Prosecution and Racial Justice
Using Data to Advance Fairness in Criminal Prosecution
By Wayne McKenzie, Don Stemen, and Derek Coursen

Prosecutors in the United States have an unrivaled level of
influence within the criminal justice system. They decide,
among other things, whether to file criminal charges, the
number and severity of offenses they will charge, whether
to offer a plea bargain, and what sentence to recommend
for defendants who are convicted. These decisions can
have a profound impact on the outcome of a case and
the life of a defendant. Yet, as they exercise this influence,
prosecutors also have unrivaled independence. Unlike officials in law enforcement and the judiciary, who have come
under varying degrees of oversight in recent years (see
box on page 2), prosecutors act with little outside scrutiny
or governance.
The discretion that prosecutors have is valuable for a
number of reasons. Most important, it provides flexibility
so they can tailor an appropriate response to individual
cases. Yet it can also lead to unfair, disparate treatment.
For many people, the possibility that minorities, especially
African Americans and Latinos, might be prosecuted differently from white defendants is of particular concern.
Statistics show that African Americans, for example, account for 39 percent of the population within the criminal
justice system, even though they make up only 13 percent
of the national population.1 The Vera Institute of Justice’s
Prosecution and Racial Justice (PRJ) program seeks to address this concern by building confidence that prosecutorial discretion is not contributing to the disproportionate
representation of people of color in the criminal justice
system.
In partnership with district attorneys in three major metropolitan counties—Milwaukee County, Wisconsin; Meck-

MARCH 2009

lenburg County (Charlotte), North Carolina; and San Diego
County, California—PRJ is piloting an internal assessment
and management procedure to help supervisors identify
evidence of possible racial bias in prosecutorial decisions
and respond appropriately when it is found. PRJ is doing
this by helping district attorneys collect and analyze data
about their office’s structures and processes which they can
then use to take corrective action when necessary—without the need for potentially costly and disruptive intervention from external entities.
PRJ’s process is based on a principle that informs all Vera
work: effective reform combines robust and careful data
analysis with a commitment to use the information to improve policies. This publication presents an overview of
how this principle is being applied to prosecutorial decision making, oversight, and supervision. The report begins
by identifying how PRJ and its partners developed their
approach to data-based reform. It then discusses lessons
that have been learned as the district attorneys began applying this methodology to ensure the integrity of the justice they deliver.

Developing a New Approach to
Prosecutorial Management
Through Data
District attorneys are elected on their promise to protect
communities from crime. Many chief prosecutors gauge
their ability to fulfill this promise by asking one question:
“How many charged crimes end in a conviction?” According to this widely accepted approach, a high conviction

Prosecution and racial justice

Prosecution and Racial Justice

rate indicates a successful office; a low rate demonstrates
discover, for example, that line prosecutors are seeking
less success. Yet closer consideration shows that the conmore severe penalties for a category of defendants who
viction rate is an incomplete measure of both performance
are otherwise similar to defendants for whom they seek a
and success. Because it reflects countless decisions at evlesser punishment. This insight could be the impetus for
ery stage of the prosecutorial process—from whether to
focusing more attention on understanding this imbalance
press charges to whether to seek specialized sentencing
and, if necessary, implementing corrective measures.
options—the conviction rate can conceal evidence of
In addition to the four key discretion points, the project
poor or unsatisfactory performance
at any point in the process. Consequently, it cannot adequately guide
Monitoring Discretion in Law Enforcement
supervisors on how well their offices
are operating and which practices
In recent years, law enforcement agencies and the courts have become
and processes warrant improvement.
increasingly subject to external constraints and oversight—due, at least
The conviction rate fails, therefore,
in part, to concerns about fairness and disparity. Many police departas a complete and useful measure of
ments have come under federal scrutiny, for example, because of patprosecutorial performance for which
terns of unacceptable practices, including allegations that they unfairly
it is often taken.
and disproportionately targeted minorities for traffic stops or searches.
Similar concerns about equal treatment led legislators in many jurisdicTo provide district attorneys with intions, including the federal government, to create sentencing structures
formation they can act on, PRJ’s staff
and guidelines designed to limit the decision-making options of judges.
of researchers and former prosecutors
developed a series of performance indicators—select statistics that provide
insight into how a system is operating—that focus on four key discretion
points in the prosecutorial process:
initial case screening, charging, plea
offers, and final disposition (see table
on page 4). When taken together, the
indicators describe with meaningful
specificity how the exercise of discretion at each point contributes to the
final outcome of a criminal prosecution, thus providing managers greater
opportunity for performance assessment and supervision.
By collecting data about, say, which
charges prosecutors decline to pursue, or which sentences they request,
supervisors can begin to identify facets of the system that appear to be
performing inappropriately and may
be in need of attention. They may



