Criminalizing Race: Racial Disparities in Plea-Bargaining
Criminalizing Race: Racial Disparities in Plea- Bargaining
Carlos Berdejó 0 1 2
Procedure Commons 0 1 2
the Law 0 1 2
Race Commons 0 1 2
0 Thi s Article is brought to you for free and open access by the Law Journals at Digital Commons @ Boston College Law School. It has been accepted for inclusion in Boston College Law Review by an authorized editor of Digital Commons @ Boston College Law School. For more information , please contact
1 Part of the Civil Rights and Discrimination Commons, Criminal Law Commons , Criminal
2 Loyola Law School
I. RACIAL DISPARITIES IN CRIMINAL CASE OUTCOMES ............................................................ 1194
A. Racial Disparities in Sentencing ...................................................................................... 1194
B. The Critical Role of Prosecutors...................................................................................... 1196
II. SENTENCING IN WISCONSIN AND DESCRIPTION OF THE DATA ............................................. 1200
A. Criminal Justice in Wisconsin.......................................................................................... 1201
B. Description of the Data .................................................................................................... 1204
III. RESULTS .............................................................................................................................. 1207
A. Racial Disparities in Sentencing ...................................................................................... 1207
B. Racial Disparities in Plea-Bargaining............................................................................. 1213
CONCLUSION ............................................................................................................................. 1240
APPENDIX 1 ............................................................................................................................... 1246
APPENDIX 2 ............................................................................................................................... 1247
APPENDIX 3 ............................................................................................................................... 1248
APPENDIX 4 ............................................................................................................................... 1249
Abstract: Most of the empirical research examining racial disparities in the
criminal justice process has focused on its two endpoints—the arrest and
initial charging of defendants and judges’ sentencing decisions. Few studies have
assessed disparities in the steps leading up to a defendant’s conviction, where
various actors make choices that often constrain judges’ ultimtae sentencing
discretion. This Article addresses this gap by examining racial disparities in
the plea-bargaining process, focusing on the period between the initial filing
of charges and the defendant’s conviction. The results presented in this Article
reveal significant racial disparities in this stage of the criminal justice process.
White defendants are twenty-five percent more likely than black defendants to
have their principal initial charge dropped or reduced to a lesesr crime. As a
result, white defendants who face initial felony charges are less likely than
black defendants to be convicted of a felony. Similarly, white defendants i-n
tially charged with misdemeanors are more likely than black defendants either
to be convicted for crimes carrying no possible incarcerat,ioonr not to be
convicted at all. Racial disparities in ple-abargaining outcomes are greater in
cases involving misdemeanors and low-level felonies. In cases involving
esvere felonies, black and white defendants achieve similar outcomes.
Defednants’ criminal histories also play a key role in mediating racial
disparities.Although white defendants with no prior convictions receive charge reductions
more often than black defendants with no prior conviction,swhite and black
defendants with prior convictions are afforded similar treatment. These
patterns in racial disparities suggest that in these “low information” cases, race
perhaps is being used as a proxy for a defendant’s latent criminality and
likelihood to recidivate.
© 2018 Carlos Berdejó . All rights reserved.
* Professor of Law and J. Howard Ziemann Fellow, Loyola Law School, Los Angeles. The
author would like to thank Jeffrey Bellin, Michael Guttentag, Justin Levitt, Alexandra Natapoff,
Elizabeth Pollman, Priscilla Ocen, Sonja Starr, Marcy Strauss, Kimberly West-Faulcon and
participants at the 2017 Conference on Empirical Legal Studies, the 2017 Annual Meeting of the
Canadian Law and Economics Association, and the Loyola Law Schoo—lLos Angeles Faculty
Workshop for extremely valuable comments on an earlier drafAt. ll errors and omissions are the
The treatment of African Americans in the criminal justice process has
been the subject of intense scrutiny in both academic literature and the
popular press. Many have argued that policing practices disproportionately
traget black individuals,1 who are also more likely to be arrested and become
defendants in criminal cases.2 These black defendants are incarcerated more
often and sentenced to longer terms in prison relative to white defendants3.
1 See, e.g., Andrew Gelman et al.,An Analysis of the New York City Police Department’s
“Stop-and-Frisk” Policy in the Context of Claims of Racial Bia,s102 J. AM. STAT. ASS’N 813,
813–14 (2012) (finding that black individuals were stopped more frequently than white
indivuidals even after controlling for a variety of factors);N.Y. CIVIL LIBERTIES UNION, STOP AND FRISK
REPORT 2011, at 7 (May 2012),https://www.nyclu.org/sites/default/files/publications/NYCLU_
2011_Stop-and-Frisk_Report.pdf [http://perma.cc/77W4-UERP] (finding that young minority
males accounted for 41.6% of stops even though they comprise 4.7% of the populatioGn)R;EG
RIDGEWAY, RAND CORP., ANALYSIS OF RACIAL DISPARITIES IN THENEW YORK POLICE
DEPARTMENT’S STOP, QUESTION, AND FRISK PRACTICES 31 (2007), https://www.rand.org/pubs/
technical_reports/TR534.html [http://perma.cc/D4LJ-4NH8] (finding that black suspects are more
likely to be frisked than similarly situated white suspects and that police arrested and used force
against nonwhite suspects more than similarly situated white suspects). In 2013, a federal district
court held that the New York City Police Department’s sto-apnd-frisk policy violated the
Fourteenth Amendment of the U.S. Constitution. See Floyd v. City of New York, 959 F. Supp. 2d 540,
661 (S.D.N.Y. 2013).
2 See, e.g., KAREN E. FIELDS & BARBARA J. FIELDS, RACECRAFT: THE SOUL OF INEQUALITY
IN AMERICAN LIFE 265 (2012) (“While Afro-Americans accounted for 14 percent of drug users in
the United States in 2006, they accounted for 35 percent of those arrested for drug offenses, 53
percent of those convicted, and 45 percent of those in prison for drug offenses aosf 2004.”);
Shima Baradaran, Race, Prediction, and Discretion, 81GEO. WASH. L. REV. 157, 187–90 (2013)
(finding that police arrest black individuals more often for drug crimes than white individu;als)
Katherine Beckett et al.,Drug Use, Drug Possession Arrests, and the Question of Race: Lessons
from Seattle, 52 SOC. PROBS. 419, 426–29 (2005) (comparing drug use data with arrest statistics
and finding disparities between the racial composition of arrestees and users); Christopher
J.yLons & Becky Pettit, Compounded Disadvantage: Race, Incarceration, and Wage Growth, 58 SOC.
PROBS. 257, 257–58 (2011) (noting that black drivers are about three times more likely to be
searched during a traffic stop than white drivers and that black individuals are twice as likely to be
arrested than white individuals);Brad Heath, Racial Gap in U.S. Arrest Rates: ‘Staggering
Dsiparity,’ USA TODAY (Nov. 18, 2014, 5:13 PM)h,ttps://www.usatoday.com/story/news/nation/
2014/11/18/ferguson-black-arrest-rates/19043207 [http://perma.cc/3CCZ-HXPJ] (examining data
filed by police departments with the FBI and concluding that blacks are more likely than others to
be arrested in almost every city regardless of the type of crime).
3 For a review of the existing literature on racial disparities in criminal sentencing sienefra
notes 35–45 and accompanying text. Racial disparities in the imposition of capital punishment are
also well-documented. See, e.g., SAMUEL WALKER ET AL., THE COLOR OF JUSTICE: RACE,
ETHNICITY AND CRIME IN AMERICA 231–45 (2000); David C. Baldus & George WoodworthR,ace
Discrimination and the Death Penalty: An Empirical and Legal Overview, in AMERICA’S EXPER
IMENT WITH CAPITAL PUNISHMENT: REFLECTIONS ON THE PAST, PRESENT, AND FUTURE OF THE
ULTIMATE PENAL SANCTION 501, 513–14 (James R. Acker et al. eds., 2d ed. 2003). For a recent
study, see KATHERINE BECKETT & HEATHER EVANS, THE ROLE OF RACE IN WASHINGTON
STATE CAPITAL SENTENCES, 1981–2012, at 2 (Jan. 2014), https://deathpenaltyinfo.org/documents/
WashRaceStudy2014.pdf [http://perma.cc/K4C6-5WMM] (finding juries were three times more
The end result of these disparites is a pair of dispiriting statistics that are
well-known to many—black males are incarcerated at a rate that is five
times that of white male,s4 and one third of black males can expect to be
imprisoned at some point in their lives.5
Much of the recent empirical work on racial disparities in thecriminal
justice process has centered on its two endpoin—tsthe arrest and initial
charging of individuals and sentencing decisions by judges. These studies
generally find that black individualasre more likely to be arrested and
charged than white individuals6 and that black defendants receive harsher
sentences than white defendants.7 The reasons for these observed disparities
are the subject of vigorous academic debate. 8
Setting that debate asdie, merely focusing on these endpoints ignores
critical steps in the criminal justice process that follow an individual’se-d
tention but precede the sentencing hearing. Critically, judges make their
sentencing decisions conditional on the crime (or crimes)for which the
defendant was convicted, which, together with other factors, determines a
sentencing range.9 The crime of ultimate convictionis the result of a process
controlled by a different set of actors in the sys1t0emS.urprisingly, few
studies have examined the role of race in determining defendants’ outcomes
in these pre-sentencing stages of the criminal justice process.11
This Article fills this gap in the literature by examining disparities in
the plea-bargaining process that precede judges’ sentencnig decisions and
constrain their sentencing discretion. Using data obtained from the
Wisconlikely to impose the death penalty when the defendant was black than in cases involving similarly
situated white defendants).
4 See ASHLEY NELLIS, THE SENTENCING PROJECT, THE COLOR OF JUSTICE: RACIAL AND ETHNIC
DISPARITY IN STATE PRISONS 3 (June
2016),http://www.sentencingproject.org/publications/color-ofjustice-racial-and-ethnic-disparity-in-state-prisons [http://perma.cc/SLW2-2CL7] (explaining
“African Americans are incarcerated in state prisons at a rate that is 5.1 times the imprisonment of
whites”); Keith Humphreys, There’s Been a Big Decline in the Black Incarceration Rate, and Almost
Nobody’s Paying Attention, WASH. POST (Feb. 10, 2016), https://www.washingtonpost.com/news/
[http://perma.cc/J33Z-A65X] (reporting that in 2014, 2,725 of every 100,000 black males were
imprisoned, while for white males the corresponding figure was 465).
5 Lyons & Pettit, supra note 2, at 257–58 (citing data from the Bureau of Justice Statistics that
suggest that “one in three black men can expect to go to prison in their lifetime”).
6 See supra notes 1–2; infra note 60 and accompanying text. Additionally, black defendants
are less likely to be released on bail and more likely to be held in remand prior to triaSle.e infra
notes 289–290 and accompanying text.
7 See infra notes 35–36 and accompanying text.
8 See infra notes 33–64 and accompanying text.
9 See infra notes 46, 82–90 and accompanying text.
10 See infra notes 48–51, 77–79 and accompanying text.
11 See infra notes 58–64 and accompanying text.
sin Circuit Courts, this Article documents striking racial disparities in these
earlier stages.12 White defendants are twenty-five percent more likely than
black defendants to have their most serious initial charge dropped oer- r
duced to a less severe charge (i.e., black defendants are more likely than
white defendants to be convicted of their highest initial charge1)3. As a
result, white defendants who face initial felony charges are approximately
fifteen percent more likely than black defendants to end up being convicted
of a misdemeanor instead.14 In addition, white defendants initially charged
with misdemeanors are approximately seventy-five percent more likely than
black defendants to be convicted for crimes carrying no possible
incarceration, or not to be convicted at all.15
More in-depth analyses reveal two patterns that may shed light as to
the underlying dynamics behind these racial disparities. First, disparities ni
plea-bargaining outcomes appear to be driven by cases in which defendants
have no prior convictions1.6 In cases involving defendants with prior
cnovictions there are no significant racial disparities pi nlea-bargaining
outcomes.17 This pattern suggests that in the absence of evidence of a
defednant’s recidivism risk (for example, when there is no criminal history),
prosecutors may be using race as a proxy for the defendant’s likelihood to
recidivate.18 Second, racial disparities in plea-bargaining outcomes are greater in
cases involving misdemeanors and lo w-level felonies relative to cases
nivolving more severe offenses.19 This second pattern suggests that
prosecutors may be using race as a proxy for a defendant’s latent criminalitfyor(
12 Not only do black defendants receive less favorable treatment in the pl-ebaargaining
process but, consistent with the existing literaturein the field, this Article also finds that
blackedfendants encounter significant disparities at the sentencing stage as well. Black defendants are
50% more likely than white defendants to be incarcerated and receive sentences that are on
average two months longer than white defendants. See infra note 143 and accompanying text.
13 See infra notes 168–169 and accompanying text.
14 See infra notes 187–190 and accompanying text.
15 See infra notes 173–175 and accompanying text. Notably, sentencing disparities are also
greater in the subset of misdemeanor cases. See infra notes 145–146 and accompanying text.
16 See infra note 198 and accompanying text. Although existing studies have controlled for
the prior criminal history of defendants when examining racial disparities in criminal case
outcomes, these have not explored the interaction of this variable with the defendant’s race.See infra
notes 70–101 and accompanying text.
17 See infra note 199 and accompanying text.
18 See Dawinder S. Sidhu, Moneyball Sentencing, 56 B.C. L.REV. 671, 675 (2015)
(criticizing predictors of recidivism that use various defendant characteristics, including race, as illegal
discrimination under the Fourteenth Amendment’s equal protection clause and incorrectly
severing the causal connection between an individual’s conduct and the punishment they receive); infra
notes 260–267 and accompanying text.
19 See infra notes 260–267 and accompanying text.
example, propensity to commit a severe offense in the future) in casesn-i
volving low-level offenses.20
In addition to uncovering racial disparities in the pl-ebaargaining
process, this Article contributes to a pair of current policy debates in the criminal
law arena. First, the evidence presented in this Article sheds light on the
question surrounding the role that the disparate impact theory should play in equal
protection claims under the Fourteenth Amendment. Historically, courts have
been reluctant to accept statistical evidence showing patterns of raciasl- di
crimination to establish an equal protection claim, instead requiring plaintiffs
to show a racially discriminatory intent or purpos2e1. Several scholars have
been critical of the high burden placed on plaintiffs as a reosufltthis
erquirement, which often renders the equal protection clause an ineffectual tool
for combating racial discrimination in the criminal justice process.22 Critics of
this evidentiary standard argue that its focus on discriminatory purpose and
intent ignores the growing evidence on the impact of implicit racial biases on
an individual’s actions, which often contradict an individual’s conscious or
express intentions.23 If subconscious racial biases do contribute to racial
dsi20 See infra notes 262–267 and accompanying text.
21 See, e.g., McCleskey v. Kemp, 481 U.S. 279, 292 (1987) (holding that evidence of the
unjustified racially disparate impact was insufficient to support an equal protection claim against a
state’s capital punishment regime unless adefendant can prove disparate treatment was based on
race in their individual case); Washington v. Davis, 426 U.S. 228, 242 (1976)(holding that
statistical evidence of the unjustified racially disparate impact of an employment policy was inis-uff
cient to mount an equal protectionchallenge to the policy); Farrakhan v. Gregoire,623 F.3d 990,
993 (9th Cir. 2010) (holding that plaintiffs challenging a felon disenfranchisement law under the
Voting Rights Act must “at least” show that a state’s criminal justice system is “infected by
intentional discrimination or that the felon disenfranchisement law was enacted with such intent”)s;ee
also Naomi Murakawa & Katherine Beckett, The Penology of Racial Innocence: The Erasure of
Racism in the Study and Practice of Punishment, 44 L. & SOC’Y REV. 695, 701–08 (2010).
22 See, e.g., MICHELLE ALEXANDER, THE NEW JIM CROW: MASS INCARCERATION IN THE
AGE OF COLORBLINDNESS 135 (2010) (discussing Supreme Court precedent preventing
challenges to racial bias in the criminal justice system and arguing that “[t]he Supreme Court has now
closed the courthouse doors to claims of racial bias at every stage of the criminal justice opcress,
from stops and searches to plea bargaining and sentencing”); Baradarsaunp,ra note 2, at 174
(“The standard of racial intent proves difficutl to establish and allows relatively unchecked police
discretion. . . . As a whole, at all steps in the proces—sstops, searches, plea bargaining, charging,
and sentencing—the Court has made it much more difficult to bring claims of racial bias, resulting
in unchecked discretion by criminal justice actors.”); Sheri Lynn JohnsonU,nconscious Racism
and the Criminal Law, 73 CORNELL L. REV. 1016, 1031 (1988) (“The dissatisfaction with the
discriminatory purpose doctrine has several facets, but a recurring themine the literature is the
difficulty of proving discriminatory purpose.”); David Rudovsky,Litigating Civil Rights Cases to
Reform Racially Biased Criminal Justice Practices, 39 COLUM. HUM. RTS. L. REV. 97, 106 (2007)
(arguing “the Equal Protection Clause requires a showing of intentional racial discriminatione-b
fore a court may consider legal remedies”).
23 See John Tyler Clemons, Blind Injustice: The Supreme Court, Implicit Racial Bias, and the
Racial Disparity in the Criminal Justice Syste,m51 AM. CRIM. L. REV. 689, 692–94 (2014)
(summarizing existing research on implicit racial bias and the perception of criminality based on
parities (for example, by the use of race as a proxy for a defendant’s latent
criminality),24 there would be no constitutional means to address these under
the evidentiary requirements articