; Fogelson, 1977; Gates, 1992:85-87; Goldstein,

1977:214; Kappeler, Sluder, and Alpert, 1994:145-167, 187-238; Kelly, 1992; Knapp

Commission, 1972; Kolts Commission, 1992; Maas, 1972; McAlary, 1989; Mollen

Commission, 1993, 1994; Murphy and Plate, 1977; Philadelphia Police Study Task Force,

Perhaps most notably, the recent Rampart scandal, in which Los Angeles police

officers allegedly were involved in trafficking and stealing narcotics, and in related

violence (Los Angeles Police Department, 2000). In addition. Philadelphia and New

Orleans have, at various times, have been marred by scandals involving profit motivated

misconduct and on-duty abuse (see, e.g., Fyfe, 1980b; New Orleans Mayor's Advisory

Committee, 1993; Skolnick and Fyfe, 1993; Thrasher, et al., 1979; United States Civil

Rights Commission, 1979, 1981; United States v. Philadelphia, 1979; Williams, 1974).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 40

1987; Rothmiller and Goldman, 1992; Rubinstein, 1973; Schecter and Phillips, 1973;

Sherman, 1978; Skolnick and Fyfe, 1993; Williams, 1974).

Despite the history of corruption scandals in New York City, Chevigny (1996:85)

writes that "[t]he NYPD is not a notably abusive department," and that a thriving "lawyer

who specializes in damage actions for police brutality in Los Angeles told [him] that he

would starve if he had to practice in New York."

Kappeler, Sluder and Alpert (1994:

145-167, 187-238) write at length regarding violence in the Los Angeles Police

Department and corruption in the New York City and Washington DC Police

Departments. In 1931, the blue-ribbon Chicago Police Committee conducted one of many

studies of corruption in that city's police department (see, also Fogelson, 1977; Goldstein,

1977:214; Williams, 1973). The presence of brutality and, until the shock of the recent

Rampart scandal, the apparent absence of money corruption during the recent history of

the Los Angeles Police Department has been reported by Bobb, et al. (1996), the

Christopher Commission (1991), Domanick (1994), Gates (1992:85-87), Rothmiller and

Goldman (1992), and Skolnick and Fyfe (1993). Although the Los Angeles County

Sheriff's Department recently suffered a narcotics-related corruption scandal, the major

focus of the Kolts Commission's study and subsequent reports has been brutality and other

abusive behavior (Kolts Commission, 1992; Bobb, et al., 1993, 1994a, 1994b, 1995,

1996), as it

In the years since Chevigny wrote this, two notorious incidents (the sodomy

inflicted upon Abner Louima in a precinct restroom and the Bronx shooting death of

Amadou Diallo) have affected this benign image.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 41

was in Cohen's (1980) study of policing in neighboring Long Beach. In New York, Daley

(1978), Kelly (1992), Maas (1972), McAlary (1989), Mollen Commission (1993, 1994),

Murphy and Plate (1977), Schecter and Phillips (1973) have all reported on corruption and

efforts to deal with it, as have the Philadelphia Police Study Task Force (1987) and

Rubinstein (1973) in their analyses of Philadelphia police.

Several likely reasons for this variation between older, eastern (and eastern-style)

cities and more recently settled areas may be identified. It is likely, for example, that this

apparent geographic variation is actually cultural, having more to do with cities' histories

and populations than with where they are. The major police scandals affecting the western

cities most similar in demographics and developmental history to eastern jurisdictions B

Seattle, Portland, San Francisco, Oakland, and Denver B have involved profit motivated

misconduct rather than use of force (see, Fogelson, 1977; Smith, 1965). In these, and in

diverse eastern jurisdictions B Boston, Chicago, Newark, New York, Philadelphia, and

Washington DC B there typically exists a tradition of Wilson's (1968) "Watchman Style"

of policing, in which officers exercise great discretion in fitting their activities to distinct

ethnic and racial communities characterized by highly decentralized political leadership

(see also Sherman, 1978). In such places, profit motivated misconduct may arise for

several reasons related to community and police organizational culture:

- Local police are mandated to enforce locally unpopular laws enacted by

distant and rural dominated state legislatures (e.g., limiting gambling; sale

and consumption of liquor; prostitution) (Haller, 1976; Wilson, 1963). In

such cases, officers are likely to cede to local will, and to engage in

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 42

pervasive organized corruption (Sherman, 1978) involving the de facto

licensing of these activities in return for bribes and kickbacks (Miller, 1975,

1977; Murphy and Plate, 1977; Wilson, 1963). The "pads" B or highly

organized monthly payments by gamblers to New York City vice officers B

exposed by Officer Frank Serpico (Maas, 1973) and the Knapp

Commission (1972) are perhaps the best-known modern example of this

form of profit motivated misconduct.

- The laws of many eastern states historically have reflected a narrow,

colionial era, religious view of propriety that not found in states in which

criminal codes and other statutes were more recently adopted. Routinely

violated "blue laws" that prohibited virtually all commerce on Sundays, for

example, were cited by the Knapp Commission, 1973) as a major source of

profit motivated misconduct in New York.

- Many of the people who populate eastern inner-cities have immigrated

from places characterized by great oppression, and have settled "among

their own," in distinct ethnic communities. Wilson (1985:162) argues that

such persons were taught by their experience in other places to distrust

government and to regard "[g]overnmental integrity [as] an implausible

abstraction." The underground cultures and economies that developed in

these conditions frequently are characterized by graft and gratuities to

untrustworthy public officials in return for permission to break laws they

had no part in enacting and that they regard as arbitrary.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 43

- The great poverty, high rates of crime and violence, and alienation in some

areas of these cities are associated with markets for drugs and other illegal

goods and services that simply do not exist in more homogenous and

uniformly prosperous jurisdictions where legitimate opportunities are in

great supply. Officers B perhaps especially young suburbanites who may

not identify with or understand underclass problems and cultures, and who

may wrongly stereotype all of their clientele as criminal B may exploit

these markets through theft, bribery, and direct involvement in drug dealing

(Kelly, 1992; McAlary, 1989; Mollen, 1993, 1994).

Until very recently, by contrast, the residents of recently developed western areas

typically have been homogenous immigrants who have left other parts of the United States

in search of the good life and good government. In post-World War II Los Angeles,

Chevigny (1996) and Domanick (1994) suggest, a major job of the police was to see that

rapid western growth and migration were unimpeded by crime or the obtrusive presence of

people some might see as undesirable. Hence, excesses in furtherance of these interests B

harassment, brutality, unlawful arrests B were tolerated. Conversely, profit motivated

misconduct B which had blotted the Los Angeles Police Department before the rapid

growth of the last half-century B was sought out and punished so vigorously that it

virtually disappeared. In Wilson's (1968) terms, William Parker's Los Angeles Police

Department (like the formerly corrupt Oakland Police Department he studied) shifted from

"Watchman" to "Legalistic" policing. Unlike the Oakland department, however, the

LAPD apparently left officers' discretion in use of force relatively unchecked.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 44

Looking for Individual Correlates of Police Misconduct

This variation over place and time does not tell us why some officers B and not

others B end their careers dishonorably. Even in the most troubled police departments, it

would appear that, while many officers may tolerate their colleagues' excesses and profit

motivated misconduct (Christopher Commission, 1991; Knapp Commission, 1972), most

officers do their work without using their hands either to brutalize citizens or to dip into

the till. Thus, officers' individual characteristics or life experiences may also distinguish

deviant officers from their colleagues, as well as among the types of deviance in which

officers participate.

Perhaps because police were so long chosen from a pool of people whose

characteristics and backgrounds did not vary much (e.g., McManus, 1969: 74-96), the

literature on the relationship of individual characteristics and deviance is somewhat

limited. Frequently mentioned in the literature of police behavior generally, however, are

such individual variables as gender, race, age, education, intelligence, physical fitness, and

length of service.

Thus B especially as police agencies have become more diverse over

An additional relationship that has been discussed in prior literature is that

between officers' attitudes toward their work and clientele and their job performance

(Friedrich, 1980; Neiderhoffer, 1969; Reiss, 1971; Muir, 1977; Worden, 1989). Because

the proposed study is retrospective, it is not possible for us to collect the attitudinal data

needed to examine this question. Even absent an empirical test, two propositions seem

reasonable: cynical officers are more likely than idealistic officers to become involved in

corrupt activities; and officers who have lost faith in the justice system may be prone to

engage in punitive on-duty abuse (see, e.g. Klockars, 1980; Skolnick and Fyfe, 1993).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 45

the last 25 years B it may be fruitful to examine the relationships between these variables

and officers' misconduct.

Gender and Police Behavior. Aside from occasional arrests of corrupt female

officers (e.g., Kappeler, Sluder, and Alpert, 1994: 226), there is little evidence concerning

the relationship of officer gender to profit motivated misconduct. Most comparisons of

male and female officers have found that male officers are more aggressive enforcers than

females (Bloch and Anderson, 1974; Forst, Lucianovic, and Cox, 1977; Melchionne, 1974;

Sherman, 1975; Worden, 1989), and that males are more likely than females to use force in

the course of their work (Grennan, 1987; Horvath, 1987). Similarly, Waugh, Ede, and

Alley (1988) reported that female officers in Queensland, Australia, were less likely than

males to be subjects of complaints, but Hickman, Piquero, and Greene (2000) found that

women officers in Philadelphia were disciplined slightly less often than their male

colleagues. Fyfe, et al., (1998), however, found no differences in the disciplinary

experiences of male and female New York City officers.

All this evidence might cause one to suspect that female officers would become

involved in misconduct less often than males and that, as the representation of female

officers, supervisors, and commanders in a department or unit increased, its culture would

become less tolerant of wrongful behavior. Waugh, Ede, and Alley (1988)also reported,

however that they women officers were just as unlikely as males to call to official

attention misconduct by their colleagues. This latter conclusion appears consistent with

one of Felkenes= (1991) key findings. He reported that female officers= adherence to a

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 46

Los Angeles Police Department culture that regarded the police as an isolated minority at

war with press and public was just as great as that of male officers. Thus, the relationships

between gender and misconduct are unclear, as are the nature of the effects upon police

culture and tolerance for wrongdoing of the presence of women in officers= ranks.

Race and Police Behavior. Cohen and Chaiken (1972, 1973) found that black New

York City officers with high IQs had above average rates of departmental misconduct.

Since this finding involves officially reported misconduct, it is possible that these officers

were victims of differential rule enforcement. It is also possible that these officers B then

even more underrepresented than now in relation to the city's African-American

population B were assigned to sensitive duties (e.g., vice; narcotics) in which opportunities

to engage in misconduct were greater than average. This explanation B race confounded

by assignment and rank B gains plausibility given two related findings. Fyfe (1980c)

found that black NYPD officers were more likely than whites to use deadly force because

of racially differing patterns of assignment, rank, and residence. On average, black

officers worked and lived in more dangerous areas than white officers, and they were more

often in front line jobs than in supervisory or staff positions. Consequently, they fired

their weapons more than their white colleagues.

More recently, Fyfe, et el., (1998)

reported that black New York officers= disparate rates of disciplinary action also were

associated with differential patterns of assignment, rank, and off-duty behavior.

Friedrich (1980) conducted a secondary analysis of observational data gathered by

Black and Reiss (1967) in an effort to link individual characteristics of police officers to

See Geller and Karales, 1981, who reported a similar pattern in Chicago.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 47

behavioral outcomes. Friedrich (1980) found that patrol teams made up of two black

officers used force against suspects at more than twice the rate (6.8%) of mixed or white

patrol teams in Boston, Chicago, and Washington DC.

All of this evidence suggests that the experience of black personnel is quite

different from that of white officers, and is certainly worth exploration in this study.

Age and Police Behavior. For generations, the conventional police wisdom held

that younger recruits made better officers. According to O.W. Wilson:

Important advantages are gained by recruiting young

men into police service. The older the man, the greater the

likelihood that he has experienced failure in some field of

activity. Successful police forces cannot be built of men who

have been unsuccessful; and who, as a consequence, may have

established undesirable patterns of thought and conduct. The

older man also is more likely to have skills and experience that

may tempt him to leave the police service when conditions of

employment in other fields seem more favorable. The man who

lacks other skills and who has not experienced service in other

activities is more likely to be contented in police service...

...The disadvantages of the immaturity of youth should

be discounted. Time and experience quickly correct the

immaturity of otherwise well-qualified men. When wisely

selected, the young recruit will not be guilty of serious errors of

judgment, and proper assignment during the early years of his

service will guard against his offending the more critical citizens

and protect him from situations where his youth may be an

advantage (Wilson, 1963: 138)

According to Leonard and More:

There is strong testimony for fixing the maximum age

limit no higher than 25. There is strong evidence indicating

candidates between 21 and 25 make the best material for

officers. They learn quicker and make better records in the

Academy than do officers who are past 25. It is the belief also,

that the younger person, after a few years of experience, is a far

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 48

better police officer than an older officer with the same amount

of experience. Young officers are more readily trained than are

men of 30 or over. Furthermore, failure to take police action has

been found to be due not so much to lack of maturity as to lack

of experience in similar situations (Leonard and More,

1993:456).

The NYPD's internal studies would justify somewhat more cautious recommendations.

In 1994, the department reported that applicants under 22 were less likely than older

applicants to be screened out of the candidate pool for "character reasons." Subsequently,

however, 20 and 21 year-old officers were disproportionately washed out of the Police

Academy; placed on suspension, modified assignment, probation, or in special monitoring

programs designed to counsel and provide close supervision of problem officers; dismissed, or

arrested. Perhaps, it was reasoned, some young candidates made it through screening simply

because they had not yet been adults long enough to demonstrate their unsuitability B and that

they subsequently did so after they had been hired (O'Sullivan, 1994, sections 1.1, 3.7.4). This

finding was consistent with an earlier finding by Cohen and Chaiken (1972, 1973) that

younger officers appointed to the NYPD in 1957 were more likely than their peers to have

been subjects of citizens' complaints between then and 1968, but Fyfe (1978) and Alpert

(1989) reported that age had no effect on officers' use of deadly force in New York and

Miami, respectively. Clearly, the data on the issue of age and police behavior are not

definitive.

Prior Employment, Life History, and Police Behavior.

Cohen and Chaiken's work

confirmed the hypothesis that past performance is a good predictor of future performance.

They reported that records of dismissal in prior jobs and military discipline were associated

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 49

with internal police rules violations (1972, 1973). They also examined other characteristics

that one might expect to be associated with performance in policing (e.g., histories of

psychological disorders; prior traffic summonses), but found nothing of significance. There is

reason to suspect that this latter finding is associated with limited range in the predictor

variables used: at that time, applicants whose psychological or driving histories were far

enough off the norm to cause concern were routinely dropped from candidate pools (see, e.g.,

Niederhoffer, 1969:148). More recently, as suggested by the Mollen Commission's

(1994:112-115) findings concerning the prevalence of arrest histories among suspended and

dismissed officers, the bounds of acceptable behavior among candidates seem to have

expanded.

Education and Police Behavior. With a few exceptions (Swanson, 1977), the literature

has supported the cause of police higher education. Consistent with earlier findings (e.g.,

Cohen and Chaiken, 1972, 1973; Bowker, 1980, Kappeler, Sapp, and Carter, 1992) reported

that college educated officers received fewer citizen complaints than lesser educated officers.

The extent to which this pattern of findings may be confounded by assignment factors,

however, is unclear. There is evidence that well-educated officers are likely to be moved out

of street assignments and into staff, supervisory, and detective assignments where exposure to

complaints is limited. In a 1996 study of a sample (n=258) of NYPD officers appointed

during 1988 and 1989, Eterno found that college educated officers, especially those who had

participated in a Police Cadet program were more likely than high school educated officers to

have advanced to detective or sergeant (see, also, Reuss-Ianni, 1983). There is also evidence

in NYPD's experience that it is difficult to distinguish among the effects of entry-level age and

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 50

education. NYPD Police Cadets B who then held baccalaureates and usually were 22 years

old or more at entry B "average fewer sick days, higher performance evaluation scores, fewer

[citizens' complaints], and receive more departmental recognition than non-cadet recruits"

(O'Sullivan, 1994, section 1.1).

Still, the evidence of an association between education and success in policing is

reasonably consistent. Most recently, Truxillo, Bennett, and Collins (1998:270) used a

retrospective longitudinal design in a cohort analysis of 84 police officers over a ten year

period in an unnamed municipal police agency. Subjects were administered a written survey

instrument, on which they were asked to indicate their level of education, grade point

averages, degrees earned, and number of credit hours of criminal justice education (Truxillo et

al., 1998). The researchers found that college training was moderately (and significantly)

associated with both promotions in rank (average r = .31), and supervisory ratings on job

knowledge dimensions (average r = .25). However, disciplinary problems among officers

were not significantly related to either a two year or four year college degree (Truxillo et al.,

1998).

Military Service and Police Behavior.

It has long been assumed that military service

provides both the discipline necessary for success in a police career and an opportunity for

police administrators to determine whether candidates for the police service take well to

service in uniform. In New York City, this view is reflected in laws and policies that have

granted extra credit for military service on police entrance and promotional examinations.

Certainly, there is much to be said for such veterans= preference simply on grounds of

fairness and desert: veterans= preference is a way of compensating military veterans for their

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 51

service to the nation and for removing themselves from civilian labor pools and career tracks

during the course of their service. Independent of this issue, however, if the assumption that

military service makes for a better police officer is correct, one would expect that military

service would be a more frequent life experience among our study=s control officers than

among its study officers.

Intelligence and Police Behavior. Cohen and Chaiken reported that, like education,

officers' scores on standardized IQ tests was associated with advancement through the ranks

and with departmental recognition (which may, in part, have been an artifact of the policy of

requiring officers to write their own requests for departmental recognition, an activity in

which verbally skilled officers presumably have an advantage). To the extent that

performance on police entrance examinations may be treated as a surrogate for intelligence,

Cohen and Chaiken (1972, 1973) reported that it predicted only performance on subsequent

promotional examinations.

Academy and Probationary Performance and Police Behavior. Cohen and Chaiken's

(1972, 1973) work suggested that officers' entry level training and probationary experiences

were predictors of career success. They reported that officers those who did well in the Police

Academy and on probationary evaluations were more likely than their colleagues to have

advanced through the ranks and win awards, and were less likely to have engaged in

misconduct or to have been disciplined.

Fitness and Police Behavior.

Eterno (1993) reported that degree of officers' physical

fitness was positively associated with rates of arrest and negatively associated with frequency

and duration of sick leave and with the frequency of police motor vehicle accidents and line of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 52

duty injuries. He reports, however, no associations between fitness level and frequency of

citizens complaints and departmental discipline.

Summary

This survey of police misconduct literature B like the findings described later in this

report B shows that police have engaged in the same kinds of rule-breaking behavior as

private citizens and, in other cases, that their positions of public trust have given them special

access to other opportunities for rule-breaking. This survey also suggests the differences

between what has traditionally been regarded as police deviance and deviance that happens to

be committed by police officers B but it also illustrates the difficulty of drawing a bright line

between the two. This section of the report also defines organizational and personal variables

that may predict police misconduct and that, therefore, help to focus our analyses.

Limited range of new officers' degree of fitness apparently is not a factor in

this finding. During 1990-93, 9,222 recruit officers were hired by the NYPD; only one

washed out for failing to meet physical standards (Eterno, 1993:2).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 53

IV. THE RESEARCH SETTING

The NYPD is the largest law enforcement agency in the United States and, by any

measure, is massive. With a personnel complement that has ranged as high as 41,000 officers

(presently about 37,000) in addition to 14,500 civilian employees, the NYPD is more than

three times as large as the Chicago Police Department, the nation=s second largest agency.

The department was founded in 1845, and is headed by a single police commissioner, who is

appointed by the city=s strong mayor.

THE POLICE COMMISSIONER

The police commissioner, or APC,@ is a civilian appointee of the New York City

mayor and is theoretically insulated from electoral politics by five-year terms of office that

overlap the four-year mayoral terms. In practice, however, newly elected mayors regard this

top police job as a key position in their cabinets, and almost invariably ask incumbent police

commissioners to resign so that they may be replaced with persons of the new mayor=s own

choosing. Mayoral involvement in day-to-day police operations has varied with the

philosophy of individual mayors, with Rudolph Giuliani surely being the most Ahands-on@

mayor in generations, even before the great increase in his national profile following the

September 11, 2001 tragedies. Both New York=s mayors and City Council apparently have

scrupulously limited their input to what Goldstein (1977) would regard as legitimate political

influences on policing, and it has been a half-century or more since New York mayors were

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 54

accused of corruptly or otherwise improperly influencing police policy or practice.

During

the period covered by this study (1975-1996), seven men served as PC.

Raymond Kelly, the

current PC, is a career NYPD officer who was appointed to the PC=s position in 1992 by

Mayor David Dinkins. He then left in 1994, when he was asked to resign by the newly

elected mayor, Rudolph Giuliani, who replaced him with William Bratton. Reputedly because

Giuliani resented the attention and credit Bratton received when New York=s crime rates

subsequently declined dramatically (Bratton, 1998), Bratton resigned and was replaced by

Howard Safir, the city=s fire commissioner, a former official of Drug Enforcement

Administration and the United States Marshals Service, and a longtime associate of Giuliani.

Upon Safir=s resignation, Mayor Giuliani appointed Bernard Kerik, his corrections

commissioner, to the PC=s position. Kerik left office at the end of Giuliani=s term, and newly

elected Mayor Michael Bloomberg appointed Kelly, who had served as his advisor on police

matters and who left a lucrative private security position to resume his former job. During his

This is not to suggest that mayors have not made serious mistakes related to

NYPD policy and practice. In the early 1970s, it was convincingly alleged that Mayor

John Lindsay ignored allegations of police corruption made by Frank Serpico and David

Durk. He allegedly did so not because he himself was also involved in improper conduct,

but because he was anxious to avoid alienating the police during a period of great civil

unrest (see Maas, 1973).

Four of the PCs who served during the period of this study (Michael Codd,

Benjamin Ward, Richard Condon, and Raymond Kelly) had been career NYPD officers

before their appointments. One had extensive police experience in other jurisdictions

(Lee Brown, who had been police CEO in Multnomah County, Oregon; Atlanta, and

Houston), and subsequently served as the mayor of Houston; and one (William Bratton)

had served as chief of the former New York City Transit Police and as police

commissioner in Boston, and is now the LAPD=s chief. One (Robert McGuire) had been

a prosecutor and successful attorney. Two other high ranking NYPD officials held the

PC=s position for a few days, as an honorific at the ends of their predecessors= terms.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 55

hiatus from the PC=s job, Kelly, who holds a master=s degree from Harvard, as well as J.D.

and L.L.M. degrees, also served with distinction at the federal level. He oversaw the

reorganization of the police service of Haiti, and was both an Undersecretary of the Treasury

and the Commissioner of the U.S. Customs Service.

NYPD commissioners are extremely powerful police chief executives, whose policies

and philosophies B or the lack thereof B have great effects on the operations of the

department. They are free to appoint their top staff, including 14 civilian deputy

commissioners,

and a raft of civilian assistant commissioners and unit directors. In addition,

all NYPD ranks above captain,

the top civil service rank, are filled at the discretion of the

police commissioner from among officers at the captain rank. This gives the PC enormous

influence over more than 800 captains and above, all of whom know that future advancement

and even continued incumbency are dependent on remaining in the commissioner=s good

graces. Reform commissioners, such as Patrick V. Murphy, who was appointed during the

Serpico-era corruption scandals, and William Bratton, who was appointed to lead Mayor

Giuliani=s war on crime, have made extensive use of their authority to reshape the agency to

their philosophies. Most others have not, and have demoted officers above captain or, more

typically, asked for their resignations on threat of demotion, only in cases of serious

misconduct or administrative failure. The ability of the current commissioner, Raymond

Kelly, to reshape the department=s

These include the First Deputy Commissioner and the Deputy Commissioners

for Administration; Strategic Initiatives; Counter Terrorism; Intelligence; Operations;

Public Information; Community Affairs; Labor Relations; Trials; Equal Employment

Opportunity; Legal Matters; Management and Budget; and Training.

These include deputy inspector; inspector; deputy chief; assistant chief; chief;

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 56

and chief of department.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 57

top rung without acting in a draconian fashion was severely curtailed by the promotion of

more than 150 people to ranks above captain during his predecessor=s last several weeks in

office.

Unlike many other large police agencies, the NYPD is not subject to a labor arbitration

process involving individual cases. Consequently, the only forum available to officers who

seek to appeal the PC=s disciplinary actions against them is the civil court. In this setting, the

PC=s disciplinary actions, including terminations and dismissals, typically are regarded as

matters of administrative discretion, and rarely are overturned.

DIVISION OF LABOR IN THE NYPD

Figure IV-1 is the NYPD=s organization chart. Although the department has become

more specialized over the last decade B creating new units to deal with specific problems;

expanding specialized units B its basic field command structures remain its 76 patrol

precincts. The Precincts

With rare exceptions, new NYPD academy graduates are assigned to work in

uniformed duty in precincts after the completion of field training. It is the precincts that

supply the core police services of preventive patrol and responding to calls. The precincts

most frequently portrayed in dramatizations of NYPD work typically are extremely busy units

in the parts of the city with which most tourists are familiar and/or which are marked by

extremes of poverty and crime.

These precincts represent only a fraction of the city. Especially in the outer boroughs

(Brooklyn, the Bronx, Queens, and Staten Island), New York City includes many areas that

are much like the residential and small business neighborhoods to be found in any American

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 58

city. Because the city does vary so much, however, it cannot be said that the work of all

police officers is comparable, in terms of volume, danger, temptation, or virtually any other

aspect. Within the NYPD, therefore, the precincts= numerical designators are a form of

shorthand that conveys a wealth of information to experienced police officers about a

community or a colleague=s assignment. The 111

Precinct B the One-Eleven B is regarded

as a plum patrol assignment by many officers, because the precinct includes relatively

prosperous and untroubled communities within easy commuting distance of the Long Island

suburbs in which many officers reside. The Four-Four, by contrast, is considered by most

officers to be a challenging assignment. Located in the diverse, and largely poor section of the

Bronx from which Yankees= owner George Steinbrenner regularly threatens to move his

team, it is a place where officers learn their business quickly, and in which violence is no

stranger. In 1990, the 44 suffered 89 homicides, while the 111 suffered four; in 2002,

following the recent dramatic decreases in New York crime, there were 23 murders and

manslaughters in the 44, compared to two in the 111. These figures are reflected in the

involuntary separation data we collected; during the years we studied, 24 officers of the 44

Precinct were involuntarily separated from the NYPD, as compared to three from the 111.

It should not be a surprise that officers in some NYPD precincts and units seem to get

into trouble more often than those from others. Fyfe (1980b) reported that the risk that police

officers would become involved in shootings varied similarly across police precincts, with the

greatest shooting frequencies occurring within inner-city precincts like the 44. Similarly, most

of the corruption scandals suffered by the NYPD have been centered in inner-city areas. The

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 59

BLANK PAGE FOR NYPD ORGANIZATION CHART

Figure IV-1: New York City Police Department Organization Chart

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 60

1970s Knapp Commission scandals had their roots in the South Bronx, where Officer Frank

Serpico disclosed corruption among plainclothes officers assigned to vice duty (Maas,

1973). The ABuddy Boys@ scandal of the 1980s (McAlary, 1989) was focused in the

Bedford-Stuyvesant area of Brooklyn that serves as the seeting for many of Spike Lee=s films,

and the more recent Mollen Commission scandals involved officers assigned to ghetto areas in

Manhattan, Brooklyn, and the Bronx (Mollen, et al., 1994). In years past, the NYPD used a

complicated regression equation, the Post Hazard Plan, to deploy officers to the precincts in

accord with empirical measures of their workload. (Fyfe, 1981b). The ordered ranking that

resulted from this was divided into three slightly uneven segments, to produce three categories

of precincts B high experience, medium experience, and low experience B in which officers

could work their way into low experience AC@ precincts (such as the 111) or special

assignments by successfully serving in high (AA@) and medium (AB@) experience precincts.

Over the years, however, the ratings came to be seen as insulting by residents of A precincts

which, with the exception of the two precincts in the Times Square area, were all largely

inner-city communities. The department itself apparently came to see these ratings as overly

rigid, and based too heavily on objective measures (e.g., street crime; calls for service) that

gave too little attention to such qualitative factors as community diversity and the tensions that

sometimes accompany it.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 61

Hence, the NYPD=s more recent assessment of the relative challenges within its

precincts is more readily grasped by who is assigned to command them. At this is writing,

eight of the 76 precincts are commanded by inspectors

(two steps above captain, one step

below deputy chief), and are considered to be the most challenging and sensitive. Beneath

them are the 24 that are commanded by deputy inspectors,

and the remaining 44 are

commanded by captains,

the lowest precinct commander rank.

Midtown South and the 34 in Manhattan; the 44 in the Bronx; the 67, 75, 77,

and 84 in Brooklyn; the 115 in Queens..

The 9, 13, Midtown North, 19, 23, 26, 28, 30, 32, and 33 in Manhattan; the 46,

47, and 52 in the Bronx; the 66, 69, 70, 72, 79, 83, 88, and 90 in Brooklyn; the 103, 109,

and 114 in Queens.

The 1, 5, 6, 7, 10, 17, 20, 22, 24, and 25 in Manhattan; the 40, 41, 42, 43, 45, 48,

49, and 50 in the Bronx; the 60, 61, 62, 63, 68, 71, 73, 76, 78, 81, and 94 in Brooklyn; the

100, 101, 102, 104, 105, 106, 107, 108, 110, 111,112, and 113 in Queens; the 120, 122

and 123, which encompass all of Staten Island.

Despite the apparently reduced reliance on crime rates as a method of distinguishing

among the precincts, it would be meaningless to classify precincts in ways that do not involve

objective differences among types. Figure IV-2 displays some of the demographics for these

three precinct types. The figure shows that, on average, Inspector and Deputy Inspector

Precincts, which have mean populations of 212,830 and 180,960, respectively, are

considerably larger than Captain Precincts (142,770). Even these differences may hide far

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 62

greater discrepancies in actual service populations: several of the Inspector and Deputy

Inspector Precincts (Midtown South, Midtown North, the 19 and 84) have very large visitor

populations that are not included in the resident population data. In addition, the figure shows

that the Inspector and Deputy Inspector Precincts include higher percentages of populations

who are poor; on welfare; living in single parent female headed households; or are foreign

born or Latino or black. Clearly, as one would expect in New York City, there is a great

range of diversity within the precincts and among these three types.

Although it has long been argued that police should be representative of the

communities they serve (Myrdal, 1944; Landrum, 1947; President=s Commission, 1967;

National Advisory Commission, 1968; Alex, 1969; Commission on Accreditation, 1994: Ch.

31), the NYPD is forbidden from making assignments to precincts or other departmental units

the basis of race or other demographic variables.

Thus, even though a look at the officers

who appear at roll calls suggests that there is a relationship between the characteristics of

precinct populations and those who police precincts, data on officers= race and assignment are

not available.

Indeed, the City recently lost a civil suit brought by a group of African-

American officers who argued that their transfers to the 70

Precinct following the

assault upon Abner Louima in the precinct=s bathroom were racially discriminatory.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure IV-2: Characteristics of New York City Precincts

49.2

29.3

33.6

24.8

10.9

37.5

17.9

24.1

13.2

21.3

9.6

28.6

14.2

18.3

12.3

11.6

22.2

16.3

35.4

10.7

25.6

24.5

23.8

% Black % Latino % Foreign Born % Unemployed % Single Fem

% Welfare HH % below

Poverty

% Low

Education

Inspector Pct D.I. Pct Captain Pct.

Fyfe and Kane -- 63

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 64

Other Departmental Units

Although patrol typically is described as the backbone or eyes and ears of policing, it

is the department=s entry level, and has historically been the assignment that many officers

have sought to leave (see, for example, Neiderhoffer, 1969). As Figure IV-1 suggests, the rest

of the NYPD includes hundreds of line and staff units that vary dramatically in their missions

and in the dangers, both physical and ethical, that they present to their personnel. To make

sense of them in the context of this work, we employed in some of our analyses a command

typology that was used by Fyfe (1981b) in analyses of shooting rates by officers in different

assignments. It is as follows:

Inspector Precincts: All precincts designated for command by inspectors.

