Audit of the Pittsburg Police Department, Part 1
by Tom Flaherty
Editor's Note: The following audit of Pittsburgh's Police
Department and the Office of Professional Standards (recently re-
organized and renamed the Office of Municipal Investigations) was
released to the public on Aug. 20, 1996. The audit, performed by
City Controller Tom Flaherty's office, was first requested early in
1996 by the Pittsburgh City Council, specifically by Councilor
Valerie McDonald. McDonald also is among the councilors (and
citizens) calling for an independent civilian police review board to
investigate citizens complaints of misconduct including physical
and verbal abuse by Pittsburgh police.
Interestingly, even though the audit reveals that 59 cops have had
five or more complaints launched against them by civilians or other
cops, and even though 40 percent of cops on the force have had at
least one complaint launched against them, Mayor Tom Murphy
claims the audit confirms his contention that problems at the
Pittsburgh Police Department amount to only "a few bad apples,"
an assertion repeated by the Pittsburgh media, most notably and
emphatically by the Pittsburgh Post-Gazette. (You'll note that the
audit itself even prescribes at times the "few bad apples"
conclusion, although the report makes much of and media accounts
have underplayed the problems of how individual complaints were
handled by police investigating themselves.)
We offer this report in an attempt to cut through media bias
(including our own) and let readers draw their own conclusions. It
is a report about something no less fundamental than justice and,
sadly, politicians and news organizations haven't given it justice.
The report below is complete save for a few things including: a
long appendix about "data verification" that explains statistical
analysis and is fairly arcane; and several graphs about Pittsburgh
crime statistics that proved prohibitively difficult to reproduce on
our web site. (A hard copy of the full report is available by calling
412 255-2055)
Finally, a word about the context of this report. In October of 1995,
police outside of Pittsburgh beat to death Jonny Gammage, a
motorist stopped for no apparent reason other than being a black
man driving an expensive car (a Jaguar) in a white neighborhood.
Also, the City of Pittsburgh is the target of a class-action suit filed
by the American Civil Liberties Union on behalf of 52 plaintiffs
alleging excessive force and verbal abuse. At this writing (August,
1996), the officers in the Gammage incident are awaiting trial while
Mayor Murphy, who first dismissed the ACLU suit out of hand, is
discussing not only police "reform" but also coming to a settlement
with the ACLU. Also, as mentioned above, the city council is
considering adopting a civilian review board, the subject of several
recent public hearings. Four council members favor a civilian
review board, which the mayor opposes, favoring the present
system of police investigating complaints internally.
Andy Newman
To read In Pittsburgh Newsweekly's ongoing coverage of police
accountability, type "police" in our search feature.
------------------------------------------------------------------------
PERFORMANCE AUDIT
DEPARTMENT OF PUBLIC SAFETY
OFFICE OF PROFESSIONAL STANDARDS
Report by the Office of the City Controller
TOM FLAHERTY
CITY CONTROLLER
Anthony Pokora, Deputy Controller
Kevin Forsythe, Esq., Management Auditor
Anabell Kinney, Esq., Assistant Management Auditor
Performance Auditors:
Gloria Novak
Joseph Chigier
Jeff Khadem
Woody Mudd
August, 1996
INTRODUCTION
From 1986 up through April 1996, the Office of Professional
Standards (OPS), within the Department of Public Safety, received,
investigated, monitored, compiled data, and rendered decisions
with respect to citizens¡ complaints about employees of the
Department of Public Safety: Police officers, fire fighters,
paramedics, building inspectors, and other administrative staff. In
addition, at the request of the City Solicitor, OPS has investigated
employees in other City departments, such as Public Works and
Parks and Recreation. Although OPS has handled complaints
involving other employees besides police officers, this report,
unless otherwise noted, deals exclusively with police officers.
This audit formally began with an entrance conference on March 6,
1996. Subsequently, in April of 1996, the Murphy Administration
began a process designed to transfer the responsibilities of OPS out
of the Public Safety Department into the Law Department, with the
new rubric of Office of Municipal Investigations (OMI). The
Administration hired a former FBI agent to head the new OMI,
replacing the departed Assistant Chief of OPS. Also during the
course of field work on this audit, City Council began a series of
neighborhood public hearings on the issue of creating a Civilian
Police Review Board. So at the writing of this audit, it is unclear
how the City will handle future citizen complaints against police
officers. The Controller¡s Office hopes that the findings and
recommendations of this report will assist the Mayor and Council
to provide the citizens of Pittsburgh and Public Safety Management
with the best possible system, regardless of its ultimate form. All
recommendations should be read as addressed to the ultimate
successor of OPS, whether that turns out to be the Law
Department¡s Office of Municipal Investigations, some form of
Civilian Review Board, or some other entity.