Prosecution and racial justice

As PRJ’s partners in Charlotte, Milwaukee, and San Diego work to enhance their policies, they are following the example of several law enforcement leaders who earlier traveled a similar path and did so without
the coercion of federal oversight.
One example is from the late 1990s. The former U.S. Customs Service, the
federal agency then entrusted with enforcing the nation’s customs laws,
faced well-substantiated accusations that it was unfairly targeting African
Americans and Hispanics for invasive searches. Raymond W. Kelly, the
agency’s chief at the time, responded by collecting data on the search
practices of customs agents. The data showed that racial disparities
arose when agents exercised independent discretion in choosing who
to search. Kelly responded by developing a policy requiring agents to
consult with a supervisor before proceeding with a search. As a result,
disparities shrank and the rate of positive searches—those that led to
recovery of contraband—increased.* Kelly also instituted the then-novel
practice of asking for daily reports on searches by race, which allowed
him to monitor patterns, identify problems, and implement solutions on
an ongoing basis.
* An in-depth description of Kelly’s changes can be found in “A Case Study: How One Police Agency
Changed for the Better” in Profiles in Injustice by David A. Harris, (New York: New Press, 2002): p. 208-22.

Using Data to Advance Fairness in Criminal Prosecution

also tracks time to disposition, which is similarly important. After all, in cases where discretion may be a factor
in disparate outcomes, it might be expressed by longer
prosecution times for certain defendants who are otherwise comparable to those whose cases are resolved more
quickly.
PRJ acknowledges that the factors it tracks in its analysis—
the performance indicators at each of the four discretion
points and time to disposition—are not comprehensive.
Neither are they all determined exclusively by prosecutorial discretion: a defendant released on bond, for example,
may feel less urgency to cooperate to swiftly resolve a case

that could result in a criminal conviction and jail time than
a defendant who is in pretrial detention facing a similar
charge and fate. Nevertheless, the indicators are useful in
identifying racially disparate outcomes and determining
whether they are the result of racial bias.
The philosophy underlying PRJ’s performance measure
approach is not unique. In 2004, the American Prosecutors Research Institute (APRI), the research arm of the National District Attorney’s Association, proposed an initial
framework for a new way of measuring prosecutors’ progress toward achieving widely accepted goals and desired
outcomes. Their strategy for a more effective measure of

Seeking Culture Change
Like any organization, a district
attorney’s office will have a distinct
culture. The adversarial nature of
most criminal justice proceedings
in the United States, the field’s traditional emphasis on convictions
as a measure of success, and the
leadership’s vulnerability to political
change all contribute to the character of this culture.
Many prosecutors and their staff
regard self-monitoring as an unnecessary and even risky proposition—especially
self-monitoring
for hints of racial bias. It is as if by
choosing to look for this evidence
they are admitting to being guilty
of the practice. Other agencies
have demonstrated, however, that
self-monitoring promises many
benefits, including a greater sense
of integrity associated with the

criminal justice system. When prosecutors hold themselves accountable for their decisions, they gain
the support of their constituencies
and the benefit of a supportive environment.
Yet, prosecutors who commit to
monitoring their discretion must
be prepared to change the culture
of their offices. PRJ facilitates this
change by creating regular meetings where discussion of race is not
only sanctioned but encouraged by
senior management. Prosecution
offices that are open to uncovering
existing racial disparities can work
more effectively to reduce such disparities in the future.
An important component of this
culture change is gaining trust.
Prosecutors can gain the trust of

community and civic organizations
by engaging in open discussions
that demonstrate their commitment to a more just criminal justice
system. Milwaukee District Attorney
John Chisholm has demonstrated
this commitment by openly discussing with community groups his
findings and the remedial steps his
office has undertaken in response
to them. Mecklenburg District Attorney Peter Gilchrist has also met
regularly with community groups
about his commitment to reducing
racial disparities in the prosecutorial process. San Diego District Attorney Bonnie Dumanis has several
community liaisons on staff and
regularly meets with various community groups to share information
and discuss concerns.

vera institute of justice



Prosecution and Racial Justice

prosecutorial performance called for a similar, multiple-indicator mechanism.2

operations. This makes it possible to identify areas that
warrant managerial attention and oversight.

The multiple indicator approach is the leading edge of
PRJ’s effort to help district attorneys identify and address
unwarranted disparities in the treatment of minority defendants. It is complemented by the project’s equal emphasis
on working with offices to ensure that the data collection
and analysis process is practical, and that the resulting information is effective in examining problem areas and developing appropriate policy responses.

One of the first important lessons learned, however, is that
prosecutors’ offices often lack the technological capacity
necessary to collect data easily, if at all. The second lesson
is that once the broad technological hurdles are surmounted, the process must be targeted to optimize the resulting
data’s usefulness.

Collecting Data Indicators:
What We Are Learning
Like APRI’s proposed alternative measure of prosecutorial
success, PRJ’s system of multiple indicators enables supervisors to monitor decisions in specific segments of their

THE NEED FOR ADEQUATE SYSTEMS TO COLLECT
DATA. Prosecution offices often use electronic case management systems to follow the progress of their cases.
Such systems are rarely designed to marshal the aggregate information required to track disparity, however. A
standard case management system may make it possible
to follow the decisions of individual prosecutors in specific
cases, but it probably cannot identify how an office of prosecutors exercised its discretion collectively. Consequently,

Performance Indicators for the Four Key Discretion Points in the Prosecutorial Process
CASE SCREENING

CHARGING

PLEA OFFERS

FINAL DISPOSITION

> Arrest charges (specific statute citation)

> Highest charge

> Highest charge offered

> Trial date

> Arrest charge level (level/class of
felony/misdemeanor)

> Highest charge type

> Highest charge offered
type

> Disposition (dismissed, found
not guilty, found guilty)

> Arrest charge crime type (person,
property, drug, public order/
government, sex, weapons, DUI/traffic)

> Highest charge crime
level

> Highest charge offered
crime level

> Highest disposition charge

> Arrest date

> Highest charge crime
type

> Highest charge offered
crime type

> Highest disposition charge
type

> Number of counts booked

> Total number of
charges

> Total number of
charges offered

> Highest disposition charge
crime level

> Case status (accepted/declined)

> Charging date

> Plea offer date

> Highest disposition charge
crime type

> Plea acceptance date

> Total number of charges
convicted

> Case deferred
> Case diverted

> Sentence recommendations

> Reason for declination

> Disposition date



Prosecution and racial justice

Using Data to Advance Fairness in Criminal Prosecution

such systems offer little support for supervisors looking to
address overall challenges.
The fact that case management systems rarely allow for
aggregating case information is but one of several related
challenges. Any system that could provide aggregate information would likely have to be computerized. Yet, prosecution offices, including our partners in Mecklenburg
County, often rely on paper files for managing their cases.
This is not uncommon in smaller jurisdictions where the
absence of a computer-based system may have a minimal
effect on overall work flow, or in offices with limited resources. Another frequent challenge is that crucial information
may be stored across different agencies. In many jurisdictions information about arrest charges and custody status,
for example, may be maintained by the sheriff, while pleas
and sentencing are recorded by the court.
Although an ideal solution would be a web-based data
collection system that integrates all agencies holding relevant information, the Mecklenburg and Milwaukee district
attorneys have demonstrated that intermediate solutions
are also viable.
As a first step in its data collecting effort, Mecklenburg
County hired data entry support staff to computerize case
information from its paper files, expediting review of initial
case screening decisions. This allowed supervisors to compare the aggregate number of drug cases in which charges
were declined to cases in which the charges were pursued.
In Milwaukee, prosecuting officials overcame the second
obstacle—data being maintained by different agencies—
by soliciting data from those agencies and incorporating
it into a single database using case identifiers that were
common to all data sets.
IMPROVING DATA QUALITY. It is not enough simply to
collect data. The data must be collected in ways that allow supervisors to compare the influence of the many factors that can affect case outcomes. PRJ has identified two
principles of data collection that substantially enhance the
usefulness of prosecutors’ databases.

First, data collection systems should be sophisticated
enough to allow prosecutors to record multiple independent variables at each discretion point. One of the earliest discretion points, for example, case screening, reflects
prosecutors’ initial judgment about cases as they enter
the office, usually from the police. The relevant information that should be collected within this indicator category
includes the total number of charges against the arrested
person (arrest charges and number of counts), the charge
type, the crime level, and the crime type. Offices must be
able to identify non-legal information about the defendant
as well, including, most saliently, the defendant’s race. If a
prosecutor chooses not to charge a case, the reason for
the declination should be recorded too.
Second, data should be collected at both the defendant
and charge levels. If data is considered only at the defendant level—tracking only what happens to the defendant’s
case as a whole—a great deal of information about separate charges is lost. A single defendant may face multiple
charges, many of which can result in different outcomes.
Some of these charges may be declined and others accepted; some may be upgraded and others downgraded.
What happens to each of these individual charges determines what happens to a defendant’s case as a whole and
determines whether two similarly situated defendants are
treated disparately. For example, two defendants may be
arrested for the same five charges. The first may have four
charges dismissed and one accepted while the second
defendant may have all five charges accepted. Considered only at the defendant level, both individuals would
be seen as accepted for prosecution. This, however, would
overlook the fact that the first defendant is only prosecuted on one charge while the second is prosecuted on five
charges—a very different outcome.
On the other hand, if data is considered only at the charge
level, there is a risk that some individuals will be counted
twice—where one or more charges are rejected and one
or more charges are accepted—resulting in another kind
of distortion. If this information is not collected specifically
and accurately, supervisors will be unable to assess fully
how decisions are made in their offices.

vera institute of justice



Prosecution and Racial Justice

The data collection issues described here are complex.
But our experience has shown that once understood, the
principles are relatively simple to begin implementing.
The district attorneys’ offices in Charlotte, Milwaukee, and
San Diego have each begun tracking prosecutorial decisions in case screening and charging, according to their
individual capacities. They have also begun to see positive
results, in terms of their ability to identify areas in need of
examination and possible managerial oversight. Some of
these gains are highlighted in the following sections of this
report. In time, PRJ’s partners expect to extend their oversight systems to include the full range of indicators at each
discretion point: case screening, charging, plea offers, and
final disposition.

charged with possession of drug paraphernalia compared
to only 27 percent of non-whites arrested for the same
crime. After looking at the data, the team considered a
number of possible explanations for this disparity. These
included policing practices, case screening procedures,
and unconscious bias based on the character of the drug
paraphernalia involved. In the course of their discussions,
the team considered whether police were treating people
differently, whether prosecutorial staff had a legally relevant reason to press or decline to press charges differently,
and whether the disproportion was based on an unconscious racial bias.

Analyzing Data:
What We Are Learning

Percent of cases declined, by race

Data alone does not provide answers. It does, however,
help to determine what additional questions should be
asked. Developing a structured, recurring way to look
at—or analyze—such data and then to apply that analysis
to managerial protocols is central to the approach developed by PRJ and its partners.
PRJ’s partner jurisdictions review data and discuss what
data mean at regularly scheduled management meetings.
Charlotte and Milwaukee have instituted new meetings
dedicated specifically to this undertaking; San Diego is integrating it into an existing general management systems
process. Whatever the venue, these discussions involve
prosecutors and a team of legal, non-legal, and technology personnel who meet to examine, analyze, and interpret
any evidence that suggests a racial bias in aggregate decision making. PRJ staff familiar with the issues are initially
facilitating these meetings to help determine which factors, such as office policies and training, may be relevant
to the disparities that have been found.f
Recently, Milwaukee County held meetings to discuss data
revealing an unexpected racial disparity in drug cases, as
illustrated in Chart A, at right. For example, Milwaukee
prosecutors chose not to prosecute 41 percent of whites



Prosecution and racial justice

In this case, racial bias was not deemed to be the most
relevant factor. The disparity was traced, instead, to pros-

White defendants

Non-white defendants

Chart A: Declination rates for
possession of drug paraphernalia

41%

27%

White defendants = 521
Non-white defendants = 846
16%
12%

12%
7%

Possession
of drug
paraphernlia

Possession of
marijuana (1st
offense)

Possession of
cocaine (1st
offense)

Chart B: Overall declination rates
40% 40%

37% 37%
32% 33%

White defendants
3,104 felonies
5,467 misdemeanors
Non-white defendants
7,642 felonies
10,091 misdemeanors

Misdemeanor

Felony

All offenses

Using Data to Advance Fairness in Criminal Prosecution

ecutors’ level of experience. Junior prosecutors had traditionally been assigned to screen misdemeanor cases. A
significant number of the paraphernalia cases that originated within the city of Milwaukee, where most of the African American population resides, involved possession
of crack pipes. However, in the suburbs of the county of
Milwaukee, the paraphernalia was more varied. Junior
prosecutors associated the crack pipes with crack cocaine
and pursued charges more aggressively. At the same time,
they viewed other forms of paraphernalia as less serious
and not worth pursuing. The result was that more African
Americans were prosecuted. More experienced prosecutors, on the other hand, tended to view possession of drug
paraphernalia charges generally as relatively minor and
not worth pursuing.
How prosecutors subsequently addressed this disparity is
the subject of the next section. It is first worth noting, however, that this disparity would have been masked if Milwaukee staff did not look into the data for the most accurate
answers. As Chart B illustrates, the overall declination rates
of whites and non-whites show no imbalance at all. It was
only when staff considered the declination rates for specific crime types, such as possession of drug paraphernalia,
that the disparity was revealed.

Lessons for Developing
Management Protocols and
Implementing Solutions
Once data has been collected and analyzed, PRJ’s process
encourages district attorneys to consider implementing
policy changes to address any imbalances that have been
identified. In the earlier example from Milwaukee County,
District Attorney John Chisholm encouraged staff to view
possession of crack cocaine paraphernalia less as a criminal matter than as evidence that the arrested individual had
a problem with drug abuse. He enacted a policy that directed staff to decline these cases whenever it was reasonable to do so and to refer the arrested individuals to drug
treatment. When prosecutors still seek to press charges,
a supervisor’s approval is required to ensure that lack of
experience on the prosecutor’s part is not a salient factor.

Although these policy changes do not directly focus on
racial issues, soon after they were implemented the racial
disparity in drug paraphernalia charges disappeared.
Similar policy changes were implemented in Mecklenburg
County where, as noted earlier, data entry personnel were
hired to input information from paper files. That process
revealed that Charlotte’s prosecutors were filing charges
in approximately 97 percent of all drug cases. This was
an extraordinarily high percentage, given that the office’s
declination rate for all cases combined was roughly 30 percent. Further scrutiny revealed that many of the charges
were eventually dismissed and a significant number of the
cases were referred to drug treatment later in the process.
Additionally, charges were being pressed for all drug cases
and every drug charge involving African American women.
Many of these cases, too, were being resolved later in the
prosecutorial process. District Attorney Peter Gilchrist responded to these findings by instituting a more vigilant
screening process that identified weak cases up front. As a
result, the declination rate in drug cases overall increased
to approximately 13 percent. Because Charlotte prosecutors were pursuing fewer minor charges that would eventually be dropped—including many against black women—more resources were available to prosecute serious
cases that went forward.
A final example, from Milwaukee, shows that supervisors
are increasingly recognizing that the interpretation of data,
and not the data itself, is the key to management and reform. During a meeting to review declination rates, a finding that minorities were less likely to be prosecuted for
property offenses was initially presented as evidence that
there was no racial bias in how such cases were handled.
Extensive discussions among managers within the office,
however, yielded several other plausible and less comforting conclusions. Perhaps there were fewer cases with
minority defendants because minority victims were reluctant to step forward, law enforcement was less willing to
treat such crimes against minorities seriously, or prosecutors were less inclined to appropriately value the property
rights of minority victims who are often demographically
similar to their victimizers.

vera institute of justice



Conclusion
Prosecutors exercise significant discretion over the cases they handle. Their choices extend from whether to
press charges at the beginning of the process, to their
role in seeking plea bargains and making recommendations about bail and post-conviction disposition (seeking
placement in an alternative to incarceration, for example).
Clearly articulated legal factors, internal policies and practices, and ethical considerations constrain these choices,
and historically this has been sufficient to sustain public
confidence in the integrity of the prosecution function.
In recent years, however, other significant actors in the
criminal justice system who once enjoyed similar autonomy have become subject to increasing levels of external
oversight. Strict guidelines have been enacted to limit
sentencing options available to judges, and many police

departments discovered to be treating people differently based on their race have come under federal scrutiny
and, in some cases, direct oversight. In both cases, a loss
of public confidence was an important catalyst for the
change. Prosecutors cannot assume that they are immune
to similar forces.
In partnering with PRJ, the district attorneys of Charlotte,
Milwaukee, and San Diego have made a commitment to
sustaining the public’s confidence. They have done so by
assuming a leadership role in ensuring that neither race
nor ethnicity are intentionally or unintentionally producing
unfair outcomes or inappropriate racial disparities. Our experience has shown, moreover, that even when a disparity
is not racially motivated, PRJ’s approach to internal oversight can enhance public confidence in the fairness of the
prosecutorial function. It is therefore an important model
for prosecutors everywhere.

endnotes
William J. Sabol, Todd D. Minton, and Paige M. Harrison, Prison and Jail Inmates at Midyear 2006 (Washington, DC, Bureau of Justice Statistics: 2007), NCJ 217675, p.9,
http://www.ojp.usdoj.gov/bjs/pub/pdf/pjim06.pdf; U.S. Census Bureau, 2005-2007 American Community Survey, http://factfinder.census.gov/servlet/STTable?_bm=y&-geo_
id=01000US&-qr_name=ACS_2007_3YR_G00_S2601A&-ds_name=ACS_2007_3YR_G00_.
1

American Prosecutor’s Research Institute, Prosecution in the 21st Century: Goals, Objectives, and Performance Measures (Alexandria, VA: American Prosecutor’s Research
Institute, 2004).
2

For More Information…
For more information about the Prosecution and Racial Justice Program, contact Wayne McKenzie at (212) 3763057 or wmckenzie@vera.org.

The Vera Institute of Justice is an independent nonprofit organization that combines expertise in research, demonstration
projects, and technical assistance to help leaders in government and civil society improve the systems people rely on for
safety and justice.
233 Broadway, 12th Floor

Tel: (212) 334-1300

New York, NY 10279

Fax: (212) 941-9407
www.vera.org