Deputy Inspector Precincts: All precincts designated for command by deputy

inspectors.

Captain Precincts: All precincts designated for command by captains.

Police Academy/Field Training Units: Includes the Recruit Training School and the

units to which officers are assigned for field training at the end of their academy

studies.

The senior officer subjects of our study were appointed in 1946, and the junior

officers were appointed in 1996. During those 50 years, the NYPD closed precincts to

which some of the officers in our study had been assigned at various times during their

careers. In such cases, we classified these precincts on the basis of the current

classifications of the precincts into which they were merged. Most of the territory

covered by the old 16

Precinct, for example, is now a part of Midtown North, a Deputy

Inspector Precinct. Thus, for our purposes, the Old 16

is also a Deputy Inspector

Precinct.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 65

Proactive Investigative Units: Includes the Narcotics Division and the Street Crime

Unit (ASCU@) and its various predecessors (e.g., City-Wide Anti-Crime Section).

These are plainclothes units that, in Fyfe=s (1981b) work, were found to have the

NYPD=s highest shooting rates. Unlike other units that typically investigate crimes

after the fact, these units engage in undercover activities designed to flush out

criminal behavior (e.g., by attempting to arrange drug buys or sales; by using decoy

techniques) or to survey suspicious persons and high crime areas in hopes that they

will encounter crime as it occurs. The spontaneous, low visibility nature of their work

historically has produced large numbers of arrests, relatively high levels of violence

and, at least in the case of narcotics enforcement, a disproportionate amount of profit-

motivated misconduct.

Detective Bureau/Warrants Squads: Includes detective units charged with

investigating crimes that have already occurred and/or with apprehending persons who

are wanted on arrest warrants. This reactive work typically lends itself to greater

planning than is possible where proactive work is concerned. Despite the glamorous

image of television and movie detectives, it also involves relatively low levels of

violence.

Organized Crime Control Bureau Units other than Narcotics: Since the 1970s, the

NYPD=s Narcotics Division has been a part of the agency=s Organized Crime Control

Bureau (AOCCB@), which has also included such units as the Social Club Task Force;

the Auto Crime Division (which focuses on stolen car rings and Achop shops@); The

Money Laundering Unit; the Public Morals Division (which enforces laws related to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 66

gambling, prostitution, and alcohol; as well as participants in joint federal/state/local

organized crime task forces. The work of OCCB=s narcotics units, however, has

historically been distinguishable from that of these other units in terms of both

volatility and exposure to opportunities for corruption. No police work is without

peril, but public morals work in New York typically involves planned encounters with

people who are unlikely to fiercely resist the police because their offenses (e.g.,

gambling; serving liquor after hours) do not carry the severe penalties associated with

drug or violent crime. Further, although OCCB work other than narcotics enforcement

historically has been a source of profit-motivated misconduct, over the last generation

the department has restructured it in ways that minimize the opportunity for

corruption.

The nature of narcotics enforcement, however, virtually dictates that

officers work in corruption-prone, low visibility, settings. It also involves encounters

Until the early 1990s Mollen Commission scandals, units charged with

enforcing gambling laws have been the focus of virtually all of the NYPD=s major

corruption scandals over the last 110 years. After the 1970s Knapp Commission

scandals, however, the department instituted a series of reforms that took away most, or

all, of gambling enforcement officers= discretion. Instead of working as independent

agents with the authority to seek out offenders on their own and to arrest or not arrest at

their own discretion, gambling enforcement officers were assigned into teams that, under

close supervision, generally conduct investigations only when citizens complain. With

these new limits went most of the opportunities to extort or accept bribes from gambling

operators. This same strategy was applied to street-level drug enforcement (see Murphy

and Plate, 1978). Over the years, however, this proved to encourage open-air drug

dealing by offenders who had come to learn that former Abuy and bust@ operations had

given way to lengthy investigations of persons higher up in the drug traffic, leaving them

effectively immune from arrest. By the late 1980s, Abuy and bust@ narcotics work

returned to New York City and is once again a major enforcement tactic (Bratton, 1998).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 67

with offenders who sometimes possess large amounts of money or fortunes in their

merchandise.

Special Patrol Units: Includes all uniformed units other than precincts. These include

Housing and Transit Bureaus, traffic units, Highway Patrol, various task forces, and

the Emergency Service, Aviation, and Harbor units.

Staff: Includes all non-field units. Headquarters units, the Police Academy, crime

laboratory, booking facilities all are included here.

The Internal Affairs Bureau

One staff unit is worthy of special note here, and should be taken into account in any

attempts to generalize from the findings in this study to the experiences of other agencies.

NYPD=s Internal Affairs Bureau (IAB) is itself larger than most U.S. police departments. In

2000, IAB processed 25,091 Alogs@ or complaints, 1,203 of which involved allegations that,

if sustained, could result in criminal charges or dismissal (NYPD/IAB, 2001:4, 20). As the

Knapp Commission (1972) suggested, however, limiting internal investigations to inquiries

into alleged misconduct that has already occurred greatly limits the effectiveness of police

integrity control efforts. Hence, the commission recommended that the NYPD adopt more

pro-active strategies and tactics that, like traditional police undercover operations, would

create the circumstances in which potential or suspected offenders would be given

opportunities to engage in corruption. IAB has embraced this recommendation with vigor,

and today engages in a variety of techniques and strategies for preventing and detecting

misconduct that are not practiced in other agencies. These should be taken into account in any

attempts to generalize from the findings of this study to the experiences of other agencies.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 68

Intelligence Section.

According to IAB:

The Intelligence Section of Internal Affairs is one of the primary

means for obtaining intelligence on corruption and misconduct,

and assisting field investigators with their cases. Its three

components are: the Voluntary Assistance Unit, which gathers

information from both uniformed and civilian members of the

service; the Field Operative Program, which relies on

cooperation from tenured members of the service; and the

Liaison Unit, which foster a strong relationship between the

Internal Affairs Bureau and federal, state, and local law

enforcement agencies (NYPD/IAB, 2001:4; emphasis in

original).

Perhaps the most interesting and controversial of these three operations is the Field

Operative Program. This involves officers and other personnel who are assigned to

departmental units but who, unknown to their colleagues, surreptitiously report to IAB on

serious misconduct and on conditions that are conducive to corruption.

Integrity Tests. Following traffic enforcement corruption scandals in the 1950s,

members of the NYPD=s internal affairs units intermittently were assigned to drive civilian

autos and to violate traffic laws in the presence of officers suspected of extorting traffic

offenders. This limited operation has since evolved into an extensive program of Aintegrity

testing.@ IAB notes:

An integrity test is an artificial situation created by investigators

to present an opportunity for a reaction by the subject

member(s) of the service. During the test, the subject is given

the opportunity to perform or fail to perform in a manner

consistent with legal and Department guidelines. Such tests

may be conducted randomly in response to a pattern of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 69

allegations, or targeted toward a specific investigation (New

York City Police Department, Internal Affairs Bureau, 2001:15).

An elaborate recent IAB integrity test involved a foot patrol officer. IAB received

information from several sources that the officer was extorting money from numbers runners

and other street-level gamblers. IAB then assigned two of its investigators to pose as

undercover OCCB officers. These two approached the officer on his post and showed him a

picture of a man they claimed was a big gambling operator who picked up money from local

establishments that quartered illegal gambling (the photo actually depicted another IAB

investigator). They asked if the officer had seen the man and, of course, he had not.

Periodically thereafter, the IAB investigators stopped by, and inquired of whether the officer

had seen the man in the photo. After several such encounters, the investigators arranged a

bogus arrest. They emerged from a local store pulling their handcuffed, shouting, and fiercely

resisting Asuspect@ behind them. The patrol officer approached, and offered to help. The

investigators gave him the keys to the suspect=s car and asked him to drive the car to the

stationhouse while they transported their prisoner there. The officer agreed but, instead of

driving to the station, made a stop on a quiet street. There, the video cameras that had been

hidden throughout the car caught him searching it and pocketing the large amounts of marked

money that had been planted in it by IAB.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 70

IAB conducts hundreds of integrity tests annually. IAB classifies the results of these

tests into four types. Passes are those in which test subjects perform as required by law and

department regulations. Procedural failures are those in which officers or other employees

are found to disobey department rules by, for example, treating citizens inappropriately.

Criminal failures are those in which, like the officer in the example above, employees take

wrongful advantage of opportunities to commit crimes. Supervisory failures are those in

which commanders or supervisors fail in their responsibilities by, for example, discouraging

or turning away citizens (actually IAB officers) who wish to lodge complaints against officers

or to complain about police service. Figure IV-3 presents the number and type of integrity test

failures during 1999 through 2001.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure IV-3: Integrity Test Failures, 1999-2001

Procedural Criminal Supervisory

1999 2000 2001

Source: New York City Police Department, Internal Affairs Bureau, Annual Report for Year

2000 (NYPD:2001, 15): Annual Report for Year 2001 (NYPD:2002, 5).

Fyfe and Kane -- 71

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 72

EDIT Program.. The EDIT Program (Enforcement, Debriefing, Intelligence, and

Testing) is an operation in which IAB conducts Aits own enforcement operations, debrief[s]

arrestees, and gain[s] valuable intelligence in those areas that have traditionally been

corruption prone@ (NYPD/IAB, 2001). Often, this occurs after patterns of allegations are

identified,

and/or involves arrests of people who are believed to be involved in corrupt

relations with police officers. After their arrests, these individuals are debriefed with an eye

toward Aturning@ them on corrupt officers (e.g., gaining their cooperation in building cases

against corrupt officers in return for promises of leniency). Figure IV-4 shows that, in 1999,

IAB conducted 280 EDIT operations involving 498 arrests; in 2000, EDIT=s 262 operations

resulted in 455 arrests; in 2001, EDIT conducted 300 operations with 543 arrests. Thus, IAB

probably is one of the few such units that makes cases against citizens, as well as police.

Certainly, it has been a major source of the disciplinary actions that resulted in the involuntary

separations studied in this report.

IAB used COMPSTAT=s computerized geographic information system

methodologies to identify such patterns.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure IV-4: Edit Programs, 1999-2001

280

262

300

498

455

543

1999 2000 2001

Operations Arrests

Source: New York City Police Department, Internal Affairs Bureau, Annual Report

for Year 2000 (NYPD:2001, 16): Annual Report for Year 2001 (NYPD:2002, 6).

Fyfe and Kane -- 73

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 74

THE NYPD=S DISCIPLINARY SYSTEM

This study does not describe the totality of the NYPD's disciplinary mechanisms.

Instead, it focuses upon separation from the service, the last stop in the department's elaborate

processes for assuring discipline. Discipline may be defined as the art and science of gaining

and maintaining compliance with official prescriptions. In the NYPD and other police

agencies, this purpose is served through:

- recruitment;

- selection;

- training and probationary screening;

- socialization and maintenance of an organizational culture supportive of

adherence to the NYPD's policies, practices, rules, and procedures;

- supervision and performance evaluation;

- career development programs;

- employee counseling and assistance; and, lastly

- punishment, including separation.

The punishments that may be administered to NYPD officers range from verbal

reprimands through suspensions and lost vacation days to involuntary separation. Because of

the interactions and interdependence of these processes, the frequency and nature of

punishment are affected by changes in earlier disciplinary mechanisms. These have varied

and evolved over time, often in ways that are hard to quantify. Consequently, some temporal

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 75

variation in disciplinary practices should also be expected. In addition, the NYPD includes

two tiers of punitive discipline: command discipline and formal charges and specifications.

Command Discipline

Command discipline is a semi-formal, non-judicial, system designed to help

commanding officers correct minor deficiencies and correct employees' behavior without

blemishing their records with the permanent stigma that may attach to charges and

specifications. The command discipline system has roots in the early 1970s recognition that

supervisors were reluctant to take punitive disciplinary action against officers because formal

charges B then the only available variety B permanently blotted officers= records for even the

most minor misconduct. Given the choice between doing nothing about minor misconduct

and permanently marking an officer=s record with a complaint B which virtually guaranteed

permanent relegation to patrol duty and exclusion from any specialized assignments B most

supervisors and commanders chose to do nothing. Thus, because the system was so

draconian, it usually was not used.

The command discipline system is highly decentralized, and allows local commanders

to punish officers for specified acts of minor misconduct that carry penalties of no more than

ten days loss of pay (e.g., tardiness; negligent vehicle accidents or loss of department

property; absence from assignment; wearing an improper uniform or equipment). Following

informal interviews with the uniformed or civilian employees involved, commanders decide

upon and impose penalties, which officers and civilian employees may decline. When they do

so, their cases go for formal proceedings at the department=s Trial Room.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 76

In 1995, in conjunction with attempts to decentralize disciplinary authority and to

enhance local commanders= accountability, the NYPD made more categories of misconduct

subject to command discipline (e.g., loss of shield, failure to safeguard a prisoner) and the

maximum penalty applicable under command discipline was increased from loss of five days

vacation to ten. Dismissible offenses are not subject to command discipline, so that this

system does not account for any of the involuntary separations in this study.

Charges and Specifications

More serious and/or chronic violations, as well as refusals or appeals of command

discipline findings or penalties,

are the province of the NYPD's formal disciplinary system,

which is where most of the officers we studied were found guilty of the offenses that led to

their separations.

Much like a military court martial system, this process includes

administrative proceedings that commence with the service upon an employee of formal

charges and specifications. Like criminal or civil actions, department charges are open to

negotiation from both parties, which may obviate the need for administrative hearings.

Employees are also free to bypass command discipline altogether and to insist

instead on the filing of formal charges and specifications and resolution of the matter via

formal administrative hearing.

Exceptions include wrongdoing by probationary officers, where the Police

Commissioner has great discretion. Although some lesser cases against probationers do

result in Trial Room proceedings, serious wrongdoing by probationers typically results in

summary termination by the commissioner=s fiat. In addition, many of the officers we

studied resigned or retired rather than face dismissible charges or were allowed to do so

in return for their cooperation in testifying against other officers, or as part of negotiated

case dispositions.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 77

The Department Advocate generally prosecutes these cases on behalf of the NYPD.

The Advocate's position is an attorney's assignment, and has been held by both uniformed

officers and civilian employees. Since 1995, all supervisors have been required to consult with

attorneys assigned to the Department Advocate's Office and receive their approval before

charges may be filed. In much the same way that assistant district attorneys consult with

arresting officers, the Advocate's staff is charged to confirm that a sufficient legal basis exists

for each charge and that all appropriate investigative steps have been completed. The

Department Advocate may decide not to proceed with charges and specifications,

recommending instead either command discipline or no charges at all.

There are two exceptions to this general pattern. Extremely serious cases (e.g.,

those also involving criminal proceedings) often are handled by a Department Special

Prosecutor rather than the Department Advocate (see NYPD, 1996). In addition, cases

substantiated by the Civilian Complaint Review Board and preferred against members in

the rank of police officer are resolved before the Office of Administrative Trials and

Hearings (OATH), a city agency independent of both the Personnel and Police

Departments.

Trial Room proceedings are open to the public, and follow rules of evidence and

procedure that differ from those found in criminal or civil courts, mainly because they allow

the admission of hearsay evidence. Department trials are heard by the Deputy Commissioner,

Trials (an attorney) or one of his Assistant Deputy Commissioners (also attorneys). Accused

employees are permitted counsel, and usually have attorneys provided by their labor

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 78

organizations. Cross-examination is permitted, and all proceedings are stenographically

recorded. The Trial Commissioner then presents a written report and recommendation to the

Police Commissioner, who renders the final decision, both as to the finding of fact and, when

guilty verdicts result, the penalty. There is no arbitration system for police administrative

disciplinary actions in New York, so that appeals must be to the courts. Although we made no

attempt to search systematically for the results of any such appeals, both general experience

and our data collection efforts suggest that reversals of NYPD dismissals and terminations are

rare.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 79

V. THE NYPD, 1975-1996

Any 20-year slice of the NYPD=s history is packed with noteworthy events. The

period we studied began at the tail end of an era of great turmoil in the NYPD. During 1971-

1973, several officers were assassinated by the radical Black Liberation Army, who engaged

in 20 or more exchanges of gunfire with NYPD officers (See Daley, 1973; Fyfe, 1978). At the

same time, the corruption scandals of the early 1970s were being closed out with the criminal

and administrative trials of many of the officers who had been implicated in the investigations

of the Knapp Commission and of the Special Investigations Unit of the NYPD=s Narcotics

Division. The period studied ended immediately after the Mollen corruption scandals, amid

accusations that the department=s aggressive law enforcement tactics were encouraging

officers to excesses.

THE KNAPP ERA

Prior to and during the Knapp/SIU era, it is fair to say, the department=s level of

integrity was highly questionable, and corruption was highly organized. The Knapp

Commission=s final report indicates that:

This historical section of this report is generally based on Bolz and Hershey, 1979; Bouza,

1990; Daley, 1973, 1978; Knapp Commission, 1972; Droge, 1973; Fogelson, 1977; Fyfe, 1978;

Gelb, 1983; Hamilton, 1975; Johnson, 1981; Lardner, 1996; Lardner and Reppetto, 2000;

McAlary, 1989; McNamara, 1967; Melchionne, 1974; Murphy, 1977; Murphy and Caplan,

1991; Murphy and Plate, 1977; NYPD Academy, 1972, 1974; Reppetto, 1978; Reuss-Ianni,

1983; Schecter and Phillips, 1973; Schlossberg and Freeman, 1974; Sherman, 1978; Walker,

1977, 1993; and on the experiences and communications of the senior author.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 80

We found corruption to be widespread. It took various forms

depending upon the activity involved, appearing at its most

sophisticated among plainclothesmen assigned to enforcing

gambling laws. In the five plainclothes divisions where our

investigations were concentrated we found a strikingly

standardized pattern of corruption. Plainclothesmen,

participating in what is know in police parlance as a Apad,@

collected regular bi-weekly or monthly payments amounting to as

much as $3,500 from each of the gambling establishments in the

area under their jurisdiction, and divided the take in equal shares.

The monthly share per man (called the Anut@) ranged from $300

and $400 in midtown Manhattan to $1,500 in Harlem.

When

supervisors were involved they received a share and a half. A

newly assigned plainclothesman was not entitled to his share for

about two months, while he was checked out for reliability, but

the earnings lost by the delay were made up to him in the form of

two months= severance pay when he left the division(Knapp

Commission, 1972:1).

***

Corruption in narcotics enforcement lacked the organization of

the gambling pads, but individual payments B known as scores B

were commonly received and could be staggering in amount...the

largest narcotics payoff uncovered in our investigation having

been $80,000.

Corruption among detectives assigned to general

investigative duties also took the form of shakedown of

individual targets of opportunity. Although these scores were not

in the huge amounts found in narcotics, they not infrequently

came to several thousand dollars.

Uniformed patrolmen assigned to street duties were not

found to receive money on nearly so grand or organized a scale,

but the large number of small payments they received present an

equally serious if less dramatic problem. Uniformed patrolmen,

particularly those assigned to radio patrol cars, participated in

gambling pads more modest in size than those received by

plainclothes units and received regular payments form

construction sites, bars, grocery stores and other business

establishments (Knapp Commission, 1972:2).

By way of comparison, a police officer=s salary at the time of the Knapp

Commission was about $1,000 per month.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 81

***

Of course, not all policemen are corrupt. If we are to exclude

such petty infractions as free meals, an appreciable number do

not engage in any corrupt activities. Yet, with extremely rare

exceptions, even those who themselves engage in no corrupt

activities are involved in corruption in the sense that they take no

steps to prevent what they know or suspect to be going on about

them (Knapp Commission, 1972:3)

These stinging revelations and the efforts of Patrick V. Murphy, the reform police

commissioner of the time, produced major changes in the NYPD during the years immediately

prior to 1975. The former APlainclothes Division,@ the gambling enforcement unit cited by

the Knapp Commission, was abolished. The Organized Crime Control Bureau was created,

with responsibility for gambling, prostitution, and alcohol enforcement, and for narcotics

enforcement, as well. The discretion of both narcotics and vice officers was greatly limited by

a general prohibition on self-initiated enforcement action. Uniformed officers were relieved

of responsibility for enforcing construction codes, liquor laws, sabbath laws (which required

most businesses to remain closed on Sundays), and most narcotics laws, all of which had been

identified by the Knapp Commission as sources of corruption. The Street Crime Unit, a

proactive unit that sought out violent crime, was created and proved extremely effective. Its

officers comprised one percent of the department=s personnel and made 16 percent of its

violent crime arrests, impressing the United States Department of Justice enough to be

designated an Exemplary Project (Abt, 1973).

For the first time in memory, police supervisors and commanders were held closely

accountable for the behavior of their personnel. Commanders were required to identify

corruption hazards in their precincts and units, and to develop integrity plans to keep their

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 82

officers in check. Commanders whose officers were implicated in scandal were transferred to

dead-end jobs, demoted, and otherwise pressured to retire.

Internal affairs was greatly

enhanced in both size and responsibility. Its proactive efforts, including the field operative

program, were put in place. The command discipline system was created, marking the first

time that supervisors and commanders could discipline officers without imposing the harsh

penalties and career-long stigma associated with Trial Room proceedings. In addition, in

response to a Knapp recommendation, Governor Nelson Rockefeller appointed a special

prosecutor, charged only with investigating and prosecuting corruption in the New York City

criminal justice system.

After a two-and-a-half-year freeze, the department began hiring vigorously in 1973,

enlisting more than 6,200 new officers in 23 months. These first post-Knapp recruits differed

in several ways from those who had gone before. Previously, the department had two titles for

those who entered the department: patrolman and police woman. Patrolmen did the uniformed

cops= job, but police women were not assigned to patrol duties, and were hired in very small

Murphy introduced an Aup or out@ policy, which presumed that 20 percent of

those at the rank of captain and above would be promoted or retired every year (see

Murphy and Plate, 1978).

During 1971 and 1972, the department hired as police officers only police

trainees (young men who held clerical police positions and who were appointed as police

officers when they turned 21 years old) and returning military veterans who were in the

service when they became eligible for police employment during earlier years.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 83

numbers to do specialized investigative work, matron duty, and service as youth officers. This

changed with the equal opportunity legislation and litigation of the early 1970s when, for the

first time, the department hired men and women on an equal basis as police officers. The first

substantial numbers of female officers entered the department at this time.

They entered to a new training curriculum, revised in accord with the

recommendations of a 1969 Law Enforcement Assistance Administration study (McManus, et

al., 1969). The training was far more rigorous than had been true in the past and, for the first

time, served as a part of the agency=s screening process. Prior to this time, appointment to the

department was a de facto lifetime sinecure, and probationary officers were terminated only

for the most serious misconduct. In some post-Knapp classes, however, recruit attrition was

as high as 20 percent, with recruits washing out because of failures in academic, physical, or

firearms and tactics training. Others were terminated as a result of poor performance in the

new three-month field training program that followed upon academic training. The New York

State Board of Regents evaluated the recruit curriculum as the equivalent of 35 undergraduate

semester hours. Firearms training changed its focus. Where it had formerly emphasized

accuracy in long distance single action shooting B which had virtually nothing to do with the

reality of New York City police work B it was now based on systematic analysis of shooting

incidents. Thus, it came to stress tactics designed to help officers avoid shootings, and to help

them survive the quick, close range encounters they could not avoid. This training, combined

with the issuance of the department=s first restrictive shooting policy in 1972, significantly

reduced both police shootings and police injuries and fatalities.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 84

The department developed a career path, designed to make eligibility for desirable and

specialized assignments contingent on successful prior service in high activity AA@ precincts.

This reduced the influence of the hook system, under which those who had connections to

highly placed police or governmental officials had historically been awarded the most

desirable assignments. The department introduced a performance evaluation system and an

early warning system designed to identify and counsel officers whose histories suggested that

they were violence prone. For the first time, it allowed officers to take educational leaves of

absence.

In addition to its firearms policy, the department issued a restrictive policy on vehicle

pursuits. Following the hostage situation depicted in the film ADog Day Afternoon@ and the

hostage taking and murders at the 1972 Munich Olympics, a veteran lieutenant, Frank Bolz,

and Harvey Schlossberg, a detective with a Ph.D. in clinical psychology, were assigned to

develop plans for police response to hostage and barricade situations. They came up with the

protocols that have since been adopted around the world.

This great emphasis on professionalism changed the culture of the department

dramatically. Elizabeth Reuss-Ianni (1983) studied the department, and concluded that its

former monolithic cop culture had split into two. The first, the street cop culture, adhered to

the old values of loyalty to one=s peers and unit, and was what remained of the department=s

pre-Knapp world view. The second, the management cop culture, was dedicated to

These protocols have been modified by the events of September 11, 2002,

which taught the painful lesson that submitting to hostage takers might no longer be the

best method of assuring survival.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 85

professional and objective standards of performance, and was practiced by the upward mobile

young supervisors and commanders who populated headquarters and who emulated Murphy=s

example. This split produced a clash in which street-level officers came to see themselves not

as part of the single entity that had existed prior to Knapp, but as an embattled minority that

was besieged not only by critics outside the department, but by their own bosses, as well.

Thus, when the first officers in our study were being forced out of the NYPD, the

department was coming out of one of the most tumultuous periods in its history. It had been

engaged in a street war with the Black Liberation Army, which ambushed and killed four of its

officers and wounded several others. It had been through two major scandals, for which many

of its officers still faced criminal and administrative trials; it had hired its first women patrol

officers; and had seen major internal reforms at virtually every operational and support level.

Then, in 1975, New York City went broke.

1975: THE FISCAL CRISIS

On June 30, 1975, after months of negotiating and begging for funds, New York City

laid off its junior 5,000 police officers, virtually all of the post-Knapp cohort in whom it had

invested so heavily. A few days later, the city announced that it had discovered the money to

rehire 2,000 of the officers, and it did so. The remaining 3,000 officers were offered

opportunities to be rehired in small groups over the next four years. Many came back, but

many others chose police careers elsewhere, or elected to permanently pursue other lines of

work.

The fiscal crisis and the layoffs had profound effects on the NYPD. During the

months that preceded the layoffs, the city and its police union had engaged in a series of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 86

negotiations in which it became apparent that the city was not operating in compete good

faith. In April 1975, the city announced that it would be forced by financial necessity to cut

300 officers. This group was identified, brought to the Police Academy on the appointed day,

and relieved of guns, shields, identification cards, and department manuals. Late in the day,

the mayor and the union president appeared on stage to announce that a last minute deal had

been struck: if the members of the Patrolman=s Benevolent Association would agree in a

ballot to work ten extra days without compensation over the coming year, the city would

somehow save enough money to avoid the layoffs.

This was obvious farce. Had police officers given up compensation for regularly

scheduled working days, the city would have saved ten days pay per officer, arguably

allowing it to set aside the funds to salvage the junior officers= jobs. That, however, did not

happen. Instead, the city agreed to a scheme that actually cost it more money: salaries and

benefits aside, putting 20,000 or more officers on duty for ten additional days is a great

expense, in terms of wear and tear on cars and other equipment, fuel, supervisory costs,

overtime that might be incurred as a result of arrest made on extra shifts, and increased

exposure to injury, disability, and other forms of liability. Nonetheless, the PBA=s members

voted to save their junior colleagues= jobs by agreeing to what they universally regarded as a

face saving sham.

Two months later, however, came the 5,000 layoffs, reduced a few days later to 3,000

by another sudden financial discovery. Prior to these layoffs, the PBA declined to take any

preventive action on grounds that the city had already broken faith with it. Worse, the city

insisted that the April ten-days-for-free arrangement was part of a separate deal, and went to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 87

court to compel the officers to stick to their end of that earlier bargain. The city prevailed and,

almost immediately, mean sick leave time for officers (who enjoy unlimited sick leave)

increased from 10 days annually to 18 days. The PBA argued that this increase was due to the

stresses of the increased workload imposed upon officers who now had to pick up on the hard

work formerly performed by their most junior colleagues. The administration of the police

department saw it, probably more accurately, as officers= way of getting back their ten free

days.

The schisms this caused were deep and lasted for many years. The cop culture cited

by Reuss-Ianni took hold and strengthened, comprised of officers who were convinced beyond

a shadow of a doubt that the city and the department had exploited them. Older officers were

angered, and resentful of their junior colleagues, for the ten extra days they had to work and

because many were transferred out of desirable jobs they believed they had earned, in order to

fill in on the departed rookies= beats. Many of the laid-off officers, including perhaps, many

with the most solid credentials and greatest career options, departed, never to return.

At the

same time, many of the managerial ranks= brightest stars left to become chiefs of police in

other jurisdictions around the country (e.g., Baltimore County; San Jose; Minneapolis;

Pontiac; Seattle; Kenosha; Racine) or to join the federal service.

The last recruit class hired prior to the layoffs entered the department in November

1974. During the next five years, as Figure IV-5 shows, the NYPD did not hire a single

Fyfe has run into laid-off officers who went on to police careers in Arizona,

California, Colorado, Florida, Kentucky, Maryland, New Jersey, Texas, Virginia, and

Washington DC.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 88

officer. During those years, it also lost the services of virtually every female officer hired

after

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 89

the court and legislative mandated that women be hired on an equal basis with men; more

than a quarter B 28 percent B of the 3,000 laid-off officers were women.

When the

Department recommenced hiring, however, it did so with vigor. Between November 1979,

when the first post-crisis class was hired, and the end of 1989, the NYPD hired 23,110 new

officers.

The layoffs had two great effects on police operations. First, they significantly

reduced the number of officers available for street patrols. Figure IV-6 shows that the number

of officers and supervisors assigned to patrol precincts declined significantly after June 30,

1975, the day of the layoffs. On that date, 17,541 sworn personnel were assigned to the

precincts; on June 30, 1977, that figure was nearly 20 percent lower (14,135; data not

available for June 30, 1976). It continued to decline until it reached its nadir in 1981 (11,453),

and returned to its former level only 20 years later, in 1994 (17,711).

At one point during this period, NYPD detectives needed a young female

officer to serve as a decoy in a plan to arrest a dentist who reportedly molested young

women patients while they were under anaesthesia. The department could not identify a

single suitable officer in its ranks, and the detectives were obliged to arrange a personnel

loan from the neighboring Yonkers Police Department.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure V-1: Police Officer Appointments to the NYPD, 1946-1996

500

1000

1500

2000

2500

3000

3500

4000

946

950

954

956

958

960

962

964

966

968

970

972

974

976

978

980

982

984

986

988

990

992

994

996

Fyfe and Kane -- 90

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure V-2: Precinct Officers and Supervisors, 1975-1996

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

975

976

977

978

979

980

981

982

1983

985

986

987

988

989

990

991

992

993

994

996

Fyfe and Kane -- 91

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 92

Second, it made police officers far less expendable than they had been in the

department=s halcyon days. Like a financially strapped university department head who

understood that she would lose a faculty line if she denied tenure to a marginal candidate, the

NYPD=s administration understood that separated officers would not be replaced. This may

have had implications for the department=s tolerance of misbehavior during those years.

THE 1980s

The 1980s were a period of rebuilding and growth for the NYPD. Hiring and

promotions proceeded apace. In 1983, the NYPD was the subject of a Congressional

investigation of allegations of brutality in the department. This preceded the appointment of

Benjamin Ward, the department=s first African-American commissioner, but produced little in

terms of discipline. Ward himself, however, was quite a disciplinarian. In 1985, a small

group of his officers was discovered used electronic stunguns to torture arrestees. The officers

were prosecuted, and Ward relieved from duty the entire supervisory and management staffs

of the precinct, division, and borough, as well as the NYPD=s Chief of Patrol. Also in 1985,

13 officers from Brooklyn=s 77

Precinct were criminally charged with stealing and

trafficking in drugs. This was a sign that, while the Knapp Commission and the reforms that

followed may have interrupted the widespread corruption related to gambling, there might still

exist pockets of organized corruption that were involved in the far more lucrative and, in the

traditional police view, far more venal drug trade.

Drug Testing

Again, in 1985, the NYPD initiated the first in a series of steps that have resulted in

one of the most comprehensive police drug testing programs in our experience. Positive

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 93

findings on drug tests result in mandatory charges and specifications, and ultimately in

separation from the department. The for cause segment of this Dole Test program charged

supervisors and commanders with responsibility for referring to testing officers and employees

when their behavior or the circumstances created reason to believe they were abusing

controlled substances. In addition, the department then began to test all applicants and

probationary police officers for evidence of drug use as part of the pre-employment screening

process, during training at the police academy, and just prior to the expiration of officers=

probationary periods.

The drug testing protocols have become increasingly more stringent over the last 18

years. In 1986, the NYPD began testing employees who were starting sensitive specialized

assignments, such as the Detective Bureau, the Internal Affairs Bureau, the Organized Crime

Control Bureau (OCCB), the Special Operations Division (including the Harbor, Aviation, and

Emergency Service Units), and the Highway Unit. In 1989, the department introduced drug

testing of all in-service personnel assigned to the OCCB, and began randomly testing ten

percent of the sworn staff of all other department units annually. In 1992, the department

doubled the size of this random sample to 20 percent. In 1993, it began to require that all

candidates for promotion to sergeant undergo drug testing; in 1996, this requirement was

extended to lieutenant candidates, and, in 1997, to captain candidates.

There is no doubt that

drug testing has greatly and directly influenced involuntary separation rates in the NYPD.

The department=s drug testing programs appear to be administered fairly and

very difficult to subvert. In related research (Fyfe, et al., 1998), we tested several

propositions about both random and for cause testing. First, we found that the annual

random samples of drug test subjects were just so, annually producing a 20 percent slice

of the NYPD that almost precisely matched the department=s demographics in terms of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 94

The Buddy Boys

The 77

Precinct encompasses a part of Brooklyn=s Bedford-Stuyvesant section and,

in the late 1980s, was found to be home to what Sherman (1974) would call a police rotten

pocket: a group of officers working together in a corrupt enterprise. The Buddy Boys= central

character was Officer Henry Winter, a fast-talking, platinum blond young man who had grown

up in New York City=s Long Island suburbs. Winter, who was assigned to steady midnight to

eight tours of duty in uniform, apparently began his career in deviance by meting out street

justice to those he believed were likely to be inadequately punished by the criminal justice

system, in the forms of beatings. Eventually, this treatment included burning drug suspects=

money or forcing them to flush it down the toilet. Having built a reputation among his

colleagues for these gaudily improper tactics, Winter soon recruited some into the same lines

of business and, in short order, his Buddy Boys (which group, in fact, included at least one

female officer) expanded their activities to theft, robbery, bribery, extortion, drug dealing. and

drug abuse. In time, other police arrested a drug dealer who claimed that he had been buying

protection from 77

Precinct officers. The then-Internal Affairs Division proceeded to

rank, gender, and ethnicity. Second, we devised and conducted a test of allegations that

the for cause system was administered in a gender and race biased way. If, as had been

asserted, women and minorities were sent for drug testing on a lesser evidentiary

standard than was true for white men, one would expect that the percentage of false

positives B for cause tests with negative results B would be higher among women and

minorities than among white men. Our analysis, however, found the reverse to be true:

the percentage of false positives was higher in tests of white men than in tests of any

other group, and lowest in tests of black and hispanic officers. The tests appear hard to

subvert because subjects are given no notice of when they will occur. Instead, at ransom

and unannounced times when they arrive at work to begin their tours of duty, employees

are directed to report immediately to the drug test site (in Queens) to be tested in a

closely supervised setting. Refusal to do so results in immediate suspension and the

beginning of dismissal proceedings; positive findings carry the same consequence.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 95

videotape Winter and another receiving payoffs from the dealer. IAD then turned Winter and

his partner, equipping them with wire recorders as they collected evidence against 36 other

officers from the 77. Thirteen officers were indicted, one of whom committed suicide rather

than surrender to authorities, and 23 were subjected to internal disciplinary procedures

(McAlary, 1989).

The Buddy Boys scandal differed substantially from most prior corruption scandals in

New York City. The Harry Gross scandals of the 1950s had involved widespread corruption

among Brooklyn gambling enforcement officers who accepted bribes in order to allow Gross

to operate (Lardner and Reppetto, 2000). Although the Knapp/Serpico scandals eventually

exposed widespread, but relatively low-level, corruption among uniformed officers, it had its

roots in long-suppressed revelations that gambling enforcement officers received regular

monthly payments from the illegal gambling operators they allowed to flourish. The 1970s

Special Investigations Unit scandals exposed theft, robbery, and drug trafficking by members

of the NYPD=s most elite drug enforcement unit (Daley, 1978; Lardner and Reppetto, 2000).

The Buddy Boys were a dramatic step from these prior experiences. While many police

distinguished dirty money from the clean money offered by gamblers who sought only to be

left alone to ply their trades, only the SIU scandal had previously involved any organized drug

corruption or other dirty money taken B by bribery, extortion, theft, or robbery B from people

who were, in effect, sticking needles into the arms of the city=s young people. But even the

SIU scandal was not a clear precedent for the Buddy Boys because it involved an elite, highly

specialized, citywide unit, rather than a group of uniformed patrol officers assigned to a single

geographic area; the Buddy Boys were local cops on the beat rather than investigative stars.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 96

THE 1990s

The Buddy Boys may have been the first in a series of episodes that have since become

unfortunately common-place in urban U.S. policing: a small band of officers assigned to some

locally-based special duty (in their case, steady night duty) who, usually led by some

charismatic deviant, systematically brutalize, rob, and traffic in drugs. In the 1990s, a similar

pattern emerged, and was investigated by both the NYPD (Kelly, 1992) and the independent

Mollen Commission (1994). Arrests were made by police in Long Island=s Suffolk County

showed that a group of Brooklyn officers, apparently led by Officer Michael Dowd, another

charismatic suburbanite, engaged in brutality, robbery, theft, drug trafficking and abuse, and

the use of their police authority to drive off rivals of the dealers who employed them (Kelly,

1992; McAlary, 1994; Mollen, 1993). Nearly simultaneously, a group of officers in Harlem=s

Precinct B Nannery=s Raiders B were discovered in the same sorts of activities, this time

under the leadership of their sergeant. An additional group, the Morgue Boys, was found in

Brooklyn=s 75th precinct, as were several officers from the 46

Precinct in the Bronx (Mollen

Commission, 1994).

Similar things were happening in other cities. In Philadelphia, the 39

District scandal

involved that patrol district=s Five-Squad, a small group of officers who reported directly to

the district=s commander and who acted as his task force for local problems. As was true in

the case of Dowd and company, the wrongdoing of these officers became public when their

activities spilled across the city line into the jurisdiction of a neighboring police department.

Led by Officer Jack Baird, another substance abusing, fast-talking platinum blond from an

outlying area of his city, this group of eight or more officers robbed, stole, dealt drugs, and

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 97

brought tainted prosecutions against at least 160 citizens (Fazlollah,1996, 1997; Kramer,

1997; McDougall, 1997). After Los Angeles Officer Rafael Perez, a Philadelphia native, was

caught stealing cocaine, he exposed a similar scandal among his colleagues in the Rampart

Division=s CRASH Unit (Community Resources against Street Hoodlums), the LAPD=s

analogue to Philadelphia=s Five-Squad (Los Angeles Police Department, 2000). At this

writing, a year-long criminal trial against three Oakland officers known as The Riders has

ended in a hung jury, and is about to begin again. Led by Officer Francisco Vazquez, a, fourth,

veteran officer who is now a fugitive, The Riders were patrol officers who worked late nights

in Oakland=s western section, and who allegedly beat and framed citizens on phony criminal

charges. Exposed by a rookie officer assigned to them for training, The Riders were sued by

102 plaintiffs in 13 separate actions, recently settled by the City of Oakland for $10.9 million,

and face a myriad of criminal charges, none of which involve any apparent money-corruption

(Anderson 2004; Chapman, 2002; Counts and Chapman, 2002; Payton, 2003).

1990s Changes in the NYPD

The revelations of the early 1990s produced some major changes in the NYPD=s

provisions for holding officers and commanders accountable for proper behavior. The former

division of responsibilities between the central Internal Affairs Division and local Field

Internal Affairs Units had left many cases, including the Dowd matter, to slip between the

cracks, and was eliminated. Instead, a new and unified Internal Affairs Bureau was formed,

and engaged in more aggressive and proactive investigations than had been true in the past.

The field associate program, which had been created in the 1970s and used officers and other

employees to systematically and anonymously report to Internal Affairs on misconduct in their

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 98

commands, had long lain fallow, and was reinvigorated. A hotline available for employees to

make anonymous reports of misconduct was adopted, and the department began to scrutinize

systematically officers= records in search of cues that they might be bound for trouble. The

department examined the records of officers implicated in Mollen-era misconduct, and found

that most fell into three categories when hired: they were less than 22 years old; they had little

or no education beyond the minimum high school equivalency diploma; and they had not

served in the military. As a result, the entrance standards were changed to require a minimum

age of 22 and a minimum of two years of college or two years of military service.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 99

VI. RESEARCH METHODS

Our review of the police literature and of the experiences of New York City and its

police led us to generate several research questions for investigation. We refined these

questions and methods of addressing them in discussions with three focus groups selected

from within the department, and in meetings with departmental officials and two separate

advisory committees. The data we collected for this study are purely archival, and are drawn

from the manual and electronic files of the NYPD. They include extensive information about

the pre-employment and career histories of all officers involuntarily separated from the NYPD

(Astudy officers@) for reasons of misconduct (excluding failures in the recruit academy or on

other non-behavioral probationary screening steps), and on a same-sized randomly selected

sample of the involuntarily separated officers= Police Academy classmates (Acomparisons@).

Differences between study officers and comparisons in the distribution of variables under

examination served as tests of the hypotheses we had generated.

We chose this study/control method for the simplest of reasons. The basis of social

science is comparison, and the most ready way to determine whether involuntary separated

officers had unique characteristics or experiences was to compare their characteristics and

experiences to those of officers who were not involuntarily separated.

IDENTIFYING THE INVOLUNTARILY SEPARATED

As indicated earlier, involuntary separation from the NYPD may take one of several

forms. Dismissals are the firings of tenured employees who have completed their

probationary periods, and who, except when convicted of certain crimes, enjoy great due

process benefits. Conviction of any felony or any of several misdemeanors involving Amoral

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 100

turpitude@ (generally, larcenies or such vice crimes as sex or drug offenses) is grounds for

dismissal and, in such cases, all the Police Commissioner must do to rid the agency of such

employees is to provide them with written notice that they have been fired on grounds of

discretion..

In other cases, the department must file formal charge and specifications against

officers. These are tried before an administrative judge (the Deputy Police Commissioner for

Trials), with prosecution brought by the Department Advocate or a designated lawyer from the

Advocate=s staff, in the Trial Room, a courtroom much like the setting of a military court

martial. These proceedings include the right of cross-examination and the right to present a

defense case with counsel, and are recorded. The high visibility of this process, and the

subsequent publication of all dismissals in Personnel Orders meant that identifying the

dismissed officers in our study was a relatively straightforward proposition.

Personnel Orders also publish notices of the terminations of probationary employees.

Because New York=s civil service laws grant agency heads great discretion where

probationers

It is because such a conviction results authorizes automatic dismissal that the

NYPD typically holds its administrative actions in abeyance pending the result of any

criminal charges against officers. Thus, an officer who is charged with, say, felonious

assault for an alleged on-duty act of brutality typically will be suspended without pay for

30 days (the maximum period allowable) before assignment to some non-street duty. If

convicted of the criminal charges involved, such an officer would then be likely to simply

be fired without a departmental proceeding.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 101

wish include simple training failures in our study, we were required to locate and review the

files on all terminated probationers in order to determine whether to include them as

separations for misconduct or to omit them from the study.

As indicated earlier, a more substantial problem involved attempts to identify officers

who had resigned or retired rather than face trial for departmental charges likely to result in

dismissal and/or as part of negotiations (usually in return for cooperation against other

officers) that would save them the stigma of dismissal. We questioned internal affairs and

Trial Room personnel for their recollections of any such cases, and reviewed carefully the

folders of all officers whose resignation notices gave any indication that they had been in

trouble when they left the department. One indicator was whether officers resigned with the

permission of the police commissioner, the NYPD=s term of art for departures in good

standing. We pulled and reviewed any file that indicated that officers had resigned without

the commissioner=s permission, and wound up including most such cases in our files. Some

few involved officers who simply stopped coming to work and who, when asked for an

explanation, indicated that they no longer wished to be in the department=s employ. Another

indicator was whether officers were the only resignations included on particular Personnel

Orders. Resignations of young officers, especially, seemed to come in bunches, as they left

the NYPD to join recruit classes in the New York City Fire Department or in other police or

government agencies. In the end, we almost certainly missed some small number of forced

resignations and retirements, but do not believe that these are sufficient in number to have had

any meaningful effects on our results.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 102

ADVISORY COMMITTEES

We assembled two advisory committees. The first included four individuals who had

enormous combined experience performing and studying police work. They are:

- Walter Connery, an attorney. Mr. Connery was a career New York City police

officer, who held a variety of patrol, investigative, command, and training

positions. He left the department as a deputy inspector, and had served as the

Department Advocate, the NYPD=s prosecutor in internal disciplinary actions.

In this position, he was responsible for many of the proceedings against

officers whose misconduct had been exposed in the Knapp and French

Connection scandals. Subsequently, he headed the Office of Professional

Responsibility in the Immigration and Naturalization Service, as well as the

INS=s New York criminal investigation unit.

- Henry DeGeneste, Vice President for International Security of Prudential-

Bache. Mr. DeGeneste worked his way through the ranks to become chief of

the Port of New York and New Jersey Authority Police, and served on the

Commission on Accreditation for Law Enforcement Agencies. He has

published extensively on policing transportation facilities and services, and has

been a faculty member at John Jay College of Criminal Justice.

- Michael Julian, then the Vice President and General Manager of Madison

Square Garden. Mr. Julian, an attorney, worked his way through the ranks of

the New York City Police Department to become its three-star Chief of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 103

Personnel, and has also served as director of security and human resources

director of Rockefeller Center.

- Jerome Skolnick, a professor at New York University School of Law and an

internationally known police scholar. Dr. Skolnick has served on the faculties

of the University of Chicago, Yale University, the University of California at

San Diego, and as Director of the Center for the Study of Jurisprudence at the

University of California, Berkeley, School of Law Boalt Hall. He has

published extensively and is a former president of the American Society of

Criminology.

We met with this committee to describe our project and to solicit their suggestions and

criticisms of our draft research questions, instruments, and analytic plan. They provided us

with excellent guidance and strengthened this report significantly.

A second advisory committee included scholars:

- Michael Maxfield, a professor at Rutgers University=s School of Criminal

Justice. Dr. Maxfield is an internationally known research methodologist and

scholar of the police, and has also served on the faculties of Indiana University

and the State University of New York at Albany.

Attorney Richard Koehler, another former NYPD Chief of Personnel and ex-

New York City Corrections Commissioner, also served on this committee briefly, but had

to leave the advisory committee because of a conflict of interest.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 104

- Joan McCord, a professor of criminal justice research at Temple University

who was internationally known for her longitudinal studies of crime and

violence, who had also served as president of the American Society of

Criminology. Professor McCord died in early 2004

This committee helped us to design our data collection instruments and to refine our

research design. They also served as valued consultants whenever we needed advice on

methodological issues.

FOCUS GROUPS

The NYPD helped us to assemble three focus groups. One consisted of eight patrol

officers, one from each of the NYPD=s Patrol Borough commands.

The second and third

included, respectively, eight patrol sergeants and eight captains who served as precinct

commanders, again one from each patrol borough. We advised these three groups of the

nature of our project, and sought their input on integrity problems and other causes of career

failure with the department.

The perceptions of the focus groups suggested that the NYPD had changed

significantly since the 1960s and 1970s, when Fyfe worked on patrol. In that earlier era,

corrupt officers had little to fear from their colleagues. Although relatively few officers then

Jeffrey Fagan, a professor at Columbia University=s Mailman School of Public

Health, also served briefly on this committee, but had to leave it because of a conflict of

interest.

The Patrol Boroughs are Manhattan South, Manhattan North, Bronx, Brooklyn

South, Brooklyn North, Queens South, Queens North, and Staten Island.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 105

were actively involved in significant corruption,

the prevailing organizational norm forbade

honest officers from giving up another cop by calling to the attention of higher authority the

improper activities of corrupt officers. Indeed, this norm was so strong at the time that it

extended to the department=s top levels, as witness the several years it took Frank Serpico and

David Durk to get any high official to pay heed to their warnings of corruption in the ranks

(Lardner, 1996; Maas, 1973). To a person, however, our focus groups insisted that corrupt

officers would quickly be given up by one or more of their colleagues, usually via anonymous

communications with the Internal Affairs Bureau.

The focus groups were less sanguine where abusive police conduct was concerned.

The patrol officers indicated that the prevailing view in all their precincts was that what

officers did to persons they had arrested was their own business, and that it was likely that

officers who witnessed, for example, backroom police brutality, would neither intervene to

stop it nor report it afterwards. Both the precinct commanders and the sergeants confirmed

this norm, indicating that it was a major problem for supervisors, commanders, and the

department=s administration. As it turned out, our focus groups were prescient: our officers

met on a Tuesday; our sergeants met on Wednesday; our captains met on Thursday. On

Friday of that same week, one or more Brooklyn officers used a wooden stick to sodomize

Abner Louima in their precinct=s restroom. These officers were caught only when Internal

Affairs received tips from non-police personnel who had learned of the nature of Mr.

Louima=s injuries, and it is clear from the conflicting police accounts that some officers=

stories are

Low-level corruption (free meals, discounted services, and gratuities at

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 106

Christmas from local merchants) was far more widespread.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 107

inaccurate. The Louima incident produced several major changes in policy and procedure. In

addition, because it involved brutality far in excess of anything anticipated in our focus group

discussions (which revolved around more traditional forms of abuse, such as the punch in the

nose or ribs), it has illustrated the danger of tolerating minor excesses. Our subsequent

conversations with department personnel suggest that this awareness has permeated informal

norms, as well. At the line level, officers tell us that the Louima incident put an end to, or

greatly weakened, the solidarity and silence that previously had characterized police officers=

culture. This is an assertion that might be taken with a grain of salt if it came only from the

line officers, but it is reinforced by IAB=s belief that more than half of the calls that come into

their anonymous reporting system originate come from the subject officers= colleagues.

This belief is based on IAB=s determination that more than half of these calls

report information that could be known only to officers= colleagues.

There was also consensus among the groups, including the officers, that local delegates

of the Patrolmen=s Benevolent Association often were elected by officers from among the

most outspoken and militant of their colleagues. The groups agreed also that the efforts of

these delegates had greatly diminished the department=s ability to effectively discipline

officers. Finally, the groups all argued that the NYPD=s entrance standards have varied

wildly over the years, and that some entering cohorts had been inadequately screened and

trained. Consequently, all three groups suggested, many officers from these cohorts have

become chronic disciplinary problems.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 108

ASSEMBLING THE DATA

The NYPD does not maintain a single file that could serve as the central source of all

the data we needed for our study. Throughout his service as a member of the department, Fyfe

had heard rumors that the NYPD maintained a large police blotter-style book in which each

new officer was assigned a page as he or she entered the department. Then, it had been

claimed by senior officers, every significant event in an officer=s career was recorded in this

book. If this were true, of course, it would have greatly eased our data collection task.

The Book

We attempted to verify rumors of this APolice Commissioner=s Ledger@ with the

NYPD=s key headquarters and personnel staff. Several senior members indicated that, like

Fyfe, they had heard that such a history existed, but that they had never seen it or any

corroboration of its existence. Thus, after tracking this rumor as far as reasonable, we wrote it

off as yet another apocryphal locker room tale, and devoted our energies to collecting data

from sources that we knew did exist.

As it turned out, we gave up too soon. Late in out data collection efforts, we filled in

some gaps in our data by searching through the NYPD=s old records storage facility, the

musty former headquarters of a wine importer, located on an out of the way dock in Brooklyn.

Rummaging through boxes of documents, we came upon a wall of the ledgers which, like

traditional police blotters, were written in usually elegant hand on heavy lined paper in

beautifully bound volumes about two feet high. Beginning with officers who entered the

department in June 1940 (when AArticle 2," the New York law that has since governed police

pension arrangements, was enacted), the book (by now, many volumes) devotes a page, in

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 109

employee number order, to every NYPD officer=s career; the data for officers whose careers

could not be described on a single page are continued onto pages indicated later in the back

pages of each volume.

Entries in the book scan officers= entire careers, and obviously have been made

contemporaneous with the events described. Thus, notations on each page might have been

made over periods of more than 40 years. The book includes information on new officers=

residence, physical descriptions, and backgrounds (in some cases, including photographs);

assignments; transfers; injuries; leaves; detailed information on departmental recognition,

complaints, and disciplinary actions; changes in residence and marital status; and performance

on promotional examinations. By pulling the appropriate volume from the shelves, Fyfe was

able to quickly examine his own career history, as well as that of his entry-level cohort: as far

as could be determined, nothing was inaccurate or missing. The book appears to have been

maintained until the late 1980s or early 1990s and includes data on about 60,000 officers. It is

doubtful that there exists anywhere in the country such a trove of information about police

officers and their careers. Unfortunately, it was found too late to be of much use in this

research and, in any event, was not in use for the last several years encompassed by the

careers of the officers we studied.

ADepartmental recognition@ is the NYPD=s term for citations for outstanding

acts or service. The NYPD has an elaborate recognition system that ranges through

several grades of award that are displayed by small enameled breastbars of varying colors

and designs worn above officers= badges, or Ashields.@

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 110

Personnel Records Unit

We did make extensive use of the files of the NYPD=s Personnel Records Unit. PRU

maintains a file for each active member of the department and, during the period of our data

collection, on all officers who left the department after 1994.

The most substantial part of

this is the PA-15, a lengthy application and information form completed by candidates for the

police job and used as the basis for pre-employment investigations. The PA-15 has changed

slightly from time to time over the years encompassed by this study, but we used it to develop

the front end of the scannable data collection instruments for study and comparison officers

included at the end of this report as as Appendices A and B. PRU=s files also contain

officers= annual job performance evaluations; reports of disciplinary actions against officers;

information concerning changes in the marital status, educational achievement, assignments,

sick leave, injuries, and various other noteworthy events (e.g., commendatory letters;

departmental recognition; serious vehicle accidents; reports of line of duty injuries). Also

included for periods covered by the study are records of the officers= performance during

recruit training.

The NYPD=s Old Records Unit in Greenpoint, Brooklyn, maintained PRU

records of officers separated from the NYPD prior to 1990; the New York City Archives,

a short walk from police headquarters, maintained records on officers separated during

1990-94. All were readily accessible to our staff.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 111

Personnel Orders Section

In addition to data from the PRU, we retrieved data from the NYPD=s Personnel

Orders Section. These included records of all transfers, assignments and, where relevant, the

manner in which officers had left the department (e.g., resigned with permission; dismissed,

etc.).

Central Personnel Index

We also checked each study and comparison officer in the NYPD=s Central Personnel

Index. CPI is a computerized database that records and classifies all Acritical events@ in

officers= careers, including complaints, Adepartmental recognition,@ designations as

Achronically sick,@ and Aunfit for duty.@

Other Units

We also received some data from the NYPD=s Management Information Systems

Division. MISD generally was a failsafe for instances in which data were not readily available

from our primary sources. Similarly, the Internal Affairs Bureau and the Department

Advocate, the prosecutor in serious internal disciplinary actions, were valuable secondary

sources of data.

NYPD officers enjoy unlimited sick and injury leave. AChronic sick@ is a

designation applied to suspected malingerers after repeated short periods of sick leave.

AUnfit for Duty@ is a formal determination made by the NYPD=s physicians that, by

reason of intoxication, an officer is not capable of performing his or ger duties. This is a

rare determination but, since officers are required by the department=s rules to be fit for

duty at all times, it can occur whether an officer is on-duty or off-duty. Usually, it

occurs when departmental surgeons are summoned by officials to police precinct houses

to assist in investigation of such events as officers= domestic disputes.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 112

One data element we would have liked to have included for analysis was the results of

psychological testing and interviews conducted by the NYPD=s Psychological Services

Unit.

We requested this information, but received our only denial of access, on grounds that

it raised unresolved issues of doctor-client confidentiality. Our suspicion, however, is that

these data would not have added much to our findings. The department apparently places

enough credence in its psychological screening procedures so that those who pass through

them and eventually enter the department include only a relatively small range of scores.

Data Collection

With the assistance of staff of the Personnel Bureau, the Employee Management

Division, and the Deputy Commissioner for Legal Matters, we assembled a data collection

staff of more than 30 NYPD personnel. They were taken on to the project as part-time

employees, and were involved in the development and refinement of the data collection

instruments. A handbook was prepared to assist them in their task and, with a short training

course, they set to working reviewing the original data sources and coding the data. Jones,

Kane, and Tillman supervised the coding process, and conducted reliability cross-checks on

randomly selected instruments with satisfactory results. After coding was complete, Jones

conducted additional reliability tests to assure data accuracy. To resolve ambiguities in the

In addition to interviews, the NYPD administers the Minnesota Multiphasic

Personality Inventory to all candidates and, at various times, has also employed other

psychological instruments on its screening procedures.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 113

data, we occasionally consulted with sources in the Internal Affairs Bureau and the

Department Advocate=s Office.

ANALYZING THE DATA

The data were analyzed using the Windows version of the Statistical Package for the

Social Sciences. Most of our analyses consisted of testing hypotheses by comparing variation

between the study and comparison groups on variables under examination. Although our

study group is a population rather than a sample (the comparison group is a sample), we

determined that, where appropriate, we would reject null hypotheses when the probability

values of the chi-square scores we derived were less than or equal to .05. In other instances,

where we present simple univariate analyses (e.g., of the nature of misconduct), simple visual

examination either supports or causes us to reject our hypotheses.

We elected to engage in the lengthy series of bivariate analyses the reader will

encounter in Chapter 8 because, absent data like those we captured for this study, so many

bivariate relationships involving police personnel have for so long been the subjects of

speculation and debate. Do the nature and frequency of police misconduct depend on who the

mayor is? Does diversity in police ranks matter? Does it make sense to require police officers

to reside where they work? Is it true that military veterans make better cops? That people

recruite into police ranks at older ages are less likely than their younger academy classmates

to get into career-ending trouble? What differences in the rates and nature of misconduct are

there among men and women, whites, blacks, Hispanics, and Asians?

We thought it necessary in this study to address these questions by conducting and

presenting bivariate analyses. We did this even though, correctly, we felt certain that some of

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 114

the significant differences we might find would wash out in the multivariate models we

intended to construct, and that also are presented in this study. In the reviews of the first draft

of this study, one reviewer found these bivariate analyses painful, unfriendly, and unnecessary

because some of our bivariate relationships did, indeed, fall out of our multivariate models.

We then generated a hypothesis concerning this reviewer: he or she is a methodological

specialist who has not administered a police agency.

Our advice to such readers is to scan

our Executive Summary, and then to skip directly to the sections of this report that present

multivariate analyses. For our other audience B police and public administrators not grounded

in Principal Component Analysis or Logistic Regression Analysis, the multivariate methods

we employed B our advice is to stay away from the multivariate sections of this report, and to

focus attention on our analyses of the questions you probably have heard and discussed about

throughout your careers.

The absence of data left us unable to test this hypothesis.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 115

VII. RESEARCH QUESTIONS

The research questions we generated fall into the following general topical headings:

- Regional and Organizational Correlates of Police Misconduct

- Community Demographics and Police Behavior

- Local Political Culture and Police Behavior

- Police Socialization, Training, and Police Behavior

- Police Demographics and Representativeness

- Individual Correlates of Police Misconduct

$ Race and Police Behavior

$ Age and Police Behavior

$ Prior Employment, Life History, and Police Behavior.

In addition, we conducted multivariate analyses designed to identify factors that might

best distinguish between the officers in our Study and Control groups.

REGIONAL AND ORGANIZATIONAL CORRELATES OF POLICE

MISCONDUCT

As indicated earlier, the literature suggests that eastern cities such as New York are

marked by profit-motivated police misconduct rather than by the brutality and other non-profit

abuses that often have scandalized departments in other parts of the country .at the trouble

most frequently affecting police involves profit-motivated corruption ((Bobb, et al., 1992,

1993, 1994a, 1994b, 1995, 1996a, 1996b; Chevigny, 1996; Chicago Police Committee, 1931;

Christopher Commission, 1991; Cohen, 1980; Daley, 1978; Domanick, 1994; Fogelson, 1977;

Gates, 1992:85-87; Goldstein, 1977:214; Kappeler, Sluder, and Alpert, 1994:145-167, 187-

238; Kelly, 1992; Knapp Commission, 1972; Kolts Commission, 1992; Maas, 1972; McAlary,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 116

1989; Mollen Commission, 1993, 1994; Murphy and Plate, 1977; Philadelphia Police Study

Task Force, 1987; Rothmiller and Goldman, 1992; Rubinstein, 1973; Schecter and Phillips,

1973; Sherman, 1974; Skolnick and Fyfe, 1993; Williams, 1974). We hypothesized, therefore,

that:

H1: More involuntary separations are attributable to profit-motivated corruption

than to brutality and other non-profit abuses of citizens.

COMMUNITY DEMOGRAPHICS AND POLICE BEHAVIOR

Community and organizational factors that may affect officers' behavior vary within,

as well as across, large and dynamic police jurisdictions and agencies. Among those of

greatest relevance to this research are community demographic and socioeconomic

characteristics, including rates of crime and violence; local political culture; police

socialization and training; and the demographics and representativeness of police departments.

Corrupt police officers cannot accept bribes from gambling operators or rob drug

dealers if such industries do not exist where they work, or if officers are not authorized by

their agencies to enforce such laws.

Brutal police officers are most prone to victimize the

disenfranchised (Reiss, 1971; Sherman, 1990; Skolnick, 1966; Skolnick and Fyfe, 1993), and

We recognize that there is considerable difficulty in demonstrating that an

officer has engaged in brutality. When reported by a resourceful officer, Bittner

(1970:56) and Toch (1995:106-107) point out, an unjustified beating or shooting can

easily be redefined as necessary and reasonable use of force. But such dissembling also

occurs among corrupt officers, who rarely report accurately on their profit-making forays.

Indeed, in an effort to reduce corruption, the NYPD actually took the step of

prohibiting officers from making on-site arrests of drug and gambling suspects (Fyfe,

1978: 329-341; New York State Commission of Investigation, 1972:60-62).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 117

are unlikely to risk their careers by attacking those with the resources to strike back in judicial

or administrative proceedings. In addition, Fyfe (1980a) has reported that rates of public

violence and arrests for violent crime have a near straight-line association with one measure of

police force, shootings by officers, across New York City police divisions, the subjurisdictions

that, at that time, each included three or four patrol precincts.

It is unlikely that the variation in demographics, crime, and police exposure to risk and

temptation across the NYPD=s 76 patrol precincts is equaled in any U.S. police jurisdiction.

This gave us an excellent opportunity to test the following:

H2: Rates at which officers are involuntarily separated for corruption and

brutality, as well as for other job-related misconduct, bear the associations indicated

with the following characteristics of the communities in which they are assigned:

H2a: rates of public homicide (positive);

H2b: FBI Index Crime rates (positive);

H2c: percentage minority population (positive);

H2d: per capita income (negative).

LOCAL POLITICAL CULTURE AND POLICE BEHAVIOR

Mayoral priorities and political exigencies shape both the broad style of policing and

the day-to-day operations, legitimate and otherwise, of line officers (see, e.g., Fyfe, et al.,

1997: 53-82; Wilson, 1968). Local elective politics also help to draw the distinctions between

acceptable police conduct and that which is officially regarded as deviant. In 1970s

Philadelphia, for example, police abuse apparently was rampant, but was neither reported nor

officially condemned by local officials. Instead, officers were free to follow their mayor's

advice to "break their heads before they break yours," and were assured by their commissioner

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 118

that he would support them as long as they believed that they had done right. When a reform

political administration cracked down on such abuse, reported brutality actually increased

(although, e.g., fatal police shootings declined 67 percent in the new administration's first

year; see Fyfe, 1988a), probably because the public grew aware that the administration was

serious about investigating and eliminating it (see Skolnick and Fyfe, 1993:138-142; U.S.

Civil Rights Commission, 1979, 1981).

Local politics also influence police priorities and operations in less extreme ways.

Mayors whose election campaigns included promises to crack down on crime (like current

New York Mayor Rudolph Giuliani and, to a lesser degree, Edward Koch, mayor 1978-1989)

or to eliminate police abuse (like David Dinkins, mayor 1990-1993) invariably choose police

leaders who share and implement their priorities (like Giuliani's appointees William Bratton

and Howard Safir, and Dinkins' appointees Lee Brown and Raymond Kelly, respectively). In

the former case, it is sometimes asserted, they are likely to encourage and tolerate aggressive

crime fighting techniques that might be regarded as abusive by mayors who had promised

more sensitive and community-oriented policing (see, e.g., Geller and Scott, 1992:300-308).

In the latter case, their chiefs are likely to implement organizational changes and restrictive

policies that limit opportunities for brutality and corruption.

H3: The rate of involuntarily separations for brutality and other non-profit abuse of

citizens was greater during the administration of David Dinkins than during either the

Koch or Giuliani administrations.

Additional environmental and organizational changes that can be expected to have

recently affected involuntary separation rates for misconduct other than brutality include the

post-Mollen Commission expansion of the NYPD=s integrity control mechanisms, and the

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 119

growth of its drug testing programs. The former Internal Affairs Division was reconstituted

into a larger Internal Affairs Bureau; much duplication B and opportunity for buck-passing B

was eliminated; and proactive operations such as stings have been increased significantly

since the NYPD=s last major corruption scandal. In addition, the NYPD=s drug testing

programs were created expanded dramatically in the latter part of in our study period, and

have been accompanied by a zero-tolerance policy that leads to automatic dismissal for failing

or refusing to undergoing to take a drug examination. Thus:

H4: Rates of involuntary separation for reasons other than brutality and non-

profit abuse were greater during the administration of Rudolph Giuliani than

during either the Koch or Dinkins administrations.

POLICE SOCIALIZATION, TRAINING, AND POLICE BEHAVIOR

Westley's pioneering 1950s work (not published until 1970) indicated that police

training and officers' introduction to their work contributed in the agency he studied to a

culture that virtually required officers to use force to demonstrate their solidarity with their

peers. Rubinstein wrote that, after brief and unrealistic training and moralizing at the training

academy, new Philadelphia officers were assigned to police districts, largely without

guidance, and found that "everywhere is bribery" (1973:404). In New York, it has long been

an article of faith that the "Class of '69," which had been hastily recruited and trained in

anticipation of continuing "long, hot summers" of riots went on to careers of special

ignominity. This concern led to the post-Knapp Commission expansion of recruit training

from 560 to 1,040 classroom hours, and to the proposal for "escort," or field, training for

NYPD rookies that was subsequently added to the expanded classroom training (see,

McManus, 1969:168-177; New York State Board of Regents, 1974). In 1979, in order to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 120

quickly reinforce badly understaffed field forces, training was decreased to 800 hours (New

York City Police Academy, 1979). Then, in 1995, the curriculum was again expanded to

1,040 classroom hours by inclusion of an extensive integrity and ethics component (New York

City Police Academy, 1995).

There is some evidence that variations in entry level class size, often caused by the

irregularity of municipal budgeting, are associated with subsequent misconduct by trainees

(Bolz, Selvaggi, and Fyfe, 1993; Fyfe, et al., 1997: 290-291). When, as in New York, police

departments hire small numbers of officers B or none at all B for several years, their resources

and ability properly to socialize new officers erode, as idle personnel and training staff are

transferred to the field. Then, when they must suddenly gear up for massive hiring and must

screen and train rapidly, such departments are caught short, and must make do with inadequate

and inexperienced staff. In such circumstances, training is weak in both substance and as a

symbol to new officers of the agency's commitment to professionalism and, in all likelihood,

in the rigor with which it is used as a device to screen out unpromising officers. When the new

officers in such large cohorts are graduated to the field, they are too great in number to be

absorbed gently into the working culture by mellower old-timers (McManus, et al., 1969).

Instead, they may redefine the police culture, often in terms more suitable to bad television

shows than to policing a democracy. Three recent examples of this phenomenon present

themselves: the Miami and New York City Police Department of the early 1980s and the

Washington DC Metropolitan in 1989-90 all hired quickly and indiscriminately, and went on

to see many of their new personnel arrested and dismissed over the next several years (see,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 121

respectively, Miami, 1984; Mollen Commission, 1994:113-115; Kappeler, Sluder and Alpert,

1994:213-238).

Thus:

For example: the Mollen Commission (1994:113-115) reported that 88 percent

of the suspended and dismissed officers it studied had not been investigated thoroughly at

appointment.

H5: Rates of involuntary separation for corruption, brutality, and other forms of

misconduct are positively associated with the size of Police Academy recruit training

cohorts.

H5a: The rate at which new officers are involuntarily separated as

unsatisfactory probationers is inversely correlated with the size of Police

Academy recruit training cohorts.

H6: Rates of involuntary separation for corruption, brutality, and other forms of

misconduct are inversely associated with the rigor and intensity of recruit training, as

measured by:

H6a: numbers of hours of training;

H6b: rates of involuntary separation for academic and physical failure, and

for disciplinary reasons.

POLICE DEMOGRAPHICS AND REPRESENTATIVENESS

There is little empirical evidence to support the common belief that a police

department that is representative of its clientele by gender, race, and ethnicity will be more

humane, honest, and better behaved than one that is not (Alex, 1969; Kerner Commission,

1968). Requirements limiting eligibility for police positions to persons residing in employing

jurisdictions frequently are upheld as means of enhancing both representativeness and officers'

identification with their constituents and their problems (Detroit Police Officers Association v.

City of Detroit,

1971; "Where New York's Finest Should Live," 1991:A18). This reasoning is

plausible, but cities that have such requirements in place (e.g., Chicago, Detroit, Philadelphia,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 122

Washington DC) have had no dearth of police scandals. Further, even though city residents

include relatively great minority representation, the Mollen Commission (1994:128) pointed

out that 77 percent of the suspended and dismissed officers in its study were New York City

residents at the time of their appointments.

It also is plausible that minority citizens may be more accepting of minority officers

(Alex, 1969; Baldwin, 1963; Fyfe, 1988b) and that representativeness may break down the us

v. them posture often described in studies of white male dominated police agencies serving

communities of color (Kelly, 1976; Westley, 1970). In the aggregate, this may reduce the

need for legitimate force and the temptation to brutality and financial exploitation generally.

We hypothesized:

H7: Rates of involuntary separation for corruption, brutality, and other police

misconduct are inversely associated over time with the percentage of non-white

officers in the department.

H8: Rates of involuntary separation for corruption, brutality, and other police

misconduct are positively associated over time with the percentage of officers who

reside outside New York City.

INDIVIDUAL CORRELATES OF POLICE MISCONDUCT

Perhaps because police were so long chosen from a pool of people whose

characteristics and backgrounds did not vary much (e.g., McManus, et al., 1969:74-96),

information about the relationship of individual characteristics and deviance is sparse.

Frequently mentioned in the literature of police behavior generally, however, are such

individual variables as gender, race, age, education, intelligence, physical fitness, and length

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 123

of service.

Thus B especially as police agencies have become more diverse over the last 25

years B we thought it fruitful to examine these variables. Taking care to separate its effects

on the department's disposition of charges that end careers, it is also useful to analyze records

of officers' police job performance prior to dismissal.

Gender and Police Behavior

An additional relationship that has been discussed in prior literature is that

between officers' attitudes toward their work and clientele and their job performance

(Friedrich, 1980; Niederhoffer, 1969; Reiss, 1971; Muir, 1977; Toch, 1995; Worden,

1989). Because the proposed study is retrospective, it is not possible for us to administer

the attitudinal meaurement instruments needed to examine this question. Even absent an

empirical test, two propositions seem reasonable: cynical officers are more likely than

idealistic officers to become involved in corrupt activities; and officers who have lost

faith in the justice system may be prone to engage in punitive brutality (see, e.g.

Klockars, 1980; Skolnick and Fyfe, 1993).

As indicated earlier, the literature on female police officers= behavior and careers is

limited. Some years ago, Martin (1980) found that new female officers were less than warmly

welcomed into the Washington DC Metropolitan Police Department. This might lead one to

anticipate that female officers= behavior, especially during probation, would be scrutinized

more closely than that of their male counterparts. Consequently, one would also anticipate

that they would more often be forced out of the police service during this early point in their

careers. Aside from news stories generated by occasional arrests of corrupt female officers,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 124

not much is known about officer gender and corruption. Most comparisons of male and

female officers have found that males are more aggressive enforcers (Bloch and Anderson,

1974; Forst, Lucianovic, and Cox, 1977; Melchionne, 1974; Sherman, 1975; Snortum and

Beyers, 1983; Worden, 1989), and are more likely to use force (Grennan, 1987; Horvath,

1987). This might cause one to suspect that female officers would become involved in

brutality less often than males and that, as the representation of female officers, supervisors,

and commanders in a department increased, its culture would become less tolerant of

excessive force. Conversely, as women became more integrated into a department, leaving

behind their "token" status (Martin, 1980), one might also anticipate that they would more

frequently become involved in profit-motivated corruption where it is approved by police

cultural norms. Further, the early results from our disciplinary study suggest hypotheses

involving involuntary separations for other reasons, such as drug abuse and rule breaking.

Thus, we hypothesized:

H9: Female officers' involuntary separation rate is positively associated over time

with the percentage of females in the department.

H10: The rate of involuntary separations of probationers is greater among female

officers than among males.

H11: The rate of involuntary separations for corruption is greater among male

officers than among female officers.

H12: The rate of involuntary separations for brutality and other non-profit abuses is

greater among male officers than among female officers.

H13: The rate of involuntary separations for drug test failures and refusals is

greater among female officers than among male officers.

H14: The rate of involuntary separations for non-line of duty criminal conduct ( e.g.,

off-duty thefts and fraud; drug crimes) is greater among male officers than among

female officers.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 125

H15: The rate of involuntary separations for administrative rule breaking is greater

among female officers than among male officers.

Race and Police Behavior.

Our review of the literature found that black officers were more likely than others to be

implicated in police misconduct and such other controversial police actions as use of deadly

force, and that such disparities may be associated with racial variations in rank and race

(Cohen and Chaiken, 1972, 1973; Fyfe, 1980c, 1981; Fyfe, et al., 1998). More recently,

however, Fyfe, et al. (1998) examined data which suggested that there is merit to the frequent

assertion that white officers are more often involved than others in brutality and abuse see,

e.g., Alex, 1969; Cohen, 1980; Domanick, 1994; Takagi, 1974, 1979). Thus:

H16: The rate of involuntary separations of probationers is greater among non-

white officers than among white officers.

H17: The rate of involuntary separations for corruption is greater among non-white

officers than among white officers.

H18: The discrepancy between white and non-white officers' rates of involuntary

separation for corruption is accounted for by differential patterns of assignment and

rank.

H19: The rate of involuntary separations for brutality and other non-profit abuses is

greater among white officers than among non-white officers.

H20: The rate of involuntary separations for non-line of duty criminal conduct (e.g.,

off-duty thefts and fraud; drug crimes) is greater among non-white officers than

among white officers.

H21: The rate of involuntary separations for drug test failures and refusals is

greater among non-white officers than among white officers.

H22: The rate of involuntary separations for administrative rule breaking is greater

among non-white officers than among white officers.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 126

H23: The discrepancy between white and non-white officers' rates of involuntary

separation for administrative rule-breaking is accounted for by differential patterns of

assignment and rank.

Age and Police Behavior

The literature on age and police behavior leads to mixed recommendations. Without

citing their sources, Wilson (1963) and Leonard and More (1993) suggest that young hires are

more likely than older recruits to become successful officers. More clearly empirical research,

however, suggests that officers who are young when recruited to the police service are more

likely than older officers to run into career trouble (Cohen and Chaiken, 1972:64-65;

Mladinich, 1995; O'Sullivan, 1994, sections 1.1, 3.7.4; see also Mollen Commission,

1994:112-115). This finding led the NYPD to increase its minimum age at entry to 22

(Mladinich, 1995), and was consistent with an earlier finding by that younger officers

appointed to the NYPD in 1957 were more likely than their peers to have been subjects of

citizens' complaints between then and 1968. To investigate this issue as it relates to career-

ending misconduct, we hypothesized::

H24: The rate of involuntary separations of probationers is greater among officers

who were less than 22 years old when appointed than among officers who were 22 or

more years old when appointed.

H25: The rate of involuntary separations for corruption is greater among officers

who were less than 22 years old when appointed is greater than among officers who

were 22 or more years old when appointed.

H26: The rate of involuntary separations for brutality and other non-profit abuses is

greater among officers who were less than 22 years old when appointed than among

officers who were 22 or more years old when appointed.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 127

H27: The rate of involuntary separations for non-line of duty criminal conduct (e.g.,

off-duty thefts and fraud; drug crimes) is greater among officers who were less than 22

years old when appointed than among officers who were 22 or more years old when

appointed.

H28: The rate of involuntary separations for drug test failures and refusals is

greater among officers who were less than 22 years old when appointed than among

officers who were 22 or more years old when appointed.

H29: The rate of involuntary separations for administrative rule breaking is greater

among officers who were less than 22 years old when appointed than among officers

who were 22 or more years old when appointed.

Prior Employment, Life History, and Police Behavior

As intuition might suggest, the research indicates that, where police are concerned, the

past is prologue. Cohen and Chaiken found that histories of unsatisfactory employment and

military discipline were associated with subsequent rule violating (1972:79-85). Honorable

records of military service, by contrast, long have been regarded as good preparation for the

discipline of police agencies

(Leonard and More, 1974:226-27, fn. 14). Cohen and

Chaiken=s analyses of police success and histories of psychological disorders and traffic

Like many other civil service organizations, the NYPD gives extra credit on

entrance and promotional examinations for military service.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 128

violating were inconclusive, perhaps because candidates who are on the extremes on these

dimensions are screened out before hiring.

We sought to investigate this issue as follows:

H30: Rates of involuntary separation for all types of police misconduct are higher

for the following categories of officers than for other officers:

H30a: officers who were dismissed by previous employers.

H30b: officers whose prior employers gave police investigators derogatory

information about them.

H31: Officers with military records that include no discipline are less likely than

officers without military experience or with military disciplinary histories to be

involuntarily separated.

A related line of scholarship (Cohen and Chaiken, 1972; Bowker, 1980; Reuss-Ianni,

1983; Kappeler, et al., 1992; Eterno, 1996; Truxillo, et al., 1998) holds that well-educated

police officers are more likely than their colleagues to enjoy successful police careers. We

hypothesized:

H32: Rates of involuntary separation for all types of police misconduct are inversely

associated with officers' years of education at entry into the NYPD.

H33: Officers= years of education at entry into the NYPD is positively associated

with movement into supervisory and management ranks.

H34: Officers= educational attainment after entry into the NYPD is positively

associated with movement into supervisory and management ranks.

Academy and Probationary Performance and Police Behavior

Cohen and Chaiken's (1972:103-110) work suggested that officers who did well in the

Police Academy and on probationary evaluations (also quantifiable) were more likely than

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 129

their colleagues to have advanced through the ranks and win awards, and were less likely to

have engaged in misconduct or to have been disciplined. Thus:

H35: Rates of involuntary separation for all types of police misconduct are inversely

associated with officers' Police Academy academic averages.

Police Job Performance

Findings to the effect that a small number of police officers accounts for a great

proportion of misconduct allegations, citizens' complaints, civil suits and judgments, and other

possible indicators of violence or corruption are by now common (Bennett-Sandler and Ubell,

1977; Christopher Commission, 1991:35-48; Kolts Commission, 1992: 157-168; Mollen

Commission, 1994:90-100; St. Clair Commission, 1992: 110-114; Toch, 1969, 1995; Toch,

Grant, and Galvin, 1975). The New York City Police Department's Michael Dowd was the

subject of 20 or more misconduct allegations in the seven years before he was arrested (Kelly,

1992:22-103); his Philadelphia counterpart, John Baird, was the subject of 26 or more

complaints before his arrest (Yatvin, 1996). These findings give rise to the following

hypotheses:

H36: Rates of involuntary separation for all types of misconduct are positively

associated with the following variables (all standardized by rates per year of service):

H36a: prior citizens' complaints:

H36b: prior supervisory disciplinary actions;

H36c: prior Abelow standards@ evaluations on performance evaluations;

H36d: prior line of duty civil suits;

H36e: prior line of duty injuries;

H36f: prior line of duty vehicle accidents;

H36g: prior designations as chronically sick.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

H37: Rates of involuntary separation for brutality and other abuses are positively

associated with rates of departmental commendations.

Fyfe and Kane -- 130

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 131

VIII. HYPOTHESIS TESTING

H1: More involuntary separations are attributable to profit-motivated corruption

than to brutality and other non-profit abuses of citizens.

In defining this hypothesis, we anticipated that we would be able to distinguish job-

related corruption from other profit-motivated crime. As explained earlier, however, this

turned out to have been a naive assumption. Consequently B although we regard this

revelation itself as a major finding of our work B we tested the hypothesis by employing our

eight-category misconduct typology. This produced Figures VIII-1 and VIII-2.

Figure VIII-1 presents the primary charges lodged against officers, defined as the most

serious or root charge that led to the officers= separations from the department.

Visual

examination indicates that the hypothesis is confirmed: during 1975-96, the NYPD separated

252 officers whose primary offense involved some profit-motivated crime, while only 58

officers were separated for offenses stemming from on-duty abuse.

An example: assume that an officer has been the subject of a drug trafficking

investigation during which it is discovered that he was part of a criminal conspiracy and

that he had also conducted his illicit business on departmental time. In such a case, he

would be charged with offenses classified under both the Aobstruction of justice@ and

Aadministrative@ categories, but his primary offense B the most serious and the one that

precipitated the NYPD=s interest in his activities B would be the Adrugs@ charges

against him. We found 58 officers who were separated primarily for on-duty abuse.

Sixty-one additional officers were dismissed in cases in which charges other than the

primary offense included on-duty abuse.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-1: Primary Charges against Study Officers

420

338

252

181

ofit-motivat

d c

off

duty cr

v p

rson

-du

er crime

on du

abus

truction of j

tice

minis

condu

t on

ion

Fyfe and Kane -- 132

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane -- 133

This same pattern appears in Figure VIII-2, which includes data on all specifications

lodged against separated officers, without regard to their seriousness.

The data again show

that profit-motivated crime specifications (n=387) greatly outnumber charges of on-duty abuse

(119). Thus, H1 is confirmed: whether using officers or the charges against them as the unit

of analysis, the number of separations involving profit-motivated offenses is three to four

times higher than the number that involved on-duty abuse.

Profit-Motivated Crime

In Figure VIII-3, we present data on the nature of the profit-motivated charges against

our study officers. Bribe taking, the solicitation or acceptance of money or some other benefit

in return for abusing one=s authority to benefit wrongfully benefit another, is the single most

frequent charge (n=72), followed closely by grand larceny, which is felony level stealing

(n=66). Twelve of the study officers were charged with defrauding the government (e.g., by

welfare or food stamp fraud); 28 were charged with burglary; seven with operating or

working for illegal gambling enterprises; 33 with insurance fraud; 28 with misdemeanor-grade

petit larcenies; ten with soliciting or receiving gratuities or accepting goods and services

without

To clarify the differences again: Figure VIII-1 considers the priority and order

of charges lodged against officers and includes data on only the first and/or worst

offense. Figure VIII-2 simply includes all charges. Thus, if an officer were charged with

trafficking drugs that he had stolen from an arrestee, he would be included in Figure

VIII-1 in the Adrugs@ category. In Table VIII-2, his case would also be included in the

Aprofit-motivated crime@ category.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-2: All Specifications Against Study Officers

286

144

119

266

742

468

387

Fyfe and Kane -- 134

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-3: Profit-Motivated Charges Against Study Officers

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



136

paying for them; 15 with improperly appropriating property in their custody; and ten

each were charged with accepting gratuities or engaging in such illegal occupations as

working in liquor-related businesses, which is forbidden to police officers by New York State

law. An additional 106 charges involved profit-motivated charges not specified on NYPD=s

coding schema (e.g., extortion; receiving stolen property; abusing or misappropriating official

resources; robbery).

Thus, while the traditional police occupational crime B bribery B is the modal category

among these offenses, it is clear that offenses not necessarily related to accused officers=

employment (e.g., insurance fraud; petit larceny) also occur with some frequency.

Why So Few Separations for On-Duty Abuse?

Clearly, therefore, we can accept this hypothesis, but the data raise other issues that are

worthy of discussion. In January 1975, when the period covered by our study began, the

NYPD employed more than 32,000 officers. Over the next 22 years, more than 45,000

additional officers joined the ranks of the NYPD, so that some 78,000 people served as NYPD

officers during the period we studied. Only 1,543 of these officers B 2.0 percent, or one in 50

B left the Department involuntarily, for reasons of improper behavior. This should be kept in

mind when reading the rest of this report because, even when we discuss apparently large

differences in rates of involuntarily separation, we are talking about only a very small

percentage of the NYPD=s total personnel complement. On an annual basis, the 1,543 officers

separated from the NYPD represented about one-tenth of one percent B one in a thousand B of

the total number of officers who served in the Department during these 22 years.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



137

Second, only 119 of these officers B about one in 563 of the officers employed by

NYPD during these years B ended their careers on charges that included on-duty abuse. For

readers familiar with media accounts of the Louima and Diallo cases and with allegations of

widespread police abuse, this may seem to be an extremely small number. Two explanations

probably apply to these abuse-related separation figures. The first is that, media accounts

aside, the actual incidence of excessive force by New York City police is far smaller than

many have been led to believe and, further, that the decreases in separations for such offenses

may represent real decreases in their frequency. There is considerable evidence in support of

this contention. As Chevigny (1996) and Domanick (1994) suggest, the NYPD has long been

noted for its attempts to control its officers= discretion in use of force. The literature suggests

that the department has been successful: the NYPD traditionally has used deadly force

proportionally less often (e.g., per 1,000 officers) than is true of most other large police

agencies (see, e.g., Geller and Scott, 1992; Washington Post 2001). Further, Skolnick and

Fyfe (1993:21-215) reported, the NYPD has been the subject of civil rights complaints to the

United States Department of Justice much less often (again, per 1,000 officers) less frequently

than any of 58 other municipal and county agencies that had been the subject of a federal

study on the subject (United States Department of Justice,1992).

Rates of civil rights complaints for the nation=s ten largest cities during 1985-

90, including New York=s, are presented in Figure VIII-7.

The second explanation involves the difficulty of proving allegations of police abuse.

This difficulty involves the dilemma of proving that officers B who are entitled by law and

policy to great discretion in use of force B have crossed the barrier from reasonable to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



138

unreasonable force. This is a very difficult burden of proof in either the criminal courts or in

police administrative proceedings. Consequently, findings of wrongdoing in either forum

usually only occur when evidence is irrefutable. This was the case in the Rodney King

matter, where the now-famous videotape contradicted the accused officers= claim that they

had used only necessary and reasonable force. It was also true of the Abner Louima matter,

in which the nature of the victim=s injuries B a ruptured colon and bladder caused by forcible

sodomy with a stick B could not be explained by any reasonable variety of police force. In the

absence of such objective evidence, accusations of brutality devolve into swearing contests

that cannot easily be resolved by finders of fact.

This is not so where profit-motivated misconduct is concerned. Since there are

circumstances in which officers can reasonably employ force, the mere fact that an officer

acted forcibly does not indicate wrongdoing. But there are no circumstances in which officers

legitimately can accept cash or benefits other than their pay checks. Consequently, the mere

proof that such an event has occurred proves dismissible wrongdoing.

Thus, what may seem like an inexplicably low frequency of force-related dismissals

probably has two explanations. The first is the comparative infrequency of force by NYPD

officers, and the second is the difficulty of proving that force has been excessive. Finally,

readers should keep in mind that, even when we are discussing involuntary separations for all

reasons, we are considering only a very small percentage of the NYPD: our data bear out the

frequent claim that the overwhelming majority of police officers are honorable public

servants.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



139

COMMUNITY DEMOGRAPHICS AND POLICE BEHAVIOR

We approached this avenue of research by using data obtained from the U.S. Census,

and the NYPD=s Crime Analysis Unit to test the following research hypothesis:

H2: Rates at which officers are involuntarily separated for corruption and

brutality, as well as for other job-related misconduct, vary over time with changes in

the following characteristics of the communities in which they are assigned:

H2a: rates of public homicide (positive);

H2b: FBI Index Crime rates (positive);

H2c: percentage minority population (positive);

H2d: per capita income (negative).

The Importance of Place and Time

Thus far, we have examined patterns of police misconduct at the individual officer

level, comparing the population of Astudy@ officers to a sample of Acomparison@ officers. As

the literature review indicates, however, a large body of research has examined the influences

of community characteristics on police coercion (e.g., Smith, 1986), deadly force (e.g., Fyfe,

1980), and even police dog deployments and bite-to-use ratios (Campbell et al., 1998),

generally showing the importance of incorporating structural (i.e., community-level) variables

as predictors of police behavior. The present study advances this research by examining the

effects of structural variables on police misconduct at the community level, largely building

on Fyfe=s (1980) examination of patterns of deadly force in the NYPD across police

divisions. It is hoped that the community-level perspective will add an important dimension to

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



140

the present study by identifying the role of place and social context in the prediction of career

ending police malpractice.

As the unit of analysis changes from the officer to the community, several theoretical

and methodological assumptions also change as the present study draws heavily on the

communities and crime research to inform its tests of the structurally-grounded research

hypotheses. Since Shaw and McKay=s (1942) seminal examination of social disorganization

and delinquency in Chicago, a series of subsequent researchers has further elucidated the

importance of examining how changes in community structure over time may predict temporal

variations in crime rates in those local areas (Bursik & Webb, 1982; Bursik, 1986; Reiss,

1986; Sampson & Groves, 1989; Bursik & Grasmick, 1993; Kubrin & Weitzer, 2003).

Moreover, as Bursik and Grasmick (1993) argued, an integrated structural model may show

that variations in social ecological conditions may explain outcomes beyond crime and

delinquency. Consistent with this reasoning, the social disorganization framework has been

specified to explain fear of crime (Taylor & Covington, 1993) and rates of violence among the

mentally ill at the neighborhood level (Silver, 2000). The present study further expands these

outcomes by examining patterns of police misconduct B a related measure of territorial

deviance B within communities over time.

Incorporating Time

Although the present study primarily examines cross-sectional research hypotheses,

several researchers have convincingly argued the importance of incorporating time into

studies of deviance that specify the community as the unit of analysis (Bursik, 1986; Bursik &

Webb, 1982; Sampson & Groves, 1989; Kubrin & Weitzer, 2003). Indeed, as Shaw & McKay

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



141

(1942) noted, social disorganization B which largely represents an opportunity theoretical

perspective B is hypothesized to predict variations in deviance patterns through disruptions of

the Amoral order@ of a community. Such disruptions are hypothesized to occur as the result of

changes in the structural (i.e., social and economic) conditions at the community level, which

are best understood when examined over time. Moreover, Shaw and McKay (1942:14)

observed, changes in neighborhood-level deviance patterns are partly a function of the

Ahistory and growth of the city and of the local communities that comprise it.@ Finding

support for this argument in a reexamination of Shaw and McKay=s Chicago data, Bursik

(1986) showed that using a cross-sectional methodology to test the ecology of crime model

can produce spurious correlations between predictor and outcome variables.

Following the lead of previous researchers who have tested the ecology of crime

model, the present study examines the effects of structural conditions on police misconduct,

which are hypothesized to create opportunity structures for occupational deviance in police

precincts over time. In conducting the longitudinal hypothesis testing (recall H2a-H2d above),

the present study employs an analytical strategy capable of distinguishing the within-precinct

changes from the between-precinct variations in the social ecological conditions that should

theoretically influence patterns of police misconduct. The following section describes these

analytical strategies in detailed form.

Analytic Procedures

The precinct-level misconduct data were submitted to a series of repeated measures

analyses using the hierarchical linear modeling procedure developed by Bryk & Raudenbush

(1992). The application of HLM to longitudinal data analysis has been referred to as growth

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



142

curve modeling (Bryk & Raudenbush, 1992; Lauritsen, 1998; Raudenbush & Chan, 1987),

which identifies variations in the outcome variable(s) of interest, then enters a series of

predictor variations in subsequent models that are hypothesized to explain the outcome

measure variations. (Roundtree et al., 1994). In replicating the procedures outlined by Bryk &

Raudenbush (1992), we examined the relationships between the community-level variations

and misconduct in two stages. The first estimation shows whether precincts have varied

significantly across each other in their levels of police malpractice at the intercept year (1986),

as well as the extent to which changes in the patterns of misconduct within precincts have

varied over time. This estimation is important because if the model shows that police

misconduct has not varied significantly over time within precincts, then there is no

justification for a longitudinal examination of the community-level variables on misconduct.

This initial model may be represented as a function of a systematic trajectory over time, plus

random error. The equation is specified as follows:

= ϑ

+ ϑ

(Time

) + ϑ

(Time

) + ϑ

(Spatial Term

) + e

Where:

= the observed career-ending misconduct rate for each precinct I (For i = 1,..., n

units) at time t (for t = -10,..., 10);

, ϑ

2i,

= the intercept, slope, and quadratic parameters, respectively, of the trend

lines;

= the spatial autocorrelation term for each areal unit at time t;

= the random within-areal unit error estimation for unit I at time t.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



143

Assuming significant within-group variance over time is observed, a series of

subsequent models will be estimated that include the measures of community characteristics

hypothesized to explain the levels of misconduct over time. These indices will be entered as

time-varying covariates, which may be expressed by the following equation:

= ϑ

+ ϑ

(Time

) + ϑ

(Time

) + ϑ

(Spatial Term

) + ϑ

+ ϑ

+ e

Where:

= the observed career-ending misconduct rate for each precinct I (For i = 1,..., n

units) at time t (for t = -10,..., 10);

, ϑ

2i,

= the intercept, slope, and quadratic parameters, respectively, of the trend

lines,

= the spatial autocorrelation term

for each areal unit at time t;

= rates of public homicide (positive);

= FBI index crime rates;

= percentage minority population (positive);

= per capita income (negative);

= the random within-areal unit error estimation for unit I at time t.

In spatial analyses of crime and delinquency, it has sometimes been found that

deviance levels in communities are influenced by deviance levels in adjoining

communities (Morenoff & Sampson 1997; Taylor & Covington, 1988). Failure to

account for this spatial dependence in the outcome measure can bias the results of the

overall model. Thus, when theory suggests that a spatial process may be influencing the

patterns under study, it is important to estimate the effects of these potential influences.

In the present study, the issue of spatial autocorrelation represents an open empirical

question. Therefore, relying in Taylor and Covington=s (1988) OLS procedure, controls

for spatial autocorrelation are included as explanatory variables in all predictive models.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



144

In the analyses that follow, the residuals are allowed to over-disperse. The pure

Poisson distribution constrains residual variance to equal the fitted mean, but actual data

generally have more variance than that. HLM has a feature that allows for the opportunity to

examine the residual variance through an over dispersion parameter. This is noted as σ

in the

tables. In addition, the time-varying covariates are group-mean centered, which allows for a

proper assessment of within-territorial unit change over time.

Group-mean centering limits coefficients of these variables to represent within-

territorial unit change, rather than mixing that with between-territorial unit differences

(Bryk & Raudenbush, 1992).

A time variable was created by subtracting the midpoint of the time series, 1986, from

each year in the study period. Therefore, 1979-1986 = -7 (which represents 7 years prior to the

midpoint), while 1989-1975 = 14 (which represents 14 years beyond the midpoint). This

variable TIME will model a linear trend. To model a quadratic parameter the variable

TIME_SQ was computed by raising TIME to the second power.

Finally, all models specify a Poisson distribution with variable exposure where the

exposure variable is the precinct population of police officers. Initially, rates of misconduct

for all precincts (and divisions) were created, but there were zeros in many of the data cells

since not all precincts experienced misconduct every year. To counter this, the number of

misconduct events (i.e., CASES) represents the dependent variable, which is weighted by the

precinct population of police officers (i.e., a count data model). This procedure decreases the

potentially adverse effect of modeling a rate-based dependent variable that contains a sizable

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



145

number of zero scores. Note that the procedures outline here are replicated for all division-

level analyses.

Independent Variables

The analyses include seven indicators of crime drawn from both manual and electronic

data sources maintained by the New York City Police Department. These include rates of

homicide, felony assault, robbery, forcible rape, burglary, grand larceny, and auto theft B all

of which are aggregated to the precinct-level. In addition, the analyses include indicators of

precinct-level per capita income and percent minority populations. Both of these structural

conditions have been associated police-citizen conflict and community dependence on police

(Jackson, 1989), and may provide opportunity structures for police misconduct independent of

precinct-level crime rates.

It should be noted that the rates of property crime were strongly intercorrelated, as

were the rates of violent crime. To reduce potential multicollinearity among the independent

variables, composite crime measures were created as summed indices via principal component

analysis. The model included precinct-level rates of homicide, forcible rape, robbery,

aggregated assault, burglary, grand larceny, and motor vehicle theft. Table VIII-1 shows the

results of this analysis.

As the data in Table VIII-1 indicate, The PCA extracted two dimensions of public

crime,

which are best described as violent crime (Eigenvalue=3.68; variance explained=52.5

%) and serious property crime (Eigenvalue=1.87; variance explained=26.7 %). To determine

The first PCA was specified to extract all dimensions with Eigenvalues greater

than one, however because the difference between the second and third dimensions was

large, we were concerned that the model Aover factored@ the data. As a reliability check,

we estimated a second PCA, forcing a two factor solution. The results of the estimations

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



146

Table VIII-1: Principal Component Analysis of New York City Crime Rates

Components

1 2

Rates of:

Homicide

Forcible Rape

Robbery

Aggregated Assault

Grand Larceny

Burglary

Motor Vehicle Theft

.900

.920

.893

.900

.076

.434

-.215

-.211

.120

.208

-.233

.784

.788

.616

were substantively equal, and we report on the initial model.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



147

the reliability of these measures, the variables loading on each component were submitted to

Cronbach=s Alpha scale reliability analysis, producing the following results:

Violent Crime index: forcible rape, homicide, felony assault, robbery; Alpha=.80

Serious Property Crime index: burglary, grand larceny, motor vehicle theft; Alpha=.75

The variables that made up each dimension were summed to create additive indices of

violent crime and serious property crime. It is understood that an additive index weights each

variable equally in contrast to a PCA, which weights each item differentially on the basis of its

contribution to the component. The purpose here was to create mutually exclusive composite

crime measures that shared little covariation. As such, we favored the additive indices over the

principle component scores because although the latter produce relatively uncorrelated

dimensions, the dimensions are not necessarily mutually exclusive; burglary, for example,

loads to some extent on both the violent crime and the serious property crime dimensions. The

relatively large reliability coefficients produced by the Cronbach=s Alpha provided the

necessary empirical confirmation to justify our decision.

HLM Results

Table VIII-2 shows the results of the unconditional repeated measures HLM. Both the

base rate parameter (i.e., cross-sectional) and the quadratic trend parameter were significant,

suggesting that police misconduct varied both between-precincts at the intercept year, and

within-precincts over time. Moreover, as the variance components show, the majority of the

variation was within-precincts over time (σ

= 1.284; p< .001), as opposed to between them.

This is a key methodological issue since the findings show the importance of modeling police

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



148

misconduct over time, rather than between precincts at a single Asnap-shot@ point (i.e., cross-

sectionally).

A note on the reliability coefficient is warranted. In a repeated measures design

the reliability coefficients have utility for showing how much of the variability in the

slope is due to random error versus parameter variance (Lauritzen, 1998:137). As the

reliability increases, so does the confidence that the outcome measure contains enough

variability to support a within-individual analysis over time. Because the reliability

coefficient produced by the precinct-level analysis was low (i.e., < .60), support for the

within-precinct growth model is necessarily qualified.

Finally, note that the spatial autocorrelation control variable was not significant,

indicating that police misconduct patterns in adjoining precincts do not influence one another.

This finding is supported by theory, as prior researchers have argued and/or observed that

police districts (i.e., station houses) within the same municipality tend to operate

autonomously of one another, and officers assigned to the districts tend to have little contact

with officers assigned to contiguous stations (Klinger, 1997; Rubenstein, 1972; Reuss-Ianni,

1983).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



149

Table VIII-2: . Initial Hierarchical Poisson Model Estimates of Within- and Between-Precinct

Police Misconduct in New York City

Fixed Effects

Predictor Coefficient S.E. T-ratio P-level

For base rate, ϑ

Intercept, β

For linear change, ϑ

Intercept, β

For quadratic effect, ϑ

Intercept, β

For Spatial Autocorrelation, ϑ

Intercept, β

-5.955

-0.011

-0.019

0.089

0.067

0.018

0.002

0.078

-87.87

-0.041

-9.02

1.132

<.001*

.968

<.001*

.258

Variance Components

Parameter Estimate X

P-level Reliability

Var (ϑ

)

0.130

1.284

123.11 <.001 .445

*p< .001

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



150

ubstantively, the data in Table VIII-2 show that, from 1975 to 1996, police misconduct

within precincts fluctuated significantly over time but that, because the linear component was

not significant, the fluctuation was neither an additive increase nor decrease. Rather, given

that the quadratic time variable was centered at the midpoint, the significant Beta indicates a

parabolic misconduct pattern within precincts over time. Graphically, the parabola would be

illustrated as either a AU@ or an inverted AU.@

Even in the absence of predictor variables, this is an interesting finding, which may be

explained by several factors. Skolnick and Fyfe (1993) reported that, in the wake of the

Serpico scandal in the NYPD, Mayor John Lindsay appointed Patrick V. Murphy as a reform

police commissioner. Skolnick and Fyfe wrote:

Even before the Knapp hearings started, Murphy had

begun changing his department irrevocably...Murphy put in

place systems to hold supervisors and administrators strictly

accountable for the integrity and civility of their personnel...He

won approval to increase the number of supervisors authorized

for the department...He rewarded cops who turned in corrupt or

brutal colleagues and punished those who...looked the other way

when they learned of misconduct.

...(Murphy) contracted with the RAND Corporation to

study his executives= jobs and to determine what knowledge,

skills, and abilities were necessary to succeed in it. He gave

those results to the American Management Association and

hired the association to assess each of his executives to identify

their strengths and weaknesses. (Skolnick & Fyfe, 1993, 179)

Murphy implemented Afield associate@ and Aintegrity testing@ programs, designed to

covertly detect and respond to corruption among police officers (Murphy & Plate, 1977).

Additionally, and perhaps most importantly, Murphy also made it impermissible for organized

crime officers to enforce laws against Avictimless crimes,@ such as prostitution and illegal

gambling unless complaints about these activities originated from outside the department

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



151

(Skolnick & Fyfe, 1993:180). Though it is not possible to test the potential influence that

Murphy=s administration may have had on the misconduct patterns of the NYPD, it is

possible that his policies led to a rapid decrease in misconduct in the immediate aftermath of

the Serpico scandal. However, as the institutional Amemory@ of Serpico and the resulting

Knapp Commission may have decreased over time, misconduct may have increased to pre-

Knapp levels, reaching a peak in 1987. The rapid decline in police misconduct after 1987 may

have been caused by either environmental or organizational conditions B or by a combination

of both. It may be the case that opportunities for police misconduct may have diminished in

the police precincts. It may also be that the observed decline in misconduct was a function of

the vigor with which the NYPD has attempted to detect and sanction police malpractice over

time.

Given that the initial HLM model showed significant variation in police misconduct

over time, a subsequent hierarchical linear model was estimated to test the research

hypothesis:

H2: Variations in community structure (i.e., per capita income,

percentage minority population) and public crime (homicide, FBI index

crimes) will predict variations in police misconduct within police

precincts over time.

Several models were estimated to test this research hypothesis. First, we examined the

relationship between the homicide rate, per capita income, and minority populations, and

police misconduct. Next, we re-estimated the model, substituting the index crime scales

(property crimes, and violent crimes) for the homicide rate, while controlling for per capita

income and minority population. Tables VIII-3 and VIII-4 show the results of these analyses.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



152

Table VIII-3: Hierarchical Poisson Estimates of the Influence of Homicide, per Capita

Income, and Percent Minority Population on Within- Precinct Police Misconduct in New York

City (No Control Variables Included)

Fixed Effects

Predictor Coefficient S.E. T-ratio P-level

For base rate, ϑ

Intercept, β

For linear change

, ϑ

Intercept, β

For quadratic effect, ϑ

Intercept, β

For Spatial Autocorrelation, ϑ

Intercept, β

-5.955

-0.001

-0.019

0.077

0.067

0.018

0.002

0.080

-87.87

-0.041

-9.029

0.961

<.001*

.968

<.001*

.337

Time-Varying Covariates

Homicide Rate,β

per Capita Income β

Precinct Minority Population,β

3.98

-1.81

0.84

.060

.642

.002

2.54

-2.82

2.37

.027**

.005*

.019**

Variance Components

Parameter Estimate X

P-level Reliability

Var (ϑ

)

0.122

1.649

124.80 <.001 .533

A summed index that includes felony assault, robbery, homicide, and rape

*p# .01; **p< .05

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



153

Table VIII-4: Hierarchical Poisson Estimates of the Influence of Index Crime, per Capita

Income, and Percent Minority Population on Within- Precinct Police Misconduct in New York

City (Control Variables Included)

Fixed Effects

Predictor Coefficien

S.E. T-ratio P-level

For base rate, ϑ

Intercept, β

For linear change, ϑ

Intercept, β

For quadratic effect, ϑ

Intercept, β

For Spatial Autocorrelation, ϑ

Intercept, β

-5.955

-0.001

-0.019

0.077

0.067

0.018

0.002

0.080

-87.87

-0.041

-9.029

0.961

<.001*

.968

<.001*

.337

Time-Varying Covariates

Violent Crime Index, β

Property Crime Index, β

per Capita Income, β

Percent Minority Population,β

2.13

1.54

-0.84

1.81

.660

.554

.002

.010

3.23

2.10

-2.37

2.82

.001*

.037**

.019**

.005*

Variance Components

Parameter Estimate X

P-level Reliability

Var (ϑ

)

0.122

1.649

124.80 <.001 .533

*p# .01; **p< .05

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



154

As the data in Tables VIII-3 and VIII-4 show, the findings from both model

estimations support the research hypothesis that indicators of community structure have

predicted variations in police misconduct over the study period. The conflict literature

suggests that geographic areas having high percentages of minority populations (Blalock,

1967; Jackson, 1989; Liska, Lawrence, and Benson, 1981) and low per capita income levels

(Jacobs & Helms, 1997) may be characterized by police-citizen as residents place increased

service demands on the police, while being resentful of the (real or perceived) associated

increases in coercion. The Apublic safety@ perspective (i.e., Fyfe, 1980) suggests that

increases in public crime across and within communities create opportunity structures for

police misconduct, generally in the presence of decreased mechanisms of informal social

control (see also: (Reiss, 1971; Sherman, 1990; Skolnick, 1966; Skolnick and Fyfe, 1993).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



155

LOCAL POLITICAL CULTURE AND POLICE BEHAVIOR

As suggested earlier, there exists in some circles a belief that adherents of community-

oriented policing are more likely than adherents of vigorous, enforcement-oriented styles of

policing to punish harshly officers who engage in non-profit abuses of citizens. Especially

during the last half of the period covered by this study, the NYPD moved between these

styles. Mayor David Dinkins and his two police commissioners, Lee P. Brown and Raymond

W. Kelly, were strong advocates of community-oriented policing. Dinkins=s successor,

Rudolph Giuliani, who eschewed community-oriented policing, won his reputation and his

office as a no-holds-barred crime fighter, and frequently was accused of tolerating police

excess in the name of law and order (Grunwald, 1999; McArdle and Erzen, 1998; Riverdale

Press,1997). Thus, assuming that Mayor Dinkins would be less tolerant than Mayor Giuliani

(or Mayor Edward Koch) of police brutality, we generated the following hypothesis:

H3: The rate of involuntarily separations for brutality and other non-profit abuse of

citizens was greater during the administration of David Dinkins than during either the

Koch or Giuliani administrations.

Conversely, to test a widespread perception about police misconduct other than

brutality, we hypothesized:

H4: Rates of involuntary separation for reasons other than brutality and non-

profit abuse were greater during the administration of Rudolph Giuliani than

during either the Koch or Dinkins administrations.

As Figure VIII-2 illustrated earlier, a total of 119 involuntary separations included

some allegation of on-duty abuse of citizens. Thus, over the 22 years studied, slightly more

than five officers per year (5.4) were removed from the department for charges that included

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-4: Primary Charge by Year

Separation Year

Number

Profit-motivated crime Drugs On-Duty Abuse

Fyfe and Kane --



156

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



157

brutality or related offenses. During these same years, 1,424 officers were separated on

charges that did not include brutality (67.4 per year). The data in Figure VIII-4 show that

separation frequencies peaked during the 1980s, in the administrations of Mayor Edward

Koch and Police Commissioners Robert McGuire and Benjamin Ward. Separations for

charges including on-duty abuse peaked during the period 1983-88, when the NYPD released

54 officers for such wrongdoing. During the four years of the Dinkins administration, (1990-

93), 15 officers were fired for charges that included on-duty abuse, while 13 were dismissed

on such charges in the three years of the Giuliani administration covered by our data.

Figure VIII-5 translates these numbers, as well as those that test H4, into annual rates

per 1,000 officers, for the three mayoral administrations included in our two hypotheses. The

figure demonstrates that, when controlling for the size of the NYPD=s personnel

complement,

the disparity between abuse separations during the Koch administration and

those during the administrations of his two successors remains. The rate during the Koch

years (0.25 per 1,000 officers) is about twice as high as the Dinkins and Giuliani rates (0.14

and 0.12, respectively). Thus, the data cause us to reject H3: there is little or no difference

between the rates of dismissals for on-duty abuse rate of dismissals in the Dinkins and

Giuliani years. Further, the rates during both these periods were much lower than was the

case while Edward I. Koch was New York=s mayor.

The NYPD averaged 26,795 officers annually during the Koch administration;

26,955 during the Dinkins administration; and 34,964 during the Giuliani years.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-5:Mean Annual Rate per1,000 Officers of Separations for On-Duty Abuse,by Mayor

0.25

0.14

0.12

Koch (1978-89) Dinkins (1990-93) Giuliani (1994-96)

Fyfe and Kane --



158

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



159

`The same figure compels us to reject H4. Here, the presentation shows that, during

the Koch administration, officers were separated from the NYPD for reasons that did not

include on-duty abuse at an annual rate of 2.88 per 1,000 officers. The comparable rates for

the Dinkins and Giuliani administrations are 1.72 and 1.88 per 1,000, respectively. Thus, it is

clear that, both in terms of raw numbers and absolute frequencies, far more officers were

forced out of the NYPD during Edward I. Koch=s 12 years in office than was true of either

the Dinkins or Giuliani administrations.

Why So Few Dismissals for Abuse?

One aspect of these data that may merit discussion is the overall frequency of

separations related to on-duty abuse. During the years covered by our study, the NYPD

employed about 78,000 officers and separated 119 on charges that included on-duty abuse.

Readers most familiar with the NYPD through reportage of its scandals may be surprised at

the disparity between these numbers. One factor that may help to explain them is the relative

infrequency with which NYPD officers use force of any kind. Paul Chevigny has noted that

the NYPD has not been known for widespread abuses, and that its administrators have long

devoted great effort to controlling street-level officers= discretion (Chevigny,1995). There is

considerable evidence that these efforts have proven successful in minimizing NYPD

officers= use of both legitimate and illegitimate force.

Figure VIII-6 is derived from a recent Washington Post series (Washington Post,

2001), and presents rates of fatal shooting per 1,000 officers in the ten largest U.S. cities

during 1990-2000. The figure shows that the NYPD, with an average annual fatal shooting

rate of 0.7 per 1,000 officers is the lowest in this group, a position it generally has held for

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

ure VIII-6: Mean Annual Fatal Shootin

Rates per 1,000 Officers

in Ten Largest U.S. Cities, 1990-2000

0.7

2.2

0.9

1.1

3.3

3.1

0.8

1.4

2.2

w Y

Source: Derived from the Washington Post, July 1, 2001.

Fyfe and Kane --



160

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



161

the last generation (see, e.g., Matulia, 1981,1985;Geller and Scott, 1992). Thus, where this

most extreme type of police force is concerned, the NYPD is at the bottom the scale.

A decade ago, in preparation for a Congressional investigation into police excesses, the

Civil Rights Division of the U.S. Justice Department prepared data that may speak to the

frequency of illegitimate police force in New York City during the period we studied. Figure

VIII-7 presents mean annual rates of civil rights complaints to the U.S. Justice Department per

1,000 officers across the ten largest cities during 1985-90.

Again, the data indicate that the

rate in New York City (0.5) is the lowest, with scores for the other cities ranging from 0.9 in

Detroit to 14.4 in San Antonio. Thus, to the extent that the infrequency of separations for on-

duty abuse are a surprise, it is likely that they reflect low rates of use of force, both legitimate

and allegedly illegitimate, that may also surprise some readers.

Prosecutorial Difficulties. Another probable explanation of the low separation rate for

on-duty abuses is the difficulty of prosecuting such cases, either in the courts or in the

NYPD=s administrative proceedings. Many of the other offense categories in our study

present prosecutors only with the challenge of proving only that the accused committed the

acts with which they are charged. Under no circumstances are officers permitted to traffic or

sell drugs,

To our knowledge, no comparative data on lesser uses of force are available

from any source.

In its report to Congress, the Justice Department included simple frequencies.

Skolnick and Fyfe (1993:213-214) standardized these by calculating the mean annual

rates per1,000 officers presented in Figure VIII-7.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-7: Mean Annual Civil Rights

Complaints per 1,000 Officers

in Ten Largest U.S. Cities, 1985-1990

0.5

1.9

1.4

4.2

1.2

4.1

3.9

5.1

14.4

0.9

ew Y

Angele

Hou

phia

San D

ego

Phoen

Dal

las

onio

oit

Source: Jerome H. Skolnick and James J.Fyfe, Above the Law: Police and the Excessive Use

of Force (New York,: Free Press, 1993) pp.213-214; derived from U.S. Department of Justice,

Civil Rights Division, Criminal Section, APolice Brutality Study FY 1985-FY1990,@ (April

1991: unpublished manuscript), App. A-4.

Fyfe and Kane --



162

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



163

to beat their spouses, or to accept money other than their paychecks, so that merely

demonstrating that such acts occurred is sufficient to result in dismissal. But on-duty officers

are permitted, and even required, to use force when necessary, so that merely proving that a

use of force occurred means nothing. Instead, criminal and departmental prosecutors must

also show that the force used was in excess of that which was reasonable and necessary under

the circumstances. This typically requires prosecutors to prove a negative: that, at the instant

officers employed force, the degree of force they used was not reasonably necessary to defend

themselves and/or to take suspects into custody.

This is a vary challenging burden, and it does much to explain why separations for

excessive force typically have occurred only in cases in which evidence of abuse has been

indisputable. bsent the now famous videotape shot by amateur cameraman George Holliday,

the beating of Rodney King would have been written off as a case of necessary force against a

large and threatening felon. In New York, the Abner Louima atrocity resulted in convictions

and dismissals only because there was no reasonable explanation for the wounds inflicted on

Mr. Louima B a ruptured colon and bladder suffered when he was forcibly sodomized with a

stick. But the more common allegations B of beatings on dark streets or in police station

backrooms B frequently devolve into swearing contests between complainants and officers

who claim that they used only reasonable and necessary force. Absent objective evidence of

what happened, these complaints almost invariably end in findings of Anot sustained,@ and

never get to either the department=s Trial Room. or the criminal courts.

Police Solidarity. The ambiguity concerning on-duty abuse creates another

prosecutorial difficulty. As we began our research B indeed, during the week immediately

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



164

prior to the Louima atrocity B we conducted three focus groups, one each for patrol officers,

patrol sergeants, and patrol precinct commanders. The consensus among all three groups was

that, while there was virtually no tolerance among patrol officers for profit-motivated

misconduct on the part of their peers, there was a tendency for officers to be more forgiving

where excessive force was concerned. All three groups told us that any officer who engaged

in money corruption or drug-related misconduct could expect to be quickly reported to

internal investigators. Where on-duty abuse was concerned, all three groups told us that there

existed a regrettably higher degree of tolerance among the ranks. This evidenced itself with

events that began a few days later, when investigators found apparent collaboration by patrol

officers to shield one or more of their colleagues from accountability in the Louima matter

(McAlary, August 13, 1997). The response to this case included an internal investigation that

won awards for its thoroughness and for penalizing those involved in concealing this episode,

as well as policy changes mandating dismissal for false statements, and has no doubt affected

this view. Still, it should be no surprise to observers of the police that there may exist a

greater degree of tolerance among police ranks for excessive force B which can be treated as

overzealousness B than for greed-induced money corruption. To the extent that it continues to

exist in the NYPD post-Louima, it certainly complicates the prosecution of brutality cases.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



165

POLICE SOCIALIZATION, TRAINING, AND POLICE BEHAVIOR

We generated several hypotheses designed to test the relationship between various

aspects of police training and rates of involuntary separation. These included the following:.

H5: Rates of involuntary separation for corruption, brutality, and other forms of

misconduct are positively associated with the size of Police Academy recruit training

cohorts.

H5a: The rate at which new officers are involuntarily separated as

unsatisfactory probationers is inversely correlated with the size of Police

Academy recruit training cohorts.

H6: Rates of involuntary separation for corruption, brutality, and other forms of

misconduct are inversely associated with the rigor and intensity of recruit training, as

measured by:

H6a: numbers of hours of training;

H6b: rates of involuntary separation for academic and physical failure, and

for disciplinary reasons.

As it turned out, our ability to adequately test these hypotheses was limited. H5a and

H6b proved untestable because it was impossible to distinguish with precision recruits who

had been involuntarily separated (by termination or by resignation in lieu of termination) from

those who simply resigned for legitimate personal reasons and/or because they believed that

they were en route to failure in the Police Academy or in a scheduled drug test. In every

recruit class, some unknown number of officers leave voluntarily and in good standing, rather

than face termination for failing an upcoming drug test or written or physical examination.

H6 and H6a proved untestable because of the imprecision or absence of Police Academy

records concerning the number of training hours for most of the years in our study. We also

encountered a methodological problem that, as far as we B and all the experts with whom we

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



166

have spoken B can determine makes it impossible to test H5 in a way that would satisfy

methodological purists.

The problem, in a nutshell, is that we have studied two different cohorts of officers.

We have collected data on officers who were separated during 1975 through 1996, but who

were hired as far back as 1946. This makes it impossible to determine the relative risk of

separation across the recruit classes included in our data because the earliest among them

entered police service a full 29 years before the start point for our separation data. Since the

average length of service before separation among our study officers is far less than this (6.8

years including officers who entered the NYPD before 1/1/75; 3.7 including only officers

hired after 1/1/75),

there is no way to compare the experiences of pre-1975 classes with

those hired later. Similarly, comparison of the histories of recruit classes hired late in our

study with those who were hired early, and who thus experienced many years of exposure to

the risk of being separated, is imprecise.

Even absent the ability to provide reliable answers to these research questions, they are

among the most important in our study and, therefore, warrant close examination. We begin

this in Figure VIII-8, which presents data on officers hired during 1965 through 1995 (thus

including 1472, or 95.4%, of our 1,543 study officers). The solid line on the chart shows that

the annual number of officers hired by the NYPD (presented in hundreds) has varied

dramatically over the years studied, ranging from none at all during the 1975-1978 fiscal crisis

The difference between these two averages, of course, is largely attributable to

the exclusion from our pre-1975 hirees of officers whose careers ended early.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-8: Officers Hired 1965-1995 and Class Separation Rates During 1975-1995

Hundreds Hired Rate per 1000

Fyfe and Kane --



167

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



168

to a peak of 3,444 during 1984. As one would expect, in large measure because of varying

periods of exposure to risk of separation among these annual cohorts, the separation rates also

vary dramatically. But even between cohorts that are closely placed, there are great

differences. The class of 1972 had a rate of 8.73 per thousand officers (two of 229), while the

classes of 1971 and 1973 have been separated at rates that are four or more times higher

(37.36, based on 17 separations among the 455 member 1971 class, and 32.59, for 115

separations among the 1973 cohort). The rate then dips again in 1974, to 13.62 (37 of 2,716).

Another dramatic change in rates occurs between the 1979 and 1980 classes: the former had a

rate of 26.57; the latter has a rate of 68.33. Clearly, therefore, there are variations in these

rates that are not explained by differential exposure by virtue of different lengths of service

during the period studied.

Our data indicate that most involuntarily separated officers leave the NYPD in the first

few years of their careers. Figure VIII-9 presents the years in service among involuntarily

separated officers who were hired between 1979

and 1996. Half (521, or 50.4%) of the

1,033 officers represented on the figure were separated during their first three years in the

department. Nine in ten (908, or 90.2%) of the separations shown on the figure occurred

during officers= nine years on the job.

Thus, our examination of the extent to which

Recall that the NYPD hired no officers during 1975-78.

As shown on this figure, the decrease in separations as years in service increase

is attributable in part to the fact that many of the officers included in the figure were parts

of classes that, by our 1996 end date, had served only short periods of time. No matter

how the data are cut, however, they show a very similar pattern: the careers of the

majority of officers who are separated end within a decade of their appointments to the

department.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-9: Recruit Class Size and Probationer Separation Rates, 1979-95

1980

1982

1983

990

1991

1992

1994

Hundreds Hired Terminated Rate per 1000

Fyfe and Kane --



169

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



170

variations in class size may have contributed to differences in class separation rates excludes

all officers who entered the NYPD prior to our study=s 1975 start date.

Variation by Class Year

Our test of H5 began with the decision to treat year of appointment to the NYPD,

rather than date of appointment, as our unit of analysis. We did this on the logic that, while

more than one recruit class frequently has been appointed during a single year, these cohorts

frequently overlapped with, or were merged into, each other. At the extreme, for example, the

NYPD hired new classes of officers in January, February, March, April, May, June, July,

September, and October 1968. In 1985, the department hired new groups in January, June,

and July; in 1987, new classes entered in January and April. Because these cohorts attended

the Police Academy alongside others, rather than one at a time, it is unrealistic to treat them as

discrete classes.

We then calculated annual involuntary separation percentages for each annual cohort

appointed during 1979-1995.

These are illustrated in Figure VIII-10, which shows them in a

cumulative format.

The figure=s tracks show some significant variation among the groups,

during its first three years in the NYPD, and had lost only one in 40 of members (2.37%) over

the 18 years included on the figure. The 1980 class, by contrast, lost one in 50 with the two

No officers appointed in 1996 had been separated by our study=s 12/31/96 end

date.

To ease interpretation, we added vertical, Axis X, gridlines to this figure. To

track each class=s experience on black and white versions of the figure, begin on its right

side with the AYear 18" gridline. The only class whose track crosses this line is the 1979

class (the only one of our cohorts in its 18

year by the study=s end date). The trend line

that ends after it crosses the AYear 17" gridline is the 1980 class; the line ending at

AYear 16" represents the 1981 cohort, and so on.,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-10: 1979-95 Classes' Cumulative Percent InvoluntarySeparations,

Career Years 1-18

r 1

1979 class 1980 class 1981 class 1982 class 1983 class 1984 class 1985 class 1986 class 1987 class

1988 class 1989 class 1990 class 1991 class 1992 class 1993 class 1994 class 1995 class

Fyfe and Kane --



171

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



172

extremes consisting of the 414-member.1979 class and the 600-member 1980 class. The

former, the first hired in five years, suffered no involuntary separations for cause of its

original members (1.83%) in its first three years. By its 17th year, it had lost one in every 12

of its original members (7.5%) to dismissals or forced resignations (1979 cumulative

percentage by Year 17=2.17).

We attempted in several ways to determine whether this variation was associated with

class size, and produced negative results, shown in Figure VIII-11, that cause us to reject H5.

We calculated a Pearson=s correlation measure between class size and cumulative separation

rates at Year 1 for all officers appointed during 1979-1995, and produced non-significant

results (r=.28; p=.29). We conducted the same analysis for 1979-1988 classes for cumulative

separation percentages at Year 9 and found similar results (r=.15; p=.68). Finally, to

determine whether class size was associated with dismissal rate after the completion of Police

Academy training, we tested the relationship between class size and cumulative separation

rates during Years 2-9. This also produced insignificant results (=.32; p=.37). Thus, it appears

that, taken alone, Police Academy class size has no measurable effects on subsequent

likelihood of involuntary separation.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

ure VIII-11: Annual Class Size (in Thousands) and Cumulative Percent

Involuntary Separations, 1979-1995 Classes

cla

81 class

las

83 cl

cla

1993

4 cl

ass

las

Hired Year 1 Year 2-9 Year 9

Fyfe and Kane --



173

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



174

Police Demographics and Representativeness

We generated two hypotheses related to whether two highly interrelated variables B

race and officers= place of residence (Fyfe, 1978, 1981b) B were associated with overall

department involuntary separation rates:

H7: Rates of involuntary separation for corruption, brutality, and other police

misconduct are inversely associated over time with the percentage of non-white

officers in the department.

H8: Rates of involuntary separation for corruption, brutality, and other police

misconduct are positively associated over time with the percentage of officers who

reside outside New York City.

As it turned out, historical data on officers= residence were extremely questionable.

Thus, we were unable to test H8 but, as the reader will find, we do examine and report on

relationships involving residence and separation later in this report.

To test H7, we obtained data on the racial composition of the NYPD during 1986-96,

excluding 1975-85 from examination because much of the data for those years was no longer

available.

We then calculated the percentages of non-white officers in the department for

Our data on officers= race come from two different sources. Since 1990, the

NYPD=s Office of Equal Employment Opportunity has annually published gender and

race specific data at January 31. The Personnel Bureau=s annual data on gender and race

are effective at June 30 of each year, and go as far back as1986. These data, however,

are not rank and race specific. Hence, in some of the analyses included in this report, we

have used a combination of EEO data for the years 1986-89 and Personnel Bureau data

for the years 1990-96.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-12: Annual Percentages of Non-White Officers

and Involuntary Separation Rates per 1,000 Officers, 1986-96

Rate of separation per 1000 officers % Black officers % All non-white officers % Other

Fyfe and Kane --



175

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



176

each year and, excluding Hispanics, Asians, Pacific Islanders, and Native Americans,

calculated the same annual percentages for black officers. Figure VIII-12 shows the results.

Before discussing B and putting in context B the test of this hypothesis, some features

of the figure are worthy of note. The figure illustrates, first of all, that blacks have represented

about half the department=s population of color, with their relative representation dipping

slightly over the years, even though their numbers have grown as a percentage of the

department. In 1986, one in five members of the department (20.8%) identified themselves as

other than white, with about one in ten (10.7%) classifying themselves as black. By 1996,

black officers comprised one in seven sworn personnel (13.7%), while all non-whites were

nearly one-third (31.6%) of the department. Thus, the representation of blacks has grown

slightly (from 10.7% to 13.7%), but the percentage of other non-white groups (largely

Hispanic and Asian) has grown more quickly (from 10.1% to 18.0%).

The tests of our hypothesis show varying relationships between separation rates and

our black and total non-white officer populations. The derived Pearson=s r describing the

relationship between percentage black and separation rate is not significant (r = -.35; p=.29).

In 1986, the NYPD employed 2,799 officers who identified themselves as

black; by 1996, this figure had almost doubled to 5,155. The 1986 NYPD roster included

2,505 Hispanics and 154 Asians; by 1996, the representation of Hispanics had more than

doubled, to 6,205,and the number of Asians had more then tripled, to 478. Some

percentage of these increases in minority representation is due to the merger of the

former, more heavily non-white, Housing and Transit police into the NYPD,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



177

The association between total non-white population also is inverse, but is even stronger (r = -

.63), and meets our .05 probability level for accepting the hypothesis (p=.04).

The relationship between overall separation rate and percentage non-white other than

blacks is even more strongly inverse (r = -.72, p=.01), driven largely by an extremely robust,

again inverse, relationship between the specific relationship between percentage non-white

other than blacks and their separation rate (r = -.78, p=.004; not shown on chart). It appears

that, as the numbers of people in Aother@ racial groups (including Hispanics, Asians, Native

Americans, and Pacific Islanders) have increased in the ranks, they have grown less likely to

be involuntarily separated from the NYPD. This experience is the converse of that of whites:

the relationship of white percentage in the department to white separation rates is positive (r =

.65; p=.03). The relationship between the percentage of black officers in the department and

their involuntary separation rate (also not shown in chart) is non-significant (r=0.29; p=0.37).

The race-specific separation rates underlying these statistics are shown in Figure VIII-

13. Here, we see that rates for all three groups have generally declined over the period studied

and that, throughout 1986-96, back rates have almost invariably been higher than those of any

other group. The black annual separation rate began at 16.1 per 1,000 officers in 1986, peaked

at 16.3 in 1987, and has since trailed downward to 6.5 in 1996. The rate for whites started at

4.9 in 1986,and has constantly decreased to 1.3 in 1996. The rate of other groups started at

6.4 per 1,000 and tracks across the Figure to 2.8 in 1996. Thus, the black rate often has been

two or more times higher than the rate for other groups and, over the figure=s last five years

(1992-96), the rates for whites and other groups have been virtually indistinguishable.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-13: Annual Involuntary Separation Rates by Race, 1986-96

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

White Rate Black Rate Other Rate

Fyfe and Kane --



178

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



179

What we see here, therefore, is that involuntary separation rates for all racial groups have

generally declined in the 1986-96 period. Further, the experience of Hispanic, Asian, and

other, more lightly represented but rapidly growing, racial groups has been more like than

of white officers than of black officers. This variation between blacks and everybody else

may be stated simply: during 1986-96, black officers were more than twice as likely as whites

or other groups to have been involuntarily separated from the NYPD. Despite this variation,

the absolute frequency of involuntary separations for all racial groups was quite low

throughout the period. Even during the worst years, only one in every 61 black officers B the

group most likely to be involuntarily separated B lost their jobs annually. In more recent

years, this has dropped to about one in every 153 black officers, while the rate for all other

groups has been one in every 357 per year. This does not paint a picture of epidemic

misconduct by any group. We will return later in this report to further discussion of these

findings and their causes and implications. For now, however, suffice to say that this is an

extremely important finding: it appears that, as the NYPD has become more diverse, it has

become significantly better behaved.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



180

Gender and Involuntary Separation

H9 posited that:

Female officers' involuntary separation rate is positively associated over time with the

percentage of females in the department.

We tested this by compiling NYPD gender data for the years 1986-96.

As Figure

VIII-14 indicates, we found that, as the percentage of females in the department increased

over the years, the rate at which they were involuntarily separated decreased (r= -.77; p =

.006). Thus, the hypothesis must be rejected: as female representation in the ranks has grown,

women officers have become markedly less vulnerable to involuntary separation.

As in the case of the preceding analysis of racial separation rates, however, it is not at

all clear that this trend is attributable to any deep-rooted sociological causes involving

relations between the genders. We included on the graph a line that tracks the separation rates

Again, data for 1975-85 were not complete. During these earlier years, 59

females were separated from the NYPD, as follows:

Year Number

1979 1

1981 5

1982 8

1983 14

1984 2

1985 26

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-14: Annual Involuntary Separation Rates by Gender, 1986-96

% Female in Dept Female Separation Rate Male Separation Rate

Fyfe and Kane --



181

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



182

of male officers over the same period. It discloses rates somewhat less than those for females

(e.g., from 8.39 and 6.06 in 1986 to 3.29 and 1.91 in 1996, for males and females,

respectively), but shows trends for both genders that are almost precisely identical (r=.96;

p=.001).

What makes them different, however, is that the separation rates for males have

decreased as their representation in the department has decreased, and that, like the rates for

Hispanics and Asians, the rates for females have decreased as the percentage of females in the

department have increased.

It appears, therefore, that our findings thus far on race, gender, and discipline are

heavily influenced by a trend over time to fewer separations generally: as the NYPD had

become more diverse, a smaller percentage of its officers of all genders and races has been

fired or forced to resign. Only the experience of black officers appears to be markedly

different from those of other identifiable groups but, even there, the trend is downward.

Gender and Probationary Terminations.

Our hypothesis on this issue is straightforward:

H10: The rate of involuntary separations of probationers is greater among female

officers than among males.

Our test of this hypothesis was limited to the years 1987 through 1995 because, prior to 1987,

data on class demographics were extremely spotty. Figure VIII-15 shows the results: in all but

two years, the female rate is higher than the male rate. Female probationers were

involuntarily separated from the NYPD at annual rates that ranged as high as 3.43 per

The mean separation rates for male and female officers during 1986-96 were

2.66 and 4.11, respectively.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

ure VIII-15: Probationar

Termination Rates b

Gender, 1987-95

0.5

1.5

2.5

3.5

male rate female rate

Fyfe and Kane --



183

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



184

hundred

(in 1988) and that, on average, were nearly twice as high as the male rate (1.54 per

100 versus 0.82 per 100). This confirms our hypothesis.

Gender and Corruption. The hypothesis we specified to examine the relationship

between gender and police corruption was

H11: The rate of involuntary separations for corruption is greater among male

officers than among female officers.

As we indicated earlier, distinguishing what is commonly understood to be police

corruption from other forms of profit-motivated misconduct by police officers is a daunting

task. When our data are examined to identify officers who had been separated for bribe

receiving, the prototypical act of police corruption, they clearly confirm our hypothesis:

during 1975-96, 65 male officers and two female officers ended their NYPD careers on

charges that included bribe receiving.

Because the career-ending corrupt acts in our data include far more than bribery, this

offense does not provide a suitable standard for determining whether males or females are

involved in entrepreneurial wrongdoing at disparate rates. Thus, we tested the hypothesis by

including for analysis cases included in our profit-motivated offense category. Again, because

of the absence of gender-specific 1975-85 data, our analysis includes only the period 1986-96.

It is shown in Figure VIII-16. As is evident from the most cursory examination of the figure,

H11 must be rejected: the female trend is significantly higher than the male trend. Although it

The base for the rates calculated in these analyses of probationary officers is

100 probationers rather than the base 1,000 we used for other analyses. This is so

because there often were far fewer than 1,000 male or female probationers active in the

department.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-16: Rates of Separation for Profit-Motivated Offenses per 1000 Officers

by Gender, 1986-96

0.5

1.5

2.5

1992

199

1996

Male Rate per 1000 Officers Female Rate per 1000 Officers

Fyfe and Kane --



185

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



186

is not shown on the figure, we derived mean annual separation rates per 1000 officers of

0.55 and 0.95 for males and females, respectively.

This finding and these data generally require some further discussion. First, the

offenses they include run a broad gamut: Figure VIII-17 shows the ten most frequent primary

charges against officers separated for profit-motivated misconduct. It describes only the most

severe charges in about three-quarters of the cases, and even these include only a small

percentage of the offenses included in the data. Second, the involvement of females in these

offenses is more frequently related to narcotics and to misconduct that, while job-related,

frequently occurs off-duty (e.g., various frauds and other larcenies) than is true of male

officers. Third, as Figure VIII-18 indicates, the rates we have derived conceal great

differences between the numbers of male and female separations for profit-motivated

misconduct: even though the female rate is higher than the male rate, the annual number of

males separated for profit-motivated misconduct typically far exceeds the number of females

separated annually. Fourth, the rates shown here reflect the downside of a pattern that peaked

in 1986, when separations for profit-motivated misconduct were at their highest level during

the 22 years studied. It was at this same time that narcotics-related separations also reached

their highest point. Both these phenomena probably are related to the fact that crack cocaine

first made its appearance in New York in 1985, and illustrate the manner in which changes in

opportunities for corruption affect its nature and frequency. It was during this period that the

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



187

ABuddie Boys@ scandal B the first involving crack cocaine related police corruption

(McAlary,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-17: Ten Most Frequent Primary Charges against Officers Separated for Profit

Motivated Misconduct, 1975-96

12 12

Fyfe and Kane --



188

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-18: Separations for Profit-Motivated Offense by Gender, 1975-96

Male Profit-Motivated Female Profit-Motivated

Fyfe and Kane --



189

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



190

1989) B came to light. Finally, 1985 was the year in which the NYPD first began its extensive

drug testing programs. This is important because, we observed during our data collection,

some of the profit-motivated misconduct included among the cases we studied came to light

after officers failed drug tests. In some instances, these were cases in which officers suspected

of profit-motivated misconduct involving drugs were subjected to drug tests as a means of

testing investigators= suspicions, and of leveraging officers= cooperation in subsequent

investigations. In other cases, failures in random or for cause drug tests led to investigations

that disclosed that officers were involved in profit-motivated misconduct.

Gender and Abuse. The images of the physically violent and intimidating male cop

and his kinder and gentle female colleague have been movie and television staples since 1971,

when Dirty Harry Callahan was appalled by his department=s attempt to slow him down by

assigning him to work with a more reasonable female partner.

This popular image has been

reinforced by spectacular cases of police brutality. Rodney King was beaten by an all-male

group of officers, and the only officer who contradicted the specious accounts of the 20 or

more officers who witnessed this event was the only female present. No female officers were

involved in the sodomy of Abner Louima, and other controversial use of force by law

Clint Eastwood fans will recall that, in mixing their message, the makers of

Dirty Harry killed off Callahan=s partner despite his hyper-aggressive policing style. A

generation later, in much the same way, NYPD Blue=s aggressive Detective Andy

Sipowicz suffered the violent death of his prosecutor wife.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



191

enforcement officers (e.g., the shooting of Amadou Diallo; the bombing of the MOVE

residence in Philadelphia) have involved no apparent participation by female officers.

To examine this issue, we hypothesized that:

H12: The rate of involuntary separations for brutality and other non-profit abuses is

greater among male officers than among female officers.

Figure VIII-19 shows the results of our first test of this hypothesis. Contrary to what

might expect, the rate per 1,000 officers at which female officers have been separated from the

NYPD on charges that included on-duty abuse is nearly twice as high as the male rate (.026

per 1,000 vs. .014 per 1,000). During the 11 years we studied, 44 men (4 per year) and 12

women (1.1 per year) were separated from the NYPD on charges that included on-duty

abuse.

These data suggest that women and men are not as different in this regard as popular

images would have it.

This is not an entirely satisfactory analysis, however, because it includes cases in

which lesser forms of abuse (ethnic slurs; discourtesy) were secondary charges in cases in

which the major offenses included profit-motivated misconduct and other serious misbehavior.

An officer who is dismissed because he or she stole cash or drugs from an arrestee whom he

or she also slapped engages in abuse not as a primary offense, but as a means of

accomplishing his or her entrepreneurial aims. In such cases, whatever on-duty abuse occurs

is merely secondary to officers= purposes and intent.

Over the entire 22 years 1975-96, 100 men and 19 women were separated in

matters that included such charges.

We adopted this posture only in cases involving on-duty abuse. In all our other

analyses (e.g., those involving drug-related or profit-motivated offenses), we included for

analysis both cases in which such offenses were the primary charge and those in which

another offense was the primary charge. We did this because we, and the NYPD, have

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



192

Hence, we took a second look at our data, including for analysis only offenses in which

the primary, or only, charge against separated officers involved nine varieties of serious on-

duty abuse. These included criminal assault; excessive force; race and ethnic slurs; race,

seen in the data situations in which on-duty abuse was a means of accomplishing other

illicit goals (e.g., profit, or drug sales or possession) rather than an end in itself. Other

offenses, however, are usually ends in themselves, rather than means to another end.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-19: Rates of Separations on Charges Including On-Duty Abuse per 1,000 Officers by

Gender, 1986-96

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

198

1990

1991

1992

1994

Male Rate Female Rate

Fyfe and Kane --



193

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



194

ethnic, and gender discrimination; improper arrest; improper stop and frisk or search;

improper summons; and verbal altercation. The data we found were so clear-cut that we did

not even cut them at 1986 to account for the absence of gender data in earlier years: we found

33 such cases during 1975-996, all of which involved male officers.

Thus, we accept the

hypothesis: Women may have been more likely than men to be separated on charges that

included some sort of abuse B but the only officers separated from the NYPD when the charge

involved on-duty abuse were men.

Gender and Drugs. We posited that:

H13: The rate of involuntary separations for drug test failures and refusals is

greater among female officers than among male officers.

The data illustrated in Figure VIII-20 confirm H13. During 1986-96, 260 men and 72

women were separated from the NYPD for failing or refusing to take NYPD drug tests.

These numbers translate to mean annual rates per 1,000 officers of 0.93 for males and 1.71 for

females. As the figure shows, this disparity has been constant (female rate exceeds male rate

in 9 of 11 years), and has followed a slight downward trend.

Gender and Non-Line of Duty Crime. We posited that:

H14: The rate of involuntary separations for non-line of duty criminal conduct (e.g.,

off-duty thefts and fraud; drug crimes) is greater among male officers than among

female officers.

As suggested earlier, it is difficult to distinguish criminal offenses that have occurred

in the line of duty from those that did not. Hence, we limited our analysis to two clearly

definable classifications of off-duty criminal offenses: crimes of violence, and criminal

violations of such public order statutes as disorderly conduct, public intoxication, and driving

Eleven were misdemeanor assaults; seven were felony assaults; 12 were

excessive force; two were discriminatory behavior; and one was a verbal altercation.

Total separations for drug test failures or refusals for 1975-96 are 376 males

and 90 females.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-20: Rate of Separations for Drug Test Failures or Refusals per 1,000

Officers by Gender, 1986-96

0.5

1.5

2.5

3.5

1988

Male Drug Failure/Refusal Female Drug Failure/Refusal

Fyfe and Kane --



195

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



196

while intoxicated. During 1986-96, this roster included 216 males and 43 females

which,

Figure VIII-21 illustrates, produces a normed trendline that, with some consistency, is higher

than that for males (mean female rate per 1,000 officers=1.02; male rate=0.77). Thus, H14

must be rejected: females are more likely than males to separated on charges that include these

two offense classifications.

Gender and Administrative Rule-Breaking. The NYPD=s manuals include myriad

rules and regulations that bound officers= behavior, and that serve as the bases for disciplinary

charges and dismissals. To test whether there existed any associated between officers=

gender and rates of separation on charges involving violations of administrative rules not

included in our other offense categories, we hypothesized:

H15: The rate of involuntary separations for administrative rule breaking is

greater among female officers than among male officers.

Figure VIII-22 shows that our hypothesis is well-founded. During 1986-96, 359 men

and 88 women lost their jobs as New York City police officers on charges that included

administrative rule breaking.

Overall, the rates per 1,000 for males and females during this

period were 1.29 and 2.09, again trending somewhat downward.

Gender and Involuntary Separations: A Summary

Except for the results of our investigation of probationary terminations (where the

1987-1995 rates per 100 probationary officers were 0.82 and 1.54 for males and females) the

mean annual separation rates for females in every misconduct category exceeds the male

Totals for 1975-96 include 349 males and 50 females.

Frequencies for 1975-96 = 618 men and 123 women.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-21: Rates of Separation for Off-Duty Violent and Public Order Offenses per

1,000 Officers by Gender, 1986-96

0.5

1.5

2.5

198

199

1996

Male Female

Fyfe and Kane --



197

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-22: Rates of Separation for Administrative Rule Violations per 1,000

Officers by Gender, 1986-96

0.5

1.5

2.5

3.5

4.5

male female

Fyfe and Kane --



198

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



199

rate. We can postulate several reasons for these disparities, and will return to discuss them

later in this report.

Race and Involuntary Separation

There is a rich literature on the role of race in police careers. A generation ago, Cohen

and Chaiken (1972:66-69) found that black New York City officers with high IQs had above

average rates of departmental misconduct. Since this finding involved officially reported

misconduct, it is possible that these officers were victims of differential rules enforcement. It

is also possible that these officers B then, even more than now, underrepresented in relation to

the city's African-American population

B were assigned to sensitive posts (e.g., vice;

narcotics) in which opportunities to engage in misconduct were greater than average. This

explanation B race confounded by assignment

B is plausible for two reasons. Fyfe (1981)

found that black and hispanic NYPD officers were more likely than whites to use deadly

force, but that these disparities were accounted for by racially differing patterns of assignment,

rank, and residence. On average, black officers worked and lived in more dangerous, higher

crime, areas than white officers and were underrepresented in supervisory and managerial

The representation of men of color and women is far higher in the NYPD than

in the City=s other major uniformed service. Black, Hispanic, Asian, and Aother@ males

comprise less than six percent of New York City 12,000 firefighters, while women are

0.2 percent of total firefighters (New York City Department of Administrative Services,

2000).

Officers' assignments and ranks, of course, are the overwhelming facts of their

professional lives, and predict degree of exposure to virtually every police hazard. The

multivariate techniques we will employ in testing all hypotheses will permit us to

determine the extent to which the relationships described are confounded by assignment,

rank, or other variables.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



200

ranks. Consequently, they fired their weapons more than their white colleagues.

This theme

of greater exposure of black officers to the risks of policing was also reported by Konstantin

(1984) who found that, nationally, black officers were far more likely than their white

colleagues to be killed in the line of duty. This disparity, Konstantin suggested, was closely

associated with the practice of assigning black officers to such high-risk tasks as undercover

narcotics enforcement.

More recently, Fyfe and his colleagues (1998) reported that disparities in the

disciplinary experience of black officers vis-a-vis others were associated with racial variation

in rank; in the frequency of off-duty misconduct; and in black officers= apparent high rates of

involvement in misconduct mandating discipline, as versus misconduct in which supervisors

were free to choose to issue formal discipline or to take informal corrective action.

The

same study also reported racial variation in the nature of misconduct generating formal

discipline, with white officers more frequently disciplined than blacks and others for brutality

and abuse. All of this led us to generate several hypotheses.

Race and Probationary Terminations. We tested the following:

H16: The rate of involuntary separations of probationers is greater among non-

white officers than among white officers.

For purposes of this analysis, we included only the years 1987-95, the only years for

which complete baseline data identifying probationers were available. We also included only

See Geller and Karales, 1981, who reported a similar pattern in Chicago.

When officers are arrested or fail or refuse to take drug tests, for example,

formal departmental discipline is mandated. In cases of such offenses as lateness,

unexcused absence, and discourtesy to citizens, however, supervisors may issue

command disciplines or simple oral or written reprimands.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



201

white, black, and Hispanic probationary officers in our analysis, electing to exclude officers

from other racial groups on grounds of their low number: only two Aother@ officers were

separated in 1990. With these provisos, the results of our test of this hypothesis are shown in

Figure VIII-23.

The figure shows mixed results regarding our hypothesis. The rate of separations per

100 probationary officers is consistently higher for blacks than for whites (annual mean white

rate per 100 = 0.83; black rate = 2.41), but the rate for Hispanics (0.79) is lower than those of

either of the other two major racial groups. Again, what we see, therefore, is an environment

in which the experiences of whites and Hispanics have come to closely parallel each other,

while black officers remain distinctly more prone than either group to this most extreme form

of discipline.

The offenses that caused these probationers= separations do not vary significantly by

race (p=.41; analysis not shown). Hispanic probationers were separated because of off-duty

violent or public order crimes less often than white or black officers (8.3% of Hispanic

separations; 22.8% and 15.6%, respectively, for whites and blacks), but were separated for

violations of administrative regulations proportionately more often (41.7%) than the other two

groups (whites = 31.8%; blacks = 29.4%). In other aspects, however, the percentages

distributions differ little among the three groups.

Thus, we accept the hypothesis where it

relates to white and black comparisons, but reject it where Hispanics are concerned.

About a quarter of the probationers in all three racial groups were separated for

drug related violations, including drug test failure (white = 26.1%; black = 27.5%;

Hispanic = 20.8%).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-23: Probationary Separation Rates

per 100 Probationary Officers by Race, 1987-95

1987

las

cla

class

1991

1992

las

ass

White rate Black rate Hispanic rate

Fyfe and Kane --



202

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



203

Race and Corruption. We hypothesized that:

H17: The rate of involuntary separations for corruption is greater among non-white

officers than among white officers.

As we have noted earlier, the NYPD did not maintain comprehensive employee race

and ethnicity data until 1986. Thus, our tests of this hypothesis included two components.

The first is an analysis of the race distributions of officers separated for profit-motivated

offenses as compared to the distributions of our sample of control officers. The second is a

more detailed analysis of data for the years 1986-96. Figure VIII-24 presents our first analysis

of H17, and suggests that our hypothesis is well-grounded insofar as it involves black officers.

The figure=s first column discloses that 10.6 percent of our1,542 randomly selected control

officers were black. By contrast, the second column indicates that 20.9 percent (14) of the 67

officers who were separated for bribe receiving were black, and the third column shows that

26.9 percent (104) of the 386 officers separated for all varieties of profit-motivated

misconduct were black. Conversely, whites are consistently underrepresented among the

offenders included in the figure. White officers comprised 78.9 percent of the control group;

73.1 percent (49) of bribery offenders and 61.4 percent (237) of those separated for all profit-

motivated offenses.

The figure is more ambiguous where Hispanic officers and those from Aother@ race

groupings are concerned. Hispanic officers constituted one in ten of our control officers

(9.6%) and officers separated for all profit-motivated offenses (11.1%), but were only one in

16 (6.0%) of those separated on bribery charges. The numbers for Aothers@ are too small for

meaningful analysis: they constituted 0.9 percent of the control group; accounted for two

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-24: Race of Control Officers and Officers

Separated

for Profit-Motivated Offenses, 1975-96

78.9%

73.1%

61.4%

10.6%

20.9%

26.9%

9.6%

6.0%

11.1%

Control (n=1542) Bribery (n=67) All Profit (n=386)

White Black His

anic Other

Fyfe and Kane --



204

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



205

(0.5%) of the officers separated for all profit-motivated offenses, and were not represented at

all among bribery offenders.

Although not directly relevant to H17, the data on bribery and other profit-motivated

offenses show interesting temporal patterns. The 1970s Knapp Commission scandals involved

profit-motivated corruption, most of which involved acts of bribery between corrupt officers

operators of such illicit activities as gambling and those who sought immunity from New York

City construction codes and colonial-era blue laws limiting hours and conditions of business

on Sundays (Knapp,1972). The opportunity for engaging in such offenses was greatly limited

by NYPD regulations that, in effect, prohibited officers from enforcing these laws and

regulations absent formal complaints from citizens. Further attempts at deterrence involved

the creation of a field associate program that employed officers who, while working alongside

others, reported on misconduct among their peers, as well as by vigorous programs that

conducted stings, rewarded officers for arresting bribers, and vigorously prosecuted bribers

(Murphy, 1977; Murphy and Plate,1977; Sherman, 1978).

Over the period we studied, however, there have emerged new forms of profit-

motivated misconduct. The investigations that have examined them suggest that they are not

nearly as wide-spread as the organized bribe takers uncovered in the 1970s (Kelly, 1992; Mc

Alary, 1989,1997; Mollen, et al., 1994),

but more recent profit-motivated offenders appear

to have been involved in activities that more violent and virulent than was true of their

Many of the officers separated from the NYPD as a result of the early-1970s

Knapp Commission/Serpico scandals had left the department by the time the period we

studied began in 1975.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



206

predecessors. These include such offenses as robbery; drug dealing and theft; theft of cash;

and various types of fraud.

These changes are reflected in Figure VIII-25. The figure shows that the frequency of

separation on charges that include bribery has diminished significantly over the years. In the

first ten years we studied (1975-84) 50 officers were separated on charges that included

bribery; over the last ten years (1987-1996), only eight officers lost their jobs on bribe-related

charges. Thus, it would appear that the department=s attempt to reduce this variety of usually

consensual misconduct have borne fruit. The trend and spikes in separations for all varieties

of profit-motivated misconduct (which include the bribery offenses tracked on the figure=s

dotted line), however, suggest the difficulty of controlling other types of profit-motivated

misconduct. Still, even with peaks following the 1980s ABuddy Boys@experience and the

1990s Mollen Commission scandal, the numbers are quite small. Only once during the 22

years we studied (1986, when 40 officers were separated) did the number of officers separated

for profit motivated misconduct represent more than one in 1,000 NYPD officers and, most

often, the rate has been far lower than even that.

To look at H17 from another perspective, we obtained NYPD=s race and rank data for

the years 1990-96. The results are shown in Figure VIII-26 and, even though they confirm the

hypothesis, show a familiar pattern. Over these last seven years, the white and Hispanic

trends have been closely parallel, resulting in similar annual rates per 1,000 officers (white =

0.26; Hispanic = 0.36), a difference that is largely accounted for by the Mollen Commission

scandals: eight of the 11 Hispanic officers represented on this figure were separated during

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-25: Separations for Bribery and Profit-Motivated Offenses, 1975-96

Bribery All Profit-Motivated Offenses

Fyfe and Kane --



207

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-26: Separation Rates per 1,000 Officers for Profit-Motivated Misconduct

by Race, 1986-96

0.5

1.5

2.5

3.5

4.5

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

White Black Hispanic

Fyfe and Kane --



208

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



209

1995 and 1996, following that expose. The black rate, however, is dramatically higher (at

1.33 per 1,000).

100

Thus, it is fair to say, our analyses confirm H17 where both black and

Hispanic officers are concerned. At the same time, our analyses show that the real racial

demarcation in officers= experiences with this most severe form of discipline lies between

blacks and others, rather than between whites and others.

Based on prior research showing that racial variations in exposure to the risks of

policing, are associated with differential patterns of rank and assignment (Alex, 1969; Fyfe,

1981a, 1981b; Fyfe, et al., 1998; Konstantin, 1984), we also generated this hypothesis:

H18: The discrepancy between white and non-white officers' rates of involuntary

separation for corruption is accounted for by differential patterns of assignment and

rank.

This hypothesis is based on two assumptions

- First, that the overall separation rates of non-white officers would vary from

that of whites largely because of differential representation in the supervisory

ranks. In general, policing is an occupation in which those at the entry level B

100

No officers of Aother@ races were separated for profit-motivated misconduct

during these years.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



210

police officers and detectives

101

B deal most directly with the public and,

therefore, have greater opportunities than supervisors and commanders, who

typically are not arresting or summonsing officers, to engage in profit-

motivated misconduct.

- Second, that even at the police officer and detective level, the opportunity to

engage in profit-motivated misconduct is not evenly distributed across all

assignments. Traffic cops rarely are implicated in profit-motivated misconduct

related to drugs because their work does not expose them to the same

opportunities or specialized knowledge that are part of the work of officers

assigned to narcotics enforcement. Patrol officers who work in outlying

middle-class communities are not exposed to street-level gambling operations

and their temptations to corruption in the same manner as officers who patrol

center city areas.

The underrepresentation of non-whites in police supervisory and command ranks is a

longterm reality throughout policing (President=s Commission,1967; Fyfe, 1981b), and

remains a characteristic of the NYPD. Figure VIII-27 traces the percentage of NYPD officers

within each of our four major racial classifications over the years 1990-96. It shows that

about one-fifth of all white officers held supervisory and command ranks (sergeant or above)

101

In the NYPD, the detective classification rank is a discretionary designation

awarded to non-supervisory personnel at the police officer rank. Historically, detectives

have been criminal investigators but, over the last generation, many detective specialist

designations have been created for personnel in elite administrative or field units such as

the Police Academy, the Internal Affairs Bureau, and the Emergency Service Unit.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-27: Percentage of Officers in Supervisory Ranks by Race, 1990-96

0.05

0.1

0.15

0.2

0.25

1990 1991 1992 1993 1994 1995 1996

White Black His

anic Othe

Fyfe and Kane --



211

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



212

during these years (low = 19.0% in 1994; high = 21.3% in 1996), typically about twice as high

as the percentages for black, Hispanic, and other race classifications (the percentage of blacks

in supervisory ranks started at 11.8% in 1990 and declined to7.7% in 1996; Hispanics

decreased slightly from 8.7% in 1990 to 7.0% in 1996; while Aothers@ increased from 8.2% in

1990 to13.2% in 1996). To examine the effect upon separation rates of this differential

distribution, we calculated rates for police officers/detectives and supervisory/command

personnel. If rank made a difference in the anticipated direction, we would expect that

officers of different races in the same ranks would similar separation rates.

Figure VIII-28 shows that this is not what we found. Instead, although the figure is

affected by the small absolute frequencies indicated, it shows that the separation rates for both

black police officers/detectives and supervisory personnel (1.54 and 2.11 per 1,000 annually)

are far higher than those for either whites or Hispanics (0.32 and 0.64 for white and Hispanic

police officers and detectives; 0.15 for white supervisors; no profit-motivated separations for

Hispanic supervisors). Thus, we must reject the section of this hypothesis which anticipated

that differential rank distributions were to account for some proportion of the variance in

separation rates among the races: the differences hold even when one derives rank specific

rates.

Our examination of the role that racially varying patterns of assignment might play in

determining separation rates among the races was inconclusive because, for the years we

studied, there were available no race-specific data on officers= assignments. Hence, all that

we could do to examine this issue is presented in Figure VIII-29. The data at the top of the

figure

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

ure VIII-28: Mean Annual Separation Rates per 1,000 Officers for Profit-Motivated

Misconduct by Rank and Race, 1990-96

0.32

1.54

0.64

0.15

2.11

White PO/Dets

(n=39)

Black PO/Dets

(n=36)

Hispanic PO/Dets

(n=18)

White Sgts.+ (n=4) Black Sgts.+ (n=5)

Fyfe and Kane --



213

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-29: Commands of Officers Separated for Profit-Motivated

Misconduct and Matching Control Officers by Race, 1975-96

1111

Insp/DI

Precincts

Captain

Precincts

PA or Training

Units

Proactive

Investigation

Det

Bur/Warrants

OCCB not Narc SpecialPatrol Staff

White Black Hispanic/Other

Fyfe and Kane --



214

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



215

present the command types to which officers of different races

102

were assigned when they

were separated. This portion of the figure shows that the modal command types for all three

racial groups of separated officers were patrol precincts. Beyond that, this band of the figure

illustrates roughly similar racial distributions across all command types: in every command

type, about 60 percent of the officers separated for profit-motivated misconduct were white.

The general pattern within the remaining 40 percent of separated is that blacks outnumber

Hispanics and others about two to one across the types.

Absent baseline race by assignment data, we elected to use as our standard of

comparison the commands of the matching control officers effective at the time the separated

officers lost their jobs. These are valuable because, as a random sample, they presumably

include a representative cross-section of the assignments held by the corrupt officers= peers.

They are presented as the figure=s lower band, and illustrate quite a different pattern. All

other things being equal, one would expect to find that study and control officers of specific

racial groups would be distributed across command types in numbers approximately

proportional to their overall representation in the study and control groups. If that is so, it

would appear that both white and, to an even greater degree, black officers assigned to patrol

precincts were overrepresented among separated officers. Among white study officers, 74

(34.7%) worked in Inspector/Deputy Inspector precincts, as compared to only 63 (22.4%) of

the white controls. Thirty-six (37.3%) of the black study officers and only 11 (29.0%) of the

black controls

102

The combined Hispanic/Other category includes 40 Hispanic officers and two

others.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



216

worked in Inspector/Deputy Inspector precincts; 27 separated blacks (28.7% of separated

blacks ) and only four black controls (10.5% of black controls) worked in Captain precincts.

Thus, while our analysis is interesting and suggestive, it is not directly responsive to

the question of whether racial variation in separation rates is attributable to variations in risk

as measured by assignment type. We will, however, return to this issue later in this report,

when we present our multivariate analyses.

Race and Abuse. H19, our hypothesis on this subject, specified that:

The rate of involuntary separations for brutality and other non-profit abuses is

greater among white officers than among non-white officers.

In New York City, as elsewhere, the most spectacular and widely publicized incidents

of controversial police use of force have involved white police officers who, almost

invariably, have injured or killed people of color. Rodney King was beaten by a group of

white Los Angeles police officers; Luis Baez died after an encounter with a white NYPD

officer; Amadou Diallo was shot to death by four white NYPD officers; Abner Louima was

brutalized by one or more white NYPD officers. Thus, this hypothesis seemed a safe bet B but

we were surprised to find that the data did not provide results that were as clearcut as one

might suspect.

We tested the hypothesis by examining data for the years 1986-96, the period for

which baseline officer race data are available. Our first analysis included involuntary

separations on charges that included any type of abuse by on-duty officers. Tested in this

fashion, the hypothesis does not hold for those years. Further, although we do not have

baseline officer race data for the first 11 years of our study (1975-85), it appears that, when

analyzed in this fashion, the hypothesis would not hold for that earlier period, as well.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



217

Figure VIII-30 demonstrates why this is so. The figures upon which it is based include

63 separations for charges including on-duty abuse during the first 11 years of our study.

Twenty-two (34.9%) of these officers were white; 35 (55.6%) were black; and six (9.5%)

were Hispanic (no Asians or others were separated for abuse during this period). This

representation of black officers is far higher than their percentage in the department during

these years and, in all probability, the percentage of Hispanic officers is somewhat higher than

their percentage among all officers. On July 1, 1974, Fyfe (1978:164) reported, white officers

constituted 88.7 percent of the NYPD=s sworn complement (n=27,314), while blacks

represented 8.0 percent (2,448) and Hispanics were 3.3 percent (1,014). By 1986, these

percentages had changed to 79.2 percent for whites (n=20,816), 10.7 percent (n=3,011) for

blacks, and 9.5 percent (2,505) for Hispanics.

103

In the intervening years, the only dramatic

change in the department=s personnel composition was the June 30,1975 lay-off of 3,000

officers. Because of last hired-first fired civil service regulations, this layoff actually reduced

the representation of non-white officers who, generally, were junior to white officers. Thus, it

is fair to say, black officers were well overrepresented among officers separated on charges

hat included on-duty abuse during 1975-85, and it is likely that Hispanic officers also were

overrepresented, albeit not to the same degree.

103

In 1974, the NYPD=s 21 Asian and other officers constituted 0.07 percent of

NYPD=s sworn officers. By 1986, their numbers had grown to 154 (0.6% of total), and,

by 1996, the department employed 478 Asians and other (1.3 % of total).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-30: Separations on Charges Including On-Duty Abuse by Race, 1975-96

444

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

White Black His

anic Othe

Fyfe and Kane --



218

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



219

We were able to more directly test the hypothesis where the years 1986-96 are

concerned. During this period, the data in Figure VIII-30 include 25 separated white officers,

20 separated black officers, nine separated Hispanic officers, and one other officer. When we

used these frequencies to calculate annual separation rates per 1,000 officers for charges

including on-duty abuse offenses, we derived rates of 0.20 for whites, 1.47 for blacks, and

0.38 for Hispanics.

104

Thus, these data do not support H19 and, qualifying our conclusions by

reference to the low numbers upon which they are based, we once again find a difference

between the experience of black officers and others.

As was the case with our analysis of gender and on-duty abuse, however, we were not

satisfied that this analysis truly answered the question at hand because it included all cases

involving on-duty abuse, regardless of whether such conduct was tangential to other, more

serious misbehavior. Consequently, we conducted an additional analysis that included only

officers separated during 1986-96 when the primary, or only, charge against them involved

on-duty criminal assault (n=10); excessive force (5); and racial, ethnic, or gender

discrimination (1).

105

Eight of these involved white officers, four involved black officers, and

four involved Hispanic officers. As Figure VIII-31 indicates, these figures translated to mean

annual rates per 1,000 officers of 0.03 for white officers; 0.11 for black officers; and 0.10 for

Hispanic officers. These rates are so small that a case or two one way or the other would

change them significantly, but they certainly do not support the hypothesis.

104

The mean annual rate for Asians and others is 0.35,but is based on the

separation of a single officer during these 11 years.

105

There were 17 such cases during the 11 years 1975-85 (white = 10; black = 4;

Hispanic = 3).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



220

Race and Non-Line of Duty Crime. Our hypothesis specified that:

H20: The rate of involuntary separations for non-line of duty criminal conduct (e.g.,

off-duty thefts and fraud; drug crimes) is greater among non-white officers than

among white officers.

Figure VIII-32 traces the frequency of separations for non-line of duty offenses between

1975 and 1996.

106

During the years 1975-85, the period for which we lack race-specific data

on all personnel, there occurred 155 separations for these offenses. Whites accounted for 53.5

percent of these (83); blacks for 37.4 percent (58), and Hispanics for 9.0 percent (14).

107

least where white and black officers are concerned, these figures again vary from the apparent

representation of each of these groups in the department as whole during those 11 years, and

suggest that our hypothesis is accurate as it relates to that period.

H20 clearly holds for the 1986-96 period. Whites accounted for 50.8 percent (124) of

the 244 separations during this period, resulting in a mean annual rate of separations per 1,000

officers of 0.88, as shown in Figure VIII-33. The black rate (3.92 per1,000; n=89) is more

than four times as high, and the Hispanic rate (0.96; n=26) is slightly higher than the white

rate, but is about what one would expect based on Hispanic representation in the NYPD

during those years: Hispanics accounted for 10.7 percent of the non-line of duty separations

and for 10.9 percent of average annual department personnel during these years. Officers

from “other” racial groups, 0.9 percent of the NYPD as a whole, accounted for 2.0 percent of

106

As in our earlier analyses of gender and non-line of duty conduct, we

calculated our dependent variable by including off-duty crimes of violence and criminal

violations of public order statutes such as disorderly conduct, public intoxication, and

driving while intoxicated into a single category.

107

No Aother@officers were separated for non-line of duty conduct during 1975-

85.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-31: Rates of Separation per 1,000 Officers

in Which Primary Charge was On-Duty Abuse,1986-96, by Race

0.03

0.11

0.10

White (n=18) Black (8) Hispanic (7)

Fyfe and Kane --



221

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-32: Separations for Non-Line of Duty Offenses by Race, 1975-96

666

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

White Black His

anic Other

Fyfe and Kane --



222

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-33: Separation Rates per 1,000 Officers

for Non-Line of Duty Offenses by Race, 1986-96

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

White Black Hispanic

Fyfe and Kane --



223

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



224

the separations, resulting in a mean annual rate of 1.73 per 1,000. This calculation, however,

is based on only five separations over an 11 year period and, therefore, is not the stuff of grand

generalizations.

Race and Drugs. H21 specified that:

The rate of involuntary separations for drug test failures and refusals is

greater among non-white officers than among white officers.

Figure VIII-34 is based on data that confirm the hypothesis. During 1986-96, 42.9

percent (96) of the 229 separations for drug test refusals or failures involved white officers;

41.1 percent (92) involved black officers; and 16.1 percent (36) involved Hispanic officers.

108

These figures result in mean annual rates of separations of 0.41, 2.45, and 0.86, respectively,

for whites, blacks, and Hispanics. Again, the pattern is familiar: the rates for all three groups

have declined, in this case from the beginning of systematic drug testing programs in the mid-

1980s; the rate for blacks is far higher than those of the either whites or Hispanics; and the

trends for whites and Hispanics became very similar during the latter years shown in the

figure. During 1991-96, the rates were 0.26 for whites; 1.39 for blacks; 0.19 for Hispanics.

108

We did not include a figure tracing drug test failures back to 1975 because of

their infrequency. During 1975-84, prior to the advent of the NYPD=s systematic drug-

testing program, 32 officers (3.2 annually) were dismissed for failing or refusing to

undergo Afor cause@ drug tests or for medical diagnoses that they had ingested drugs;

another 22 were terminated in 1985, the first year of systematic testing. Of the 54 officers

separated during 1975-1985, 28 were white, 19 were black, and one was@other.@ No

Aother@officers were separated for these reasons during 1986-96.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-34: Separation Rates per 1,000 Officers

for Drug Test Failures or Refusals, 1986-96

0.5

1.5

2.5

3.5

4.5

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

White Black His

anic

Fyfe and Kane --



225

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



226

Race and Administrative Rule-Breaking. In H22, we predicted that:

The rate of involuntary separations for administrative rule breaking is greater among

non-white officers than among white officers. Our analysis of this question requires an

explanation. Nearly half of the separations in our study (741) could be included in the

Aadministrative rule-breaking@ category because such charges routinely are added to the

catalog of offenses with which officers whose major sins have been drug or criminal offenses

are prosecuted. It is impossible to commit a violent felony without simultaneously engaging

in conduct unbecoming an officer, for example. In such cases, the administrative violations do

not accurately describe the heart of the issue. Consequently, we included in our examination

of analysis of H22 just the 242 cases in which the charges included only violations of the

NYPD=s administrative rules.

The results are presented in Figures VIII-35 and VIII-36. The former shows the

absolute frequencies of separations for rules violations during 1975-96, and shows that these

separations have not been constant. Instead, they hit high points in the mid-1980s (peaking at

32 in 1986). Keeping in mind that blacks likely represented somewhat less than ten percent of

the department during the years for which we lack base race data, Figure VIII-35 they were

overrepresented during this early period: 22, or 20.6 percent of the separations, involved black

officers. Hispanics accounted for 7.5 percent (8) of the separations which, again, is about

what one would expect given their likely representation in the NYPD.

Figure VIII-36 shows the trends for the years 1986-96, and is derived from data that

confirm the hypothesis. During these years, the separation rates per 1,000 officers for whites,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-35: Separations for Administrative Rules Violations by Race, 1975-96

White Black Hispanic Other

Fyfe and Kane --



227

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



228

blacks, and Hispanics were 0.32, 1.12, and 0.38, respectively. Once again, therefore, we see a

familiar pattern: although absolute frequencies are small,

109

the experience of black officers is

markedly different from that of their white and Hispanic colleagues.

By way of explaining this disparity among the races, we offered H23:

The discrepancy between white and non-white officers' rates of

involuntary separation for administrative rule-breaking is

accounted for by differential patterns of assignment and rank.

Analysis verified that vulnerability to separation for administrative rule-breaking is

greatest at the department=s lowest levels. Only 3.4 percent (8) of the personnel separated for

administrative violations during 1975-96 had been promoted to third grade detective (4) or

sergeant (4), and none had attained any ranks higher than those. By contrast, 14.7 percent (32)

of their matched control officers had made detective or supervisory ranks, with several going

beyond the detective 3 designation and sergeant rank.

110

During 1990-96, the period for which

we have baseline rank and race data, the separated personnel were probationary police officers

(20), police officers (22), or detectives (1); none was a supervisor or commander. The matched

controls were slightly different, including 24 probationers, 16 police officers, one third grade

detective, two second grades, and a sergeant.

109

Over the 11 years studied, white frequency=75; black=42; Hispanic=16. We

also calculated an AAsian/other@rate of 0.69 per 1,000, which is based on only two cases.

110

In the NYPD, Adetective third grade@is the entry level detective designation,

carrying with it the coveted Agold shield@and a salary increase of about 15 percent.

Detective second grade and detective third grade are higher paying, but still non-

supervisory, designations awarded for excellent performance. None of the four separated

detectives had gone beyond third grade, but five of the 13 detectives in the matched

control group had attained second or first grade. Four of the 19 control officers were

lieutenants, but all of the four separated supervisors were sergeants.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-36: Separation Rates per 1,000 Officers for Administrative Rules

Violations 1986-96, by Race

0.5

1.5

2.5

3.5

198

199

1994

199

White Black Hispanic

Fyfe and Kane --



229

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



230

When we constructed separation rates per 1,000, therefore, we included as our base only

personnel at the police officer and detective ranks, on grounds that the supervisory rates for all

racial groups were zeroes. These are shown in Figure VIII-37 in which, were our hypothesis

well-based, one would expect to find similar rates across the racial groups. This is not the

case: instead, we see that the mean annual rates for whites, blacks, and Hispanics over the

seven years are 0.19, 0.43, and 0.28, respectively.

111

This again repeats the familiar pattern

we have seen throughout these analyses of race and separation. It also compels us to accept

the null hypothesis.

Our attempt to test the portion of the hypothesis that racial variations in rates of

separation for administrative violations were associated with assignment was complicated by the

infrequency of such cases during the years for which we have baseline data (43 over the seven

years 1990-96). Thus, we could address it only inferentially. For Figure VIII-38, we collapsed

our command typology into four categories (Patrol; Training; Investigative; Staff) and compared

the assignments of officers separated for administrative violations by race over the entire study

period 1975-96 with those of their matched control officers.

The figure shows that, across the races, most of the officers in both the study and control

groups were assigned to patrol or training when they study officers were terminated, and that the

percentages of officers assigned to investigative or staff jobs for every race is higher among

control officers than among study officers. The figure shows that officers who work in

specialized tasks appear less likely than officers who are still in patrol or training to be

111

Two Asians/Others were separated during this period.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VII-37: Mean Annual Separation Rates

per 1,000 Police Officers/Detectives

for Administrative Rules Violations by Race, 1990-96

0.19

0.43

0.28

White Black Hispanic

Fyfe and Kane --



231

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



232

have been fired. Its raw frequencies, shown at the bottom of each set of columns, also indicate

that blacks are overrepresented among study officers (57 study officers and only 24 control

officers were black);and that both whites and Hispanics were more likely to be found among

control officers than among those separated from the NYPD. Still, it cannot be said to resolve

our hypothesis.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

10%

20%

30%

40%

50%

60%

70%

80%

Study White

(n=138)

Control White

(n=167)

Study Black

(n=57)

Control Black

(n=24)

Study Hispanic

(n=19)

Control

Hispanic (n=24)

Patrol Trainin

Investi

ative Staff

Figure VIII-38: Commands of Officers Separated for Administrative Rule-Breaking

and Matched Control Officers by Race, 1975-96

Fyfe and Kane --



233

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



234

Race and Involuntary Separations: A Summary.

This section of this study is likely to be its most controversial. The data we analyzed

show that black officers are involuntarily separated from the NYPD at rates far

disproportionate to other groups, for virtually all offense types. It does not appear that these

disparities are closely associated with rank or assignment, two variables that have explained

much racial variation in prior studies of other varieties of police conduct. One explanation of

this disparity is that the system discriminates against black officers. Certainly, our statistical

analyses may hide instances of discrimination, but three considerations suggest that any

discrimination is isolated, rather than systemic.

First, discrimination is an abuse of discretion: in most of the cases we studied, most

discretion is exercised at the policy level, rather than at the level of individual cases. Our data

included 466 separated African-American officers. Only 107 of these officers were separated

for matters in which the primary charge was a purely administrative offense. In the remainder

B more than three-quarters of the cases B the primary charge apparently involved criminal

behavior or drug-related offenses. In 168 cases, the primary charge against black officers was

a drug offense (possession, sale; failure or refusal to take a drug test); in 60, it was an off-duty

violent crime; in 55, it was a profit-motivated crime; in 18, it was perjury, conspiracy, or some

other offense designed to obstruct justice; in 16, it was an off-duty public order crime. Once

legislators and NYPD leaders decide that these are firing offenses, those who administer the

NYPD=s internal justice system have few options but to separate officers found to have

committed them.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



235

Second, there is a considerable degree of due process involved in involuntary

separations of NYPD officers, especially after they have completed their probationary

periods. They are entitled to notice of formal charges; to representation; to full, recorded,

public trials involving confrontation and cross-examination of witnesses and presentation of

their own witnesses and evidence. Those who lose these trials are entitled by statute to appeal

in the civil courts. Thus, there are enough checks on the system to make it unlikely that it

involves systematic discrimination.

Third, the contrast between the disparity in the study group of African-American

officers and the recent common experience of white, Hispanic, and Asian officers suggests

that, to the extent racial discrimination exits, it must be directed only at black officers, while

granting equal treatment to Hispanics and Asians. This also is unlikely.

Thus, these findings concerning African-American officers are troubling not because

they suggest systemic discriminatory behavior by the NYPD, but because they suggest

differential patterns of behavior among the races.

At the same time, we are heartened by other findings regarding both officers= race

and gender. Over the 1986-96 period for which we have data on officers= race and gender,

the NYPD became a more diverse agency. In 1986, 26.3 percent of all the uniformed

members of the NYPD were other than white men, and the department involuntarily separated

167 officers, for a rate of 6.4 per 1,000 officers. By 1996, the percentage of persons other than

white men in the department had increased to 38.4 and the rate of separations had dropped to

2.7 (n=83). Figure VIII-39 presents this relationship graphically, and shows that it is

extremely robust.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



236

Most specifically, the figure shows that half of the variation in the overall rate of

involuntary separation rate during 1986-96 is accounted for by the percentage of white male

officers in the NYPD (r=.71; r

=.50). In other words, as diversity has increased, career

ending police misconduct has decreased. This obviously is a finding that requires further

exploration and that may be confounded by other variables B but it is too strong to be ignored,

and is the most convincing evidence we have yet seen of apparent positive effects of diversity

in police ranks.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

r=.71

=.50

p=.01

Figure VIII-39: Separation Rate per 10,000 Officers and

Percentage White Male Officers, 1986-96

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

Rate per 10,000 % White Males

Fyfe and Kane --



237

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



238

Age and Involuntary Separation

In 1994, the NYPD increased its minimum age for entry into the department from 20 to

22. The NYPD did this on grounds that the candidates it was excluding were more likely

than older recruits to end their careers under less than honorable circumstances (O'Sullivan,

1994; Mollen Commission, 1994; Mladinich, 1995). To determine whether this policy change

was consistent with the experience reflected in our data, we hypothesized that:

H24: The rate of involuntary separations of probationers is greater among officers

who were less than 22 years old when appointed than among officers who were 22 or

more years old when appointed.

H25: The rate of involuntary separations for corruption is greater among officers

who were less than 22 years old when appointed is greater than among officers who

were 22 or more years old when appointed.

H26: The rate of involuntary separations for brutality and other non-profit abuses is

greater among officers who were less than 22 years old when appointed than among

officers who were 22 or more years old when appointed.

H27: The rate of involuntary separations for non-line of duty criminal conduct (e.g.,

off-duty thefts and fraud; drug crimes) is greater among officers who were less than

22 years old when appointed than among officers who were 22 or more years old when

appointed.

H28: The rate of involuntary separations for drug test failures and refusals is

greater among officers who were less than 22 years old when appointed than among

officers who were 22 or more years old when appointed.

H29: The rate of involuntary separations for administrative rule breaking is greater

among officers who were less than 22 years old when appointed than among officers

who were 22 or more years old when appointed.

We tested these hypotheses by constructing chi-square tables in which we

dichotomized age at hiring into values of under 22 and 22 and over and compared the

percentages of officers subsequently separated for each of these offense types with their

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



239

matching control officers. The short answer concerning what this showed is that, although

statistical tests (here applied to a population of separated officers and a matched random

sample of comparison officers) show significance only for all offenses in the aggregate

(p=.04) , there is a consistent pattern across all but one offense type (drug test failures and

refusals; see Figure VIII-43) in which the percentage of separated officers who were 20-

21years old when hired is higher than the percentage of control officers who were 20 and

21when hired).

Figure VIII-40 shows this aggregate analysis, with age categories at 22 or higher further

disaggregated into values of 22-23, 24-25, and 26+ (which produces a chi square significance

level of .09). The figure shows that 27.7 percent of the separated officers we studied were

under 22 when hired, as compared to 24.4 percent of the control officers. This difference is

almost precisely reversed for officers hired at ages 22-23 (24.5% of separated officers; 27.7%

of control officers). The distributions of separated and control officers at ages 24 -25 and 26+

are, for all intents and purposes, virtually identical.

Similar patterns appear in Figures VIII-40 through VIII-45. Although none of these

analyses produces a significant probability level, the trend is clear and explains why, in the

aggregate, there is a statistically significant difference between those hired at under 22 years

of age and those who are 22 or more when hired. In Figure VIII-41, we see that 20-21 year

old hires represented a quarter (25.1 percent) of the officers separated for profit-motivated

offenses, and a fifth (21.5 percent) of the matched controls. Figure VIII-42 indicates that a

third (33.6%) of the officers separated for on-duty abuse, versus a quarter (25.4%) of their

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



240

matched controls were 20-21 when hired (the relationship was similar when we included for

analysis only officers against whom the primary charge was on-duty abuse).

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-40: Separated and Control Officers' Age at Appointment, 1975-96

24.5%

19.3%

28.4%

24.4%

27.7%

28.9%

27.7%

19.0%

20-21 22-23 24-25 26+

Separated (n=1536) Control (n=1539)

Fyfe and Kane --



241

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-41: Separated and Control Officers' Age at Appointment,

Profit-Motivated Offenses, 1975-96

25.1%

23.1%

22.0%

29.8%

21.5%

27.7%

22.5%

28.3%

20-21 22-23 24-25 26+

Profit-Motivated (n=386) Profit-Controls (n=382)

Fyfe and Kane --



242

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-42: Separated and Control Officers' Age at Appointment,

On-Duty Abuse, 1975-96

33.6%

24.4%

14.3%

27.7%

25.4%

27.1%

16.9%

30.5%

20-21 22-23 24-25 26+

On-Duty Abuse (n=119) On-Duty Abuse Controls (n=118)

Fyfe and Kane --



243

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



244

In Figure VIII-43, we observe that three in ten (29.3%) of officers separated for drug sales

were 20 or 21 when hired, versus two in ten (21.4%) of their controls. Figure VIII-44 shows

that a third (32.9%) of officers separated for non-line of duty offenses were 20 or 21when

hired, as versus a quarter (26.5%) of their controls. Only Figure VIII-45's analysis of officers

separated for drug test failures and refusals runs contrary to this trend (22.4% of separated

officers and 26.8% of control officers were under 22 when hired). Figure VIII-46 shows only a

very minor difference between the percentage of 20 and 21 year-old hires separated for

administrative rules violations and their matched controls (30.2% versus 27.2%).

Thus, the chi square tests we conducted suggest that people who were hired by the

NYPD at ages 20 and 21 were more likely than their older Police Academy classmates to have

ended their careers with involuntary separations. This confirms our H24. Although we did

not derive statistically significant differences in the offense-specific analyses conducted to test

H25 through H29, when taken in toto, our analyses suggest that the NYPD was wise to raise

its minimum age at entry from 20 to 22.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-43: Separated and Control Officers' Age at Appointment,

Drug Sales, 1975-96

29.3%

25.9%

13.8%

31.0%

21.4%

30.4%

17.9%

30.4%

20-21 22-23 24-25 26+

Drug Sale (n=58) Drug Controls (n=56)

Fyfe and Kane --



245

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-44: Separated and Control Officers' Age at Appointment,

Non-Line of Duty Offenses, 1975-96

32.9%

25.6%

17.1%

24.4%

26.5%

27.3%

18.4%

27.8%

20-21 22-23 24-25 26+

Non-Line of Duty Non-Line of Duty Controls

Fyfe and Kane --



246

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-45: Separated and Control Officers' Age at Appointment,

Drug Test Failures and Refusals, 1975-96

22.4%

24.9%

16.7%

35.9%

26.8%

27.2%

17.8%

28.3%

20-21 22-23 24-25 26+

Drug Test Failures/Refusals (n=281) Drug Test Controls (n=276)

Fyfe and Kane --



247

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-46: Separated and Control Officers' Age at Appointment,

Administrative Rules Violations, 1975-96

19.8%

19.0%

31.0%

27.2%

28.0%

16.7%

28.0%

30.2%

20-21 22-23 24-25 26+

Admin Rules (n=242) Admin Rule Controls (n=239)

Fyfe and Kane --



248

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



249

Prior Employment, Life History, and Involuntary Separation

We hypothesized that:

H30: Rates of involuntary separation for all types of police misconduct are higher

for the following categories of officers than for other officers:

H30a: officers who were dismissed by previous employers.

H30b: officers whose prior employers gave police investigators derogatory

information about them.

H31: Officers with military records that include no discipline are less likely than

officers without military experience or with military disciplinary histories to be

involuntarily separated.

Prior Employment.

Our tests of H30a and H30b appear in Figure VIII-47, which also

includes data on four other salient background variables B whether officers previously had

been rejected by civil service employers, whether officers had been disciplined by their

employers before they were hired by the NYPD, whether officers had been hired over the

objections of one or more of the personnel investigators assigned to their cases, and whether

officers were hired only after civil service appellate boards overturned negative

recommendations by the NYPD=s Applicant Processing Division.

The figure shows that both hypotheses are confirmed, and that the analyses of the other

four variables follow the same trend B that, however measured, unsatisfactory performance in

prior jobs is associated with career failure in the NYPD. Officers who previously had been

fired from one or more jobs (maximum = 5 previous job dismissals) were found in the study

group approximately twice as often as in the comparison group (15.3% v. 7.7%), as were

officers whose prior employers gave investigators derogatory information about them (10.0%

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-47: Separated and Control Officers' Prior Employment

and Pre-Entry Experiences, 1975-96

15.3%

10.0%

15.3% 15.3%

16.3%

15.2%

7.7%

4.7%

8.7%

9.3%

6.8%

7.3%

Previously Fired Derogatory

Information

Prior Job Discipline Civil Service

Rejections

Reject

Recommendation

Appointed after

Hearing

Study Control

Fyfe and Kane --



250

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



251

v. 4.7%);

111

officers who previously had been disciplined in other jobs (15.3% v. 8.7%);

officers who had previously been refused civil service employment (15.3% v. 9.3%);

officers whose assigned personnel investigators recommended that they not be hired as police

officers (16.3% v.6.8%); and officers who were hired after appeals boards overturned the

Applicant Processing Division=s recommendation that they not be hired (15.2% v. 7.3%).

These are all quite strong associations, producing chi square statistics significant

below the .0001 level. Expressed another way: our data include a number of officers who had

previously been fired (n=323); worked for people who gave them poor ratings (n=202) or

disciplined them (n=296); been rejected for civil service employment (n=303); recommended

for rejection by their personnel investigators (n=315); or hired only on appeal from the

NYPD=s rejection recommendation (n=307). All other things being equal, one would expect

that these groups would be equally split between our study and control groups. Instead, two-

thirds of the members of each group are study officers, indicating that officers who share any

of these experiences are more likely than their colleagues to be fired or forced to resign.

Military Experience. Our analyses concerning prior civilian employment produced

results consistent with what would expect on the basis of both theory and common experience,

but our analysis of military experience produced surprises. Contrary to common belief, we

found that officers whose pre-employment histories included military service were more likely

than non-veterans to have been involuntarily separated from the NYPD. Figure VIII-48 shows

111

The derogatory information consisted of assessments of police candidates=

character and reliability, as well as responses to inquiries about whether employers would

rehire them if given the opportunity.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-48: Separated and Control Officers' Military Experiences, 1975-96

59.2%

40.8%

32.0%

8.7%

67.5%

32.5%

27.8%

4.7%

No Military Military Military/No Discipline Military/ Discipline

Study Control

Fyfe and Kane --



252

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



253

three in five of our study officers (59.2%) and two-thirds of the comparison officers (67.5%)

entered policing without any history of military service. The converse, of course, is that 40.8

percent of the separated officers and 32.5 percent of the comparison officers had served ion

the military before becoming NYPD officers. This association is significant at the .001 level,

in the direction opposite that anticipated in our hypothesis.

Moving further to the right on the figure shows that officers who served in the military

without being disciplined are more frequently found in the study group than in the control

group (32.0% v. 27.8%; p < .001), as are officers who were disciplined in the military (8.7%

v. 4.7%; p < .001). All of this compels us to reject H31. In order to accept it, we would have

to find, first, that the percentage of control officers had served in the military without being

disciplined was lower than the percentage of control officers without military records. It is

not. Sliced another way, the data in Figure VIII-48 disclose that 54.8 percent of the 1,862

non-veterans we studied were members of the control group, as compared to 48.1 percent of

the 876 veterans who had not been subjects of military discipline. Thus, even though we did

reach the second finding predicted in H31 B only 36.4 percent of the 195 veterans with

military discipline were in the control group, significantly lower than the 48.1 percent of

veterans without discipline, we must reject H31. Instead, it appears that, with or without

histories of military discipline, veterans were more likely than non-veterans to have been

involuntarily separated by the NYPD during 1975-96.

This finding runs contrary to what one would expect based on widely held beliefs and

policies that grant extra credit or even absolute preference to veterans who compete for civil

service jobs like policing. In searching to explain it, we considered whether it might be

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



254

attributable to spikes in separations, perhaps caused by the introduction into the NYPD of

veterans whose experiences in the military, or some other consideration, should have kept

them out of policing. Most specifically, much has been written on Vietnam-era veterans and

their difficulty in adjusting to civilian life and, presumably, to sensitive careers like police

work (see, for example, Kendal, 1978; Mitchell, 1988; Otten, 1978; Salasin,1981; Silva, et al.,

2001). The3 general tenor of this literature B that Vietnam veterans are especially vulnerable

to maladjustment, addiction, criminality, and other manifestations of posttraumatic stress

disorder B would lead one to expect that officers who served in the military during the

Vietnam era would subsequently be most likely to be separated from the NYPD.

We examined this theory in Figures VIII-49 through VIII-51. Figure VIII-49 tracks

the percentages of military veterans among officers separated during each year and the

percentage of veterans among each annual cohort=s matched control officers. Starting at

Figure 49's left side, for example, we see that 75 percent of the officers separated in 1975

(n=55) were military veterans, as compared to 69 percent of their matched control officers.

After that year, the data for each group follow very much the same trend (r=0.65), with the

percentage of military veterans among study officers consistently higher than the percentage

of veterans in the comparable control groups. Only in the figure=s very last year (1996, when

75 percent of the 83 study officers and 15.5 percent of the controls were veterans) is there any

suggestion that the trend might be changing. Thus, there is no evidence in the figure that

officers separated in any particular year or groups of years account for the disparity in military

service among the study and comparison officers.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-49: Percentage of Study and Control Officers with Military Experience,

by Study Officer Separation Year

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1975

1978

1982

1983

986

198

990

1993

1996

Study Officers Control Officers

Fyfe and Kane --



255

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



256

Table VIII-50 also offers no support to any theory about unfit Vietnam veterans. Here,

we again see that the percentage of veterans among each annual cohort=s study group officers

typically is higher than that of the control officers, but that the correlation between the two is

extremely high (r = 0.87).

112

Table VIII-51 suggests that the issue of when officers entered the military is more

salient than when they entered or left the NYPD. Here, we can see that the number of study

and control officers who entered the military was remarkably consistent during 1942 through

1973 (r=0.90), but that this relationship becomes irregular thereafter (r for 1974-95 = 0.09).

All other things equal, the random manner in which our control group was selected would

cause one to expect that approximately the same number of study and control group veterans

would have entered the military during each year. Throughout the first 32 years covered by

Figure VIII-51, the annual frequency of study group veterans entering the military typically

outnumbered control group veterans by a slight margin. This continued throughout the

Vietnam era, but changed in . 1974. In 1974, 33 of the officers we studied entered the

military; 30 of these were control officers, and only three were study officers. This was the

year before the formal cessation of hostilities, and several years after the 1967-69 period in

which the war=s heaviest fighting and casualties were suffered. After that, the pattern became

extremely irregular. In 1975, 17 study officers and seven control officers entered the military;

in 1976 the split was 59 study officers and 11 control officers; in 1980, nine study officers and

112

During 1946-60, 34 veterans entered the NYPD and wound up in the study

group, while our control officers for the same years included 31veterans. For purposes of

conciseness, we collapsed them into one category on Figure VIII-49.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-50: Percentage of Study and Control Officers with Military Experience,

by Year Officers Appointed to NYPD

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1946-60

1961-6

1966

1979

1983

198

1990

991

1993

1994

1995

Military No Military

Fyfe and Kane --



257

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-51: Year in Which Study and Control Veterans Entered the Military

1942

1950

1954

1980

Study Control

Fyfe and Kane --



258

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



259

48 control officers. Despite these annual fluctuations, the ratio of study:control officers

remained relatively constant over the period studied (e.g., 52.9 percent of the veterans who

entered the military during 1942-73 found their way into the study group, versus 54.1 percent

during 1974 through the end of the study).

We then attempted to determine whether explanations for this might be associated with

variations in officers= military experience. Figure VIII-52 describes the branches of the

service, highest military ranks, and length of active duty among the study and control officers

who served in the military. There are significant differences on all three dimensions. The

figure shows that Navy and Air Force veterans are more frequently found among control

officers than among study officers, and that Marine Corps veterans are overrepresented among

study officers. Fewer than 15 percent of the study veterans had served in the Navy (14.8%),

as compared to 21.4% in the control group. Similarly, 12.9 percent of the study veterans and

17.1 percent of the control veterans had been in the Air Force. Conversely, 25.3 percent of the

study veterans and only 15.6 percent of the control veterans were Marine Corps alumni.

113

The representation of Army veterans in both groups is virtually identical (study=47.0%;

control=45.8%). Thus, while many former Marines have achieved great success in the

113

Excluded from the figure are small percentages of study and control officers

who had served in the Coast Guard or in other countries= military services.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-52: Military Experiences of Study and Control Officers

14.8%

25.3%

12.9%

69.6%

0.9%

21.4%

77.6%

22.4%

21.4%

17.1%

40.5%

7.1%

29.5%

47.0%

9.9%

90.1%

14.4%

52.3%

15.6%

45.8%

Air Force

Pvt

Cpl/Seama

Sgt/

ipli

0-4

ars act

years active

Study Control

Fyfe and Kane --



260

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



261

NYPD,

114

it also appears that Marine veterans are more likely than veterans of other services

to be involuntarily separated from the NYPD.

There also are significant differences between the military ranks achieved by separated

and control officers, and they are in the direction one might expect: separated officers

generally achieved less success in the military than did their control group colleagues. Seven

in ten of the study group officers (69.6%) never rose above the rank of private, private first

class, corporal or seaman, as compared to half (52.3%) of the control officers. Conversely,

four in ten control officers (40.5%) and three in ten study officers (29.5%) achieved sergeant

or petty officer ratings. Fewer than one in a hundred of the study group (0.9%) achieved

commissioned officer status in the military, while control officers accomplished this

distinction eight times as frequently (7.1 percent).

Finally, the right side of the figure shows a very strong association between length of

active military service and whether officers were separated. Here, we see that the percentage

of study group veterans who served a normal full four-year enlistment or less (77.6%, n=433)

is lower than that of the control officers (90.1%, n=436). Conversely, of course, study officers

who served five or more years in the military (and who therefore presumably reenlisted)

114

Distinguished Marine Corps alumni dominate the NYPD=s top command.

The current Police Commissioner, Deputy Commissioner for Counter Terrorism, Chief of

Department, Chief of Patrol, Chief of Personnel, and many other top administrators are

Marine Corps veterans.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



262

outnumber control veterans with such long service by more than two to one (22.4%, or 125,

versus 9.9%, or 48).

Thus, our analysis of military experience among the officers we studied has produced

some complicated and surprising results. First, military experience bears a relationship with

involuntary separation from the NYPD that is the converse of what we expected B military

veterans are more likely than non-veterans to have been separated by the department during

the period we studied. But this difference is largely accounted for by officers who did not

advance in the military, and/or who served more than one enlistment, and/or who were

disciplined in the military, and/or who served in the Marines. We shall return to examine this

issue later in this report, when we construct multivariate models that distinguish between

study and control officers. Here, we find that much of the relationships between military

experience and police career success may be attributable to other, confounding, variables.

Education and Involuntary Separation

We postulated that:

H32: Rates of involuntary separation for all types of police misconduct are inversely

associated with officers' years of education at entry into the NYPD.

H33: Officers= years of education at entry into the NYPD is positively associated

with movement into supervisory and management ranks.

H34: Officers= educational attainment after entry into the NYPD is positively

associated with movement into supervisory and management ranks.

115

Our original proposal also hypothesized some relationships between officers=

scores on standard IQ tests and their dismissal rates. We have since found that the NYPD

has not regularly tested or recorded IQ scores, and have removed this discussion from the

current proposal.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



263

We were able to test H32 and H33 but, because of the sketchy nature of data on post-

entry educational attainment among officers who had left the NYPD by the time we began our

research, we were unable to test H34.

Education and Separation. We tested H32 by aggregating officers= educational levels

at entry into four categories: General Equivalency Diploma/Less than High School Diploma;

Four-Year High School Graduate; Some College (less than an associate=s

degree);Associate=s Degree; and Baccalaureate and Beyond. The results are shown in Figure

VIII-53, and confirm the hypothesis (p < .001). Beginning at the left, the figure shows that

17.2 percent of the study officers and 12.5 percent of the control officers had a General

Equivalency Diploma or had not finished high school when they became police officers.

116

Forty two percent (41.7%) of the study officers and 35.9% of the control officers entered the

department holding high school diplomas and no further education. The figure=s central bars

show no real

differences, with about about 35 percent of each group having completed some college not

leading to any degree (study=34.7%; control=36.9%). Differences show up in the figure=s

two right side bars, with the percentages of control officers who had achieved associate or

baccalaureate degrees (7.1% and 7.6%, respectively), more than twice as high as the

comparable percentages for study officers (3.1% and 3.3%).

116

The data include seven officers, all hired in the 1950s, who entered the

department without having completed high school educations or equivalency diplomas.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

igure VIII-53: Separated and Control Officers' Educational Levels

at Entry to the NYPD

17.2%

41.7%

34.7%

3.1%

3.3%

12.5%

35.9%

36.9%

7.1%

7.6%

GED/Less than H.S.

Grad

H.S. Grad Some College Associate Degree Baccalaureate +

Study Comparison

Fyfe and Kane --



264

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



265

Education and Rank To test H33, we included in analysis only officers who had

completed five or more years of service prior to the 1996 end of the period we studied. We

did this on grounds that it has been rare for officers to achieve supervisory rank in less time

than this. We also split officers on the study/control dimension. As we have already seen,

there are differences in rank between the two; further, most control officers= careers

continued beyond their matched study officers= separation dates. The results appear in

Figure VIII-54, which confirms the hypothesis. The figure shows that 3.4 percent of the study

officers and 2.8 of the control officers with GEDs or less had attained supervisory or

command rank (n=148 and 143, respectively) and that these percentages trend upward as

educational level increases. By the time we reach the highest level, we find that 13.3% of

study officers (n=15) and 18.3 percent of control officers (n=93) holding baccalaureate

degrees at their appointment went on to achieve rank in the NYPD.

Education and Career Success. The clear conclusions of our tests of these two

hypotheses are, first, that marginally educated officers were significantly more likely than

their better educated colleagues to end their NYPD careers with involuntarily separations; and,

second, that better educated officers are more likely than their colleagues to advance in rank.

These are findings consistent with a string of prior research (Cohen and Chaiken, 1972, 1973;

Bowker, 1980; Murrell, 1982; Reuss-Ianni, 1983; Hayeslip, 1989; Kappeler, Sapp, and Carter,

1992; O=Sullivan, 1994; Palombo, 1995; Eterno, 1996; Smith and Aamodt, 1997; Johnson,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-54: Percentage of Study and Control Officers with Five+ Years Service

in Supervisory or Command Ranks, by Educational Level at Entry into NYPD

3.4%

5.8%

6.5%

8.3%

13.3%

2.8%

9.1%

7.7%

9.2%

18.3%

GED/Less than H.S.

Grad

H.S. Grad Some College Associate Degree Baccalaureate +

Study Control

Fyfe and Kane --



266

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



267

2000; Polk and Armstrong, 2001; Johnston and Cheurprakobkit, 2002; Lersch and Kunzman,

2002),

117

as well as with the recommendations of virtually every authority and scholar who

has ever spoken or written on the issue (American Bar Association, 1973; Clark, 1970;

diGrazia, 1977; Eskridge, 1989; Fosdick, 1920; Goldstein, 1977; Guyot, 1979; Leonard and

More, 1964; McManus, et al., 1970; Murphy and Plate, 1977; Myrdal, 1944; National

Advisory Commission on Civil Disorders, 1968; National Advisory Commission on Criminal

Justice Standards and Goals, 1973; Philadelphia Police Study Task Force, 1987; President=s

Commission, 1967; Saunders, 1970; Sherman, et al., 1978 ;Vollmer, 1972; Wilson, 1950;

Witham, 1985). Certainly, our findings ratify the NYPD=s decision to require that new

officers have completed 60 college credits, approximately the equivalent of an associate

degree, before entering the department.

Academy and Probationary Performance and Involuntary Separation

We hypothesized that:

H35: Rates of involuntary separation for all types of police misconduct are inversely

associated with officers' Police Academy academic averages.

We present data responsive to this hypothesis, as well as data related to study and

control officers= performance on the Police Academy=s exit physical strength and agility

examination in Figure VIII-55. To avoid confounding our results, we excluded probationers

117

Worden (1990) is one of few studies that did not report a relationship between

police educational attainment and some aspect of police job performance.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-55: Study and Control Officers's Academic and Physical Scores in

Police Academy (Excludes Separated Probationers)

79.9%

85.8%

83.4%

88.3%

Academic Average Physical Education Score

Study Control

Fyfe and Kane --



268

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



269

from the figure, on grounds that their academic and physical test performances might affect

decisions about whether to terminate them for misconduct. Thus, the figure includes only

officers who were separated from the NYPD after they had successfully completed their

probationary periods as the result of decisions made by officials who, as is the practice, would

not have taken Police Academy performance into account in making disposition decisions.

The figure shows confirms the hypothesis. It shows that the study officers for whom

Police Academy data were available attained a mean academic average of 79.9 percent, as

compared to 83.4 percent for control officers. Thus, H35 is confirmed. The figure also

shows that study officers did not perform as well as control officers on the Police Academy=s

exit physical examination (mean scores = 85.8% and 88.3%).

These trends are repeated in virtually all objective assessments of study and control

officers= performance as recruits. Figure VIII-56 presents data on the Police Academy sick,

lateness, and disciplinary histories of study officers who subsequently were separated after

completing their probationary periods, along with their matched control officers. The figure

demonstrates that, during their recruit training, study officers were more likely than control

officers to have reported sick with line of duty injuries (5.0% v. 1.0%); to have reported sick

with non-line of duty afflictions (30.7% v. 19.8%); to have been late (21.7% v. 19.5%); to

have been subjects of demerits (64.5% v. 58.6%); command disciplines (19.0% V. 11.6%);

and department charges (2.6% v. 0.7%). Study officers were also more likely than control

officers to have been held back from graduation with their class (3.9% v. 1.2%).

In Figure VIII-57, we see that this pattern includes the frequency with which study and

control officers reported sick or late or were disciplined while in recruit training. Study

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-56: Study and Control Officers' Sick, Late, and Disciplinary Records

in Police Academy (Excludes Separated Probationers)

0.05

0.31

0.22

0.65

0.19

0.03

0.04

0.01

0.12

0.59

0.20

0.01

Line of Duty Sick

Leaves

Non Line of Duty

Sick Leaves

Latenesses Demerits Command

Disciplines

Department

Charges

Held Back

Study Control

Fyfe and Kane --



270

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-57: Frequency of Study and Control Officers' Sick and Late Reports,

and Disciplinary Records in Police Academy (Excludes Separated Probationers)

0.15

0.98

0.43

2.58

0.38

0.09

0.01

0.43

0.29

1.52

0.26

0.02

N. Line of Duty

Sick Leaves

N. Non Line of

Duty Sick Leaves

N. Latenesses N. Demerits N. Command

Disciplines

N. Department

Charges

Study Control

Fyfe and Kane --



271

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



272

officers reported sick in the line of duty far more often than control officers (means=0.15 and

0.01). They reported sick for non-line of duty reasons more than twice as often as control

officers (means=0.98 and 0.43). Study officers were counted half again as often as control

officers among those who reported to work late (means=0.43 and 0.29), received demerits

(means=2.58 and 1.52), or were subjects of command discipline (means=0.38 and 0.26).

Finally, study officers were brought up on department charges more than four times as often

as control officers (0.09 v. 0.02).

In Figure VIII-58, we see that this trend is interrupted where the subjective evaluations

of recruits officers prepared by Police Academy staff are concerned. The figure=s first three

sets of columns show instructors= ratings of recruits on specific and clearly defined

performance dimensions. They show that control officers received excellent ratings (either

Aextremely competent@ or Ahighly competent@) nearly twice as often as study officers (means

= 3.80 and 2.10), and that study officers were slightly more likely than control officers to

receive poor (Aextremely low@ or Alow@) ratings (0.26 v. 0.23).

On the basis of such objective measures of their performance as academic and physical

school grades, lateness, sick reports, and discipline B where they consistently did more poorly

than control officers B one might expect that study officers would also be given low ratings by

their recruit instructors. This is not the case: instead, study officers were given higher

subjective ratings than control officers by their instructors in all three academic disciplines

(Law, Police Science, Social Science) and the Physical School, as well by their official (home

room) instructors. There clearly is a disconnect here between the objective and subjective

ratings. It suggests strongly that the subjective evaluations of instructors B which should be

based, insofar as possible, on objective performance measures B have little or no validity.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-58: Study and Control Officers' Mean Scores on Police Academy

Instructional Staff Ratings (Excludes Separated Probationers)

2.10

4.76

0.26

2.74

2.79

2.65

2.64

2.80

3.80

2.58

2.43

2.42

2.57

2.50

0.23

4.40

N. Excellent

Ratings

N. Competent

Ratings

N. Poor Ratings Law Rating Police Science

Rating

Social Science

Rating

Physical Rating Overall Rating

Study Control

Fyfe and Kane --



273

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



274

Police Job Performance and Involuntary Separation

We hypothesized that:

H36: Rates of involuntary separation for all types of misconduct are positively

associated with the following variables (all standardized by rates per year of service):

H36a: prior citizens' complaints:

H36b: prior supervisory disciplinary actions;

H36c: prior Abelow standards@ evaluations on performance evaluations;

H36d: prior line of duty civil suits;

H36e: prior line of duty injuries;

H36f: prior line of duty vehicle accidents;

H36g: prior designations as chronically sick.

H37: Rates of involuntary separation for brutality and other abuses are positively

associated with rates of departmental commendations.

We encountered problems testing these hypotheses as they were stated. We found that

data on evaluations were spotty and inconsistent, due largely toi varying evaluation forms

used at different times by the three agencies (NYPD and the former Transit and Housing

Authority police agencies) that were merged to form the current NYPD. This made analysis

of H37c impracticable. Further, at the time we formulated our hypotheses, we were led to

believe that the NYPD=s automated systems were better developed than they were. Thus,

when we began our data collection effort, we found that we were unable to collect data on line

of duty injuries,

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



275

line of duty vehicle accidents, and chronic sick designations.

118

We also were unable to

retrieve data on civil suits against officers. This left us unable to test H36d-H36h or H37.

Prior Complaints.

Finally, we also found it impossible to retrieve data that

distinguished citizens= complaints from those filed by ranking officers in the department.

Consequently, we combined H36a and H36b into a single hypothesis, worded as follows:

H36a: Rates of involuntary separation for all types of misconduct are positively

associated with records of prior complaints when standardized by rate per year of

service.

The data confirm this hypothesis. We found that, for most officers, complaints were a

rarity, but that the study officers accumulated them more quickly than did the control officers.

The study officers had as many as 19 prior complaints, but for purposes of presentation, we

collapsed the data into six annual rate categories.

119

Figure VIII-59 shows that fewer than

half of the study officers (46.5%) had not accumulated any complaints by the times of their

separations. We collected data on complaints received by control officers through the dates at

which their matched study officers left the department, and found that three-quarters (74.6%)

had received no complaints. As we move to the right of the figure, the disparity

118

Much of this difficulty also was attributable to the difficulty of reconciling the

records of the NYPD and the former Transit and Housing Authority police agencies

following their 1994 merger. The senior author=s more recent experiences indicate that

these data are now retrievable, but such data collection is beyond the scope of this study.

119

We attempted to exclude from this analysis any complaints that were the direct

causes of study officers=s separations from the NYPD, and are comfortable that few, if

any, slipped through our net..

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Figure VIII-59: Annual Rates of Complaints against Study and Control Officers

Prior to Study Officers' Separation

15.4%

4.4%

7.6%

6.4%

74.6%

15.6%

1.3%

46.5%

10.6%

0.8%

1.6%

6.1%

None .01-.25 .26-.50 .51-.75 .76-1 More than 1

Study Comparison

Fyfe and Kane --



276

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



277

between study officers and control officers grows, resulting in annual means (not shown in

figure) of 0.28 complaints for study officers and 0.08 complaints for control officers. Clearly,

there were precursors to the conduct that finally ended our study officers= careers.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



278

IX. MULTIVARIATE ANALYSIS

To this point, this report has analyzed a series of bivariate relationships in a broad

examination of police misconduct in New York City. These relationships with significance tests

and odds ratios are summarized in Table IX-1.

Although the study/control design lends itself well to bivariate comparisons, it is

nevertheless useful to submit the misconduct data to multivariate analyses in order estimate the

extent to which some of the bivariate relationships may have been confounded by exogenous

factors. An additional justification for conducting a multivariate examination of the misconduct

data is to move beyond single item indicators that may not fully capture the extent to which

complex social processes may have influenced variations in misconduct patterns.

For example, the data in Table IX-1 show that pre-employment arrests for violent,

property, and public order crimes are all significantly associated with police misconduct. What

is not readily known, however, is whether the criminal history indicators might converge to

suggest discrete dimensions of prior criminal involvement that may differentially predict police

misconduct. Uncorrelated composite indices derived from data reduction techniques may predict

patterns of police misconduct in ways that differ from the single item indicators. Thus, given the

depth and range of the data under analysis, the present section investigates complex and

potentially subtle relationships that the simple bivariate analyses may have been unable to

identify.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



279

Table IX-1. Comparisons Between Study and Comparison Officers Along Predictor Variables of

Interest

Research Group

Variables

Study Comp. Odds

Pre-Employment Personal History

Female

White

Black

Hispanic

Other

Born in New York City

Born Elsewhere in New York State

Born in other US State

Born in Puerto Rico

Born outside US

Less than 22 Years Old at Appointment

Never Married

Divorced/Separated

Children at Appointment

Reside Outside City

Arrested

Violent Crime Arrests

Property Crime Arrests

Public Order Crime Arrests

JD/YO Findings

Misdemeanor Convictions

Moving Violations

Parking Violations

Criminal/TAB Summonses

Own a Motor Vehicle

Driver License Suspended

GED or Less Than High School Grad

Associate Degree or Higher

Actively In School at Appointment

Unemployed for 30 Days or More

Professional or Technical Occupation

Fired from Job

Negative Comments by Prior Employer

Disciplined by Prior Employer

Workers' Compensation Claims

Previous Civil Service Rejections

Previously Test for NYPD

Prior Police Experience

15.1

56.8

30.5

11.9

0.5

75.8

5.7

11.3

1.9

3.2

27.7

64.0

4.3

26.2

14.8

23.3

3.1

7.6

14.6

9.7

4.7

60.8

28.9

14.4

46.2

17.9

17.2

6.4

7.2

63.4

5.7

15.3

10.0

15.3

5.5

15.3

7.9

3.6

13.4

78.9

10.6

9.6

0.5

80.3

9.0

5.3

2.3

2.5

24.4

66.7

2.1

20.3

29.2

13.9

1.5

3.6

5.0

4.2

1.9

55.2

30.8

8.8

53.9

11.8

12.5

14.7

9.3

58.3

8.3

7.7

4.7

8.3

5.2

9.3

7.9

4.4

n/s

.001

.048

n/s

.034

.001

n/s

.040

n/s

.001

.003

.001

.002

n/s

.001

.050.

.006

.017

.001

n/s

.001

n/s

1.13

0.72

2.88

1.24

1.00

0.94

0.63

2.13

0.83

1.28

1.14

0.96

2.05

1.29

0.51

1.68

2.07

2.11

2.92

2.31

2.47

1.10

0.94

1.64

0.86

1.52

1.38

0.44

0.77

1.09

0.69

1.99

2.13

1.84

1.06

1.65

1.00

0.82

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



280

Previously a Transit Police Officer Pre-merge

Previously a Housing Police Officer Pre-merge

Merged from Transit Police in Unification

Merged from Housing Police in Unification

Previously an NYPD Civilian Employee

Previously an NYPD Cadet or Trainee

Owed More than $10,000 at Appointment

Held Home Mortgage at Appointment

Held Vehicle Loan at Appointment

Held Student Loan at Appointment

Held Credit Card Debt at Appointment

Held Personal Loan at Appointment

Outstanding Judgments at Appointment

Father an NYPD Officer

Members of Immediate Family Been Arrested

NYPD Background Investigator Recommend Not

Hiring

Any NYPD Staff Recommend Not Hiring

Reside in City When Last Known

Married When Last Known

Single When Last Known

Children at Separation

Below Standards Performance Dimension Ratings

Below Standards Overall Performance Ratings

Military History

Military Service

Among Veterans Only:

Army Veteran

Navy Veteran

Marine Veteran

Air Force Veteran

Private/PFC/Corporal or Seaman at Discharge

Sergeant/Petty Officer at Discharge

Officer at Discharge

Court Martialed/Disciplined

Police Academy Performance

Extremely/Highly Competent Overall Ratings

Academic Average

Extremely/Highly Competent Law Ratings

Extremely/Highly Competent Police Science Ratings

Extremely/Highly Competent Social Science Ratings

Physical Education Average

Extremely/Highly Competent Physical Education

0.9

0.3

1.0

0.9

2.8

5.3

3.3

1.9

12.7

12.0

7.3

15.1

1.7

8.0

25.8

15.4

16.3

71.2

42.0

48.7

31.0

21.7

13.1

40.8

43.6

13.7

23.5

11.9

69.6

29.5

0.9

21.4

19.7

80.4

27.9

22.0

32.1

86.6

2.5

0.6

1.0

0.9

1.7

4.2

4.1

2.6

13.4

12.4

7.9

13.4

0.7

13.0

20.3

7.0

6.8

53.3

65.5

26.9

23.7

12.4

3.1

32.5

44.7

20.9

15.2

16.7

52.3

40.5

7.1

14.4

38.0

84.1

44.2

38.0

48.9

89.9

.002

n/s

.01

.001

.003

.001

0.36

0.50

1.00

1.65

1.26

0.80

0.73

0.95

0.97

0.92

1.13

2.43

0.62

1.27

2.20

2.40

1.34

0.64

1.81

1.31

1.75

4.23

1.26

0.98

0.66

1.55

0.71

1.33

0.73

0.13

1.49

0.52

0.96

0.63

0.58

0.66

0.96

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



281

Ratings

Number of Extremely/Highly Competent Ratings

Number of Low/Extremely Low Ratings

Recruit Demerits

Command Disciplines

Formal Department Charges

Non-Line of Duty Sick Leaves

Line of Duty Sick Leaves

Latenesses

Departmental History

Prior Complaints

Assigned to Patrol at Separation

Assigned to Investigative Unit at Separation

Assigned to Administrative Staff Job at Separation

Sergeant or Higher at Separation

42.3

2.07

0.50

71.7

26.6

5.7

36.3

3.6

26.1

59.8

59.3

5.2

7.1

3.2

56.1

3.22

0.20

64.4

11.7

0.6

22.4

1.6

18.9

47.9

51.1

11.7

13.1

8.2

.001

0.75

0.64

2.50

1.11

2.27

9.50

1.62

2.25

1.38

1.25

1.16

0.44

0.54

0.39

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



282

ANALYTICAL PROCEDURES

Independent Variables and Principal Component Analysis

In an effort to protect against excessive mulitcollinearity and to identify discrete

dimensions of officers’ personal histories, the independent variables under consideration were

submitted to a series of data reduction analyses using the principal component extraction

method. Three sets of logically related variables were entered into three separate analyses in

order to identify relatively uncorrelated dimensions of officers’ criminal histories, social

conditions at appointment, and employment histories. The following sections report the findings

of these analyses.

Criminal History. The first principal component analysis (PCA) entered all criminal

justice contact variables into a model that ultimately extracted three components. The variables

included numbers of violent and property crime arrests, juvenile delinquency findings,

misdemeanor arrests and convictions, public order arrests, moving and parking violations, and

driver’s license suspensions. Table IX-2 shows the results of this analysis.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



283

Table IX-2. Principal Component Analysis of Criminal History Variables

________________________________________________________________________

Components

1 2 3

EV=2.81 EV=2.00 EV=1.82

Variables

%Var.=20.1 % Var.=14.3 % Var.=13.0

Violent Crime Arrests .684 -- --

Property Crime Arrests .703 -- --

Juvenile Delinquency Findings .514 -- --

Misdemeanor Convictions .496 .398 --

Public Order Arrests -- .948 --

Moving Violations -- -- .732

Parking Summonses -- -- .627

Driver’s License Suspensions -- -- .545

________________________________________________________________________

1. All variables indicate the number of occurrences prior to officer’s appointment to the NYPD

As the data in Table IX-2 show, the PCA extracted three components. These explained

47.4 percent of the cumulative model variance. The first may be considered a traditional

criminal history dimension on which violent, property, and misdemeanor crimes loaded, as well

as juvenile delinquency findings. Officers scoring high on this index would have engaged in

criminal or delinquent activity that began during their teenage years. The second component

may be regarded as a public order dimension of criminal activity on which only arrests for public

order and misdemeanor convictions loaded. The final dimension is associated with vehicle code

violations on which numbers of traffic and parking summons, as well as driver’s license

suspensions loaded.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



284

All components were saved as variables to be entered as covariates into the logistic regression

analyses.

Work History. The next principal component analysis examined groupings of officers’

work history indicators and included the following variables: numbers of jobs, workman

compensation claims, 30 day (or longer) periods of unemployment, jobs from which fired, work-

related disciplinary actions, and derogatory comments by prior employers. Table IX-3 shows the

results of this analysis.

As the data in Table IX-3 show, the current PCA extracted two dimensions of officer

work history, explaining 49.1 percent cumulative variance. The first may be regarded as a pre-

employment disciplinary problems component on which numbers of jobs from which officer was

fired, job-related disciplinary actions, and derogatory

Table IX-3. Principal Component Analysis of Work History Variables

Components

Variables

EV=1.91

% Var.=31.9

EV=1.04

% Var.=17.2

Jobs from which officer was fired

Number of job-related disciplinary actions

Derogatory comments by prior employers

Workman compensation claims

Number of jobs held

30 day (or longer) periods of

unemployment

.765

.748

.718

.525

.640

.685

1. All variables indicate the number of occurrences prior to officer’s appointment to the NYPD

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



285

comments by prior employers loaded. The second may be regarded as an employment reliability

construct on which numbers of workman compensation claims, jobs held, and 30 (or longer)

periods of unemployment loaded. The two components were saved as variables to be entered into

the logistic regression analyses that follow.

Social Condition. The final PCA examined the groupings of variables describing officers’

social circumstances at the time of their appointment

120

and included marital status (coded as a

series of binary variables), number of kids, and whether officer was actively enrolled in school at

time of appointment. Table IX-4 shows the results of this analysis.

Table IX-4. Principal Component Analysis of Social Condition Variables

Components

Variables

EV=2.47

% Var.=27.5

EV=1.09

% Var.=12.1

Married

Number of children

Single (never been married)

Actively enrolled in school

Divorced

Separated

Living with partner

.916

.704

-.891

.546

.706

.543

1. All variables indicate the number of occurrences prior to officer’s appointment to the NYPD

120

This analysis was conducted in part because the initial logistic regression models found no

associations between marital status variables and misconduct. In an effort to more fully capture

officers’ social conditions at the time of their appointment, marital status was combinedwith other

social indicators in an attempt to identify the more complex construct of social condition

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



286

The results in Table IX-4 show that the PCA extracted two dimensions of officer social

condition at time of appointment, explaining 39.6 percent cumulative variance. The first may be

considered a conventional family construct on which married and number of children loaded.

The second is more open to interpretation, but suggests a post marriage living situation

associated with school enrollment. This second component, which explains just over 12 percent

of the model variance, may represent “second start” indicator for many officers in the sample. As

with in the previous analyses, the dimensions extracted by the current estimation were saved as

variables to be entered as covariates into the logistic regression models that follow.

Logistic Regression Analysis

In an attempt to distinguish study officers from comparison officers along the specified

set of predictor variables, the present study employed logistic regression modeling, which is

useful estimating the independent effects of a set of predictor variables on a binary outcome,

while controlling for the influences of other covariates. The analysis used all 3,085 cases and set

the dependent variable to predict the study group (i.e. comparison officer=0; study officer=1).

All categorical variables were entered into the model using indicator contrasting, which

designates a reference category against which the remaining categories are compared. For

example, the variable MILITARY included the categories, army, navy, air force, marine corps,

other, and no military. The reference category was no military, so that the potential impact of

military service on

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



287

misconduct will be contrasted against no military service. The indicator categories for all

nominally measured variables are noted in Table IX-5.

FINDINGS

At step 0 the model produced a -2 Log likelihood function of 2,128.61 and classified 51.6

percent of all cases correctly based on the intercept alone. At step 1 (during which all

independent variables were simultaneously entered), the model produced a final -2 Log

likelihood function of 1,604.77 (p<.001) and correctly classified 86.5 percent of all cases

correctly. The final correct classification percentage coupled with the Cox & Snell R

and

Nagelkerke R

(.55 & .73, respectively) suggest a highly internally reliable model with

exceptional predictive utility. Table IX-5 shows the results of the logistic regression analysis.

As the data in Table IX-5 show, the logistic regression model identified both risk and

protective factors for career ending police misconduct. While controlling for all other covariates,

the strongest risk factors for misconduct were black officer (B=1.19), an average of one or more

complaints per year of service (B=1.11), and working in inspector precincts at time of incident

leading to separation (B=.91). Specifically, black officers were 3.27 times more likely than white

officers to be dismissed for job-related misconduct; officers receiving an average of one or more

complaints per year were 3.03 times more likely to be dismissed for misconduct compared to

officers who received between zero and one complaint per year; and officers assigned to

inspector precincts were 2.48 times more likely than officers assigned to staff units to be

dismissed for occupational misconduct.

This document is a research report submitted to the U.S. Department of Justice. This report has not

been published by the Department. Opinions or points of view expressed are those of the author(s)

and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

Fyfe and Kane --



288

Table IX-5. Logistic Regression Estimates and Odds Ratios Predicting Police Misconduct

Variable B S.E. Sig. Odds

Dad NYPD Officer

Military Service

Army

Navy

Air Force

Marine Corps

Command Risk

Inspector Precinct

DI Precinct

Captain Precinct

Police Academy/Field Training Unit

Proactive Investigation (Narc/SCU)

Detective Bur/Warrant Squads

OCCB Field Units

Special Patrol Units

Supervisory Rank

Age at Appointment

Average Number Complaints per Year

Education Level at Appointment

College Study (No Degree)

Associate’s Degree

Bachelor’s Degree or Higher