Originally named the Office of Professional Responsibility, OPS
received its present title in April of 1989. Its offices were located
on the seventh floor of the Public Safety Building adjacent to
administrative offices of the Department of Public Safety. Earlier
this year the Assistant Chief, who had been the head of OPS since
1989, left to pursue a career opportunity in the private sector. The
remaining staff consisted of one Civilian Coordinator, two civilian
investigators, two detectives, two police officers and one clerk-
stenographer. Citizens filed complaints at this location where an
intake clerk interviewed the citizen and filled in a complaint sheet
like the one on the next page.
The complaint was then assigned to an investigator who would
generally schedule another interview with the complainant, at
which time he/she would attempt to identify witnesses and gather
other pertinent information. Upon completion of the investigation
OPS issued a final report with one of five dispositions: Sustained,
not sustained, unfounded, closed, or exonerated.
When OPS made a final disposition of sustained, it meant that,
based on its review of the evidence, OPS had concluded that the
officer against whom the complaint was filed has acted improperly.
Proscribed conduct could include the violation of police
regulations such as excessive use of force, verbal abuse, under
the influence of alcohol while on duty, or the catch-all, conduct
unbecoming an officer. It could also involve violations of state or
federal statutes and/or depriving a complainant of state or federal
constitutional rights.
OPS then sent the sustained complaint up the chain of command,
starting with the Commander of the police officer in question.
From the Commander the complaint moved on to the Assistant
Chief of Police, next it went to the Chief of Police, then to the
Director of Public Safety. At each step in this chain, the reviewing
supervisor could reject or accept the OPS finding. If the OPS
finding of sustained was accepted, each representative of Public
Safety Management could suggest or select an appropriate
discipline including verbal reprimand, written reprimand, one-day
suspension, three-day suspension, or five-day suspension pending
dismissal. In some cases Public Safety Management would direct
the officer to attend counseling or training sessions. Final
responsibility on all discipline rests with the Director of Public
Safety.
The police officer then had the option of accepting the disciplinary
measure or challenging it in one of two ways: By filing a grievance
through the police union, or by taking the disputed disciplinary
action through statutory arbitration (in reality a hybrid of State
law and an arbitration award handed down November 26, 1990). In
its 10-year career, OPS witnessed the dismantling of the Police
Trial Board system in 1990 and its replacement by this current
statutory arbitration scheme under which the police union chooses
one arbitrator, the City (Public Safety Management) chooses
another and a third, neutral arbitrator is selected by the first two
arbitrators from a panel provided by the American Arbitration
Association. This is a change from the old Trial Board which was
composed exclusively of police officers.
The police union grievance procedure can also end up eventually in
arbitration, if the parties fail to reach an agreement after a series of
meetings and procedures spelled out in the bargaining agreement.
Unlike the statutory arbitration scheme outlined above, this
proceeding will be held before a single American Arbitration
Association arbitrator. It is unnecessary for purposes of this report
to distinguish further between these two types of arbitration. The
important point is that any discipline given to an officer as a result
of Public Safety Management¡s agreement with an OPS finding of
sustained, could be challenged and taken to arbitration. The
decisions of these arbitrators could, under very limited
circumstances, be appealed to the Allegheny County Court of
Common Pleas. The Pennsylvania Commonwealth Court hears any
appeals from Common Pleas Courts; and from there, appeal may be
taken to the Pennsylvania Supreme Court, but, again, only under
certain circumstances.
It is crucial to note with regard to OPS findings of sustained, that
OPS did not have any power to discipline officers. It did not have
any power to compel Public Safety Management to mete out a
certain type of discipline. It did not have any recourse if Public
Safety Management decided to ignore or overturn its findings. It
did not even recommend types of discipline. It simply and
exclusively was a fact finding agency within the Department of
Public Safety. Its sole job was to make an initial determination as
to whether a violation of applicable law and/or departmental
regulations had occurred. OPS did, however, have the authority to
initiate an investigation against an officer, without requiring a
complaint to be filed by a citizen.
An OPS finding of not sustained meant that on the basis of its
investigation OPS concluded that the complainant failed to
establish the charges with the requisite degree of certainty. This is a
concept similar to burden of proof. In these instances OPS was
not saying that the officer did nothing wrong. It was saying that it
couldn¡t, on the basis of the evidence before it, make a
determination of improper conduct. Findings of not sustained
often emerge in cases where the complainant is the only witness to
the alleged misconduct. In other words, there is no corroborating
evidence, and in these situations OPS invariably opted to accept the
officer¡s version of the incident (primarily because no one in Public
Safety Management and no arbitrator would reprimand an officer
solely on the basis of uncorroborated evidence). Or the complainant
may have brought forward other witnesses, but in the opinion of the
OPS investigator these witnesses were biased or unreliable for one
reason or another (often because they are relatives or because they
were arrested along with the complainant on similar charges).
A finding of unfounded meant that on the basis of its
investigation OPS had determined that the story of the complainant
had no merit. This is very different than cases deemed not
sustained, in which the story of the complainant may have had a
good deal of merit, but lacked a threshold level of competent
evidence. Unfounded cases are much more like decisions in favor
of the officer and against the complainant. Typical unfounded
situations occur when the complainant offers no evidence in
support of the charges, or when independent corroborating
evidence (uninvolved bystanders, for example) clearly demonstrates
the truth of the officer¡s version of the story. The police collective
bargaining agreement dictates that anonymous complaints with no
corroborative evidence must be classified as unfounded. It also
mandates that complaints which, even if true, would not lead to
criminal charges against the officer, must be classified as
unfounded if they are made more than 90 days after the incident.
Finally, the F.O.P. contract directs that all unfounded OPS
complaints must be removed from OPS records and the police
officer¡s personnel file after one year.
Sometimes a complainant would refuse to cooperate with OPS
investigators after the initial intake interview. These instances were
classified as closed and no further action was taken by OPS or by
Public Safety Management.
A finding of exonerated meant that on the basis of its
investigation OPS had determined that, even if everything alleged
in the complaint was true, the officer acted properly under the
circumstances.
By Resolution No. 18 of 1996 (enacted January 23, 1996 and
effective, February 2, 1996) Pittsburgh City Council directed the
City Controller¡s Office to conduct a performance audit focusing
on the number of disciplinary actions, types of discipline
administered to police employees, and any disciplinary patterns
originating from investigations of citizens¡ complaints filed with
OPS (and also the Pittsburgh Human Relations Commission). A
copy of the Resolution appears on the following page.
In preparing for this performance audit we reviewed the City
Controller¡s 1994 Performance Audit on Police Officer Hiring; the
current collective bargaining agreement between the City and the
Fraternal Order of Police, Fort Pitt Lodge No.1 (Working
Agreement, expiration date, December 31, 1997); the Disciplinary
Manual of the City of Pittsburgh; OPS quarterly and year-end
reports 1993, 1994, and 1995; a Recent Chronology on Police
Civilian Review Board prepared by the Community Relations
Committee of the Pittsburgh Human Relations Commission, April
25, 1996; Year-end Reports on Intake Activity prepared by the
Pittsburgh Human Relations Commission for the years 1994 and
1995; a Memorandum dated January 16, 1996 to Sal Sirabella,
Deputy Mayor, from David Miller, Managing Director of the
Pennsylvania Economy League, recommending that the City
expand the mission of OPS to include investigating allegations of
misconduct for all City employees; The Bureau of Police, 1994 and
1995 Statistical Reports; numerous journal and newspaper articles.
We also attended four of City Council¡s neighborhood public
hearings on the Citizen Police Review Board.
OBJECTIVES
1. To conduct random testing on a sample of the OPS database
received from Public Safety Management in order to verify its
accuracy.
2. To quantify the total number of citizen complaints filed against
police officers with OPS and their disposition by OPS, from its
inception in 1986 up through the first two months of first quarter,
1996; and to present this data by year and overall.
3. To generate a series of statistical reports using as variables race,
gender, and age, of the complainant, and race, gender, and starting
date of the officer, in order to identify possible patterns; presenting
this data from 1986 up through the first two months of first quarter,
1996 by year and overall.
4. To organize data on the number of officers against whom
multiple citizen complaints have been filed from 1986 up through
the first two months of first quarter, 1996.
5. To evaluate the performance of Public Safety Management in
administering discipline in response to OPS findings of sustained
during the years 1991-1995.
6. To present the above data in the context of other relevant
statistics on crime rates and police activity.
7. To make constructive recommendations for improvement.
SCOPE
The scope of this performance audit is the entire tenure of the
Office of Professional Standards, from its inception in 1986 up
through the first two months of first quarter, 1996. Resolution No.
18 of 1996 specified a scope of 1992-1995. In the course of
discussions among representatives from the City Controller¡s
Office, City Council, the Law Department and the Department of
Public Safety, it was mutually agreed to expand the scope to the
entire time period OPS had been in existence. Then, during the
course of field work the scope of Objective No. 5 was subsequently
limited to the years 1991-1995, for reasons explained below in the
Methodology Section.
This scope is limited by the Objectives stated on the previous page
and in several other important respects:
1. Unless otherwise noted, this audit deals only with complaints
filed against police officers, and not against any other employees of
the Public Safety Department.
2. This audit seeks to evaluate the response of Public Safety
Management (during the specific time period of 1991-1995) to
OPS findings of sustained. It does not attempt to second guess the
work or conclusions of the OPS investigators.
3. This audit expresses no opinion upon, and should not be read as
a comment on, the degree of police misconduct in the City of
Pittsburgh. We took the level of citizen dissatisfaction as we found
it, reflected, for purposes of this audit, in the records of OPS.
4. This audit expresses no opinion upon, and should not be read as
a comment on, any past, present, or future lawsuits filed against the
City of Pittsburgh and/or City police officers or other personnel.
5. This audit expresses no opinion upon the advisability of either
the new Office of Municipal Investigations or a Citizen Police
Review Board.
METHODOLOGY
I. Database Verification (Objective No. 1)
In the course of preliminary discussions among the auditors, Public
Safety Management, and City Council Members, we learned of an
OPS database. This computer aided record-keeping program had
been used by OPS since its inception in 1986 to organize and store
data on all its cases, involving all Public Safety employees; police
as well as fire fighters, paramedics, building inspectors, etc. It
existed on the hard drive of the City Information Services (CIS) D-
Base Level 6 system. OPS investigators and data entry clerks
accessed this database through terminals at their office. As cases
progressed through their various stages, these OPS personnel made
appropriate entries into one of over 40 different fields in the
database program.
We first notified Public Safety Management in January of 1996
that we would be needing this database for purposes of our
performance audit. A series of meetings and additional discussions
(including the March 6, 1996 entrance conference) ensued in which
we further defined the audit¡s scope and addressed concerns of
Public Safety Management with regard to security and
confidentiality of police personnel records under the terms of the
current collective bargaining agreement. The Law Department also
became a participant in this process at this point.
In the first week of April, 1996 we received the entire OPS
database on 3 1/2 inch hard disc. We then used this database on
Controller¡s Office computers, utilizing software programs in D-
Base, Lotus, and SPSS (Statistical Package for the Social
Sciences) to generate the reports which form the heart of this audit.
But before we could rely on the data we had received, we verified it
by randomly generating a 2.6% sample (originally 3%, see
Appendix, page i) and testing it for accuracy against the actual OPS
case files (hard copy).
The database contains 46 different informational entries, or fields.
[Some of the fields for example, program, status, DAR number,
and trial board were no longer used by OPS and considered
obsolete. We did not use these fields in our verification
comparison]. The database consists primarily of information which
was first recorded on an intake form by OPS personnel, then
entered into their computer. When this intake form was available in
the case file, we used it for direct comparison. When it was not
available, database information was compared to whatever could be
found in the case file (complaint form, police report, etc.) that
might verify the entries.
The results of our verification testing appear in the Appendix to
this audit. On the basis of those results we concluded that the OPS
database we had received from Public Safety Management was
sufficiently reliable to allow us to proceed with the remainder of
the audit.
II. OPS Statistical Overview (Objective No. 2)
Having satisfied ourselves of a sufficient degree of accuracy in the
OPS database, we proceeded to the audit¡s second Objective: To
quantify the total number of citizen complaints filed against police
officers with OPS and their disposition by OPS, from its inception
in 1986 up through the first two months of first quarter, 1996; and
to present this data by year and overall.
At this point, in order to capture as many records as possible, we
manually searched the database for incomplete entries in the fields
of bureau code, officer I.D. number, officer race, and officer
gender. We found 28 of these. If we had done nothing further to
them, they would not have been picked up by the various sorts we
intended to run. We were able to correct all 28 of these incomplete
entries. This was usually possible through an alphabetical sort
which listed all complaints against a particular officer in a row. If,
for example, the officer appeared four times as a white male, and a
fifth time as blank, and if all five entries had the same officer
identification number, then we felt justified in completing the
database at that point by adding W and M.
Although we had done the verification testing with the entire OPS
database, from this point on we were concerned only with police
officers. We accordingly sorted the data using the bureau letter
code, P, and obtained a listing of only those complaints filed with
respect to Police Bureau personnel. We then cross checked this
listing to identify all records for which the accused had no I.D.
number. The lack of this badge number indicated it might not be a
police officer. It might, for example, be a parking meter reader who,
up until their transfer to the Public Parking Authority of
Pittsburgh, were within the Police Bureau. Working record by
record with the Coordinator of OPS, we identified 14 of these that
did not involve a police officer and identified the rest as police
officers. We eliminated the non-officers from the database. We now
had an extremely close approximation to a database containing only
complaints filed against police officers.
Next we separated out all complaints initiated by Public Safety
Management. We did this by sorting out all complaints with the
case-type letter codes of BI (Bureau Initiated) and SI (Special
Investigations). Typical examples of these would have the
complainant listed as Dir Cannon or Car Gedman. This step was
taken for two reasons: First, because it was our objective to find the
total number of citizen complaints, and second, because these
management-initiated complaints were irrelevant to the next
Objective of the audit involving racial, gender, and age/experience
patterns.
This is a crucial point to keep in mind when reading the pages that
follow. Management-initiated complaints might deal with matters
like a police officer with a DUI; alcohol on the breath at the
station; dress code violations; tardiness; etc. These usually do not
provide a helpful indicator of citizen dissatisfaction with the
police. Citizens complain about excessive use of force, verbal
abuse, refusal to fill out a police report, conduct unbecoming an
officer, etc. This distinction will be maintained throughout this
audit, and whenever relevant we will present the data both ways
(i.e., looking only at citizen complaints against police, then doing
the same analysis using total complaints against police).
We had now corrected the OPS database to an optimum level of
accuracy and distinguished between citizen complaints and
management-initiated complaints. From this we ran a report listing
dispositions (sustained, not-sustained, closed, exonerated)
and compared disposition rates for management complaints to those
for citizen complaints. Next, with citizen complaints only, we
focused on the various types of police misconduct alleged, for
example, verbal abuse, excessive force, etc. Sorting with the
database field, offense code, we generated two reports: The first
showing the percentage breakdowns for the most frequent offenses,
and the second comparing disposition rates for these offenses.
Our findings with respect to Objective No. 2 appear at pages 15-22.
III. Race, Gender, Age/Experience Analysis (Objective No. 3)
Our next Objective was: To generate a series of statistical reports
using as variables race, gender, and age, of the complainant, and
race, gender, and starting date of the officer, in order to identify
possible patterns; presenting this data from 1986 up through the
first two months of first quarter, 1996 by year and overall.
For this part of the audit we used our modified database containing
only citizen complaints against the police. Using management-
initiated complaints in this portion of the audit would only skew
the results. The fact that a former Chief of Police was a black male,
or that a former Director of OPS was a white female, is irrelevant to
the type of patterns which this part of the audit was designed to
identify; viz., patterns of possible racial or sexual
discrimination/harassment in the relationships between the police
and the citizenry. Using SPSS software on the citizen complaint
database, we generated reports for the total audit period and for
each year, showing first the race of complainant vs. the race of the
officer, then gender of complainant vs. gender of officer.
Next, we sorted the citizen complaints by age of the complainant
and, with SPSS software, calculated an average age for each of the
years in the audit period.
The database had no field for age of the police officer, but it did
have one for starting date. We decided to use this to calculate
average experience level of the officers and to chart any significant
change in this variable over time. We performed this operation by
converting the citizen complaint database from D-Base to Lotus
then subtracting date of complaint from starting date of the police
officer. This gave us the number of days of experience of the officer
at the time the complaint was filed; an extremely accurate
measurement. We next divided this number by 365 to give years of
experience, then finally ran an average for each year in the audit
period. The main purpose for these calculations was to test the
hypothesis that the upswing in citizen complaints beginning in
1994, could be attributed to the influx of less experienced, newly
hired police officers.
Reports and findings with respect to Objective No. 3 appear at
pages 23-30.
IV. Multiple Citizen Complaints Against an Officer (Objective No. 4)
We now used the same modified database of citizen complaints
only, for the audit¡s fourth Objective: To organize data on the
number of officers against whom multiple citizen complaints have
been filed from 1986 up through the first two months of first
quarter, 1996.
We sorted this database by accused officer¡s name and by officer
identification number. This identified those police officers against
whom more than one citizen complaint had been filed for the audit
period of just under 10 years. Through a conversational shorthand
this portion of the audit came to be known as the bad apple
analysis. The purpose was to see whether or not our data lent
support to the claim that a small number of rogue cops were
responsible for the perceived problems which led to the call for this
audit.
We first generated a number of findings for the 10 year audit period
as a whole. Next, we compiled year by year breakdowns for the
years 1991-1995 sorting now by ID Nos. of officers with two or
more complaints filed against them for each year, and by the
starting date of each officer so identified. This allowed us to
observe any trends in the multiple complaint problem over time.
Auditors manually sifted through various print-outs of the modified
database described above to gather the findings which appear at
pages 30-35.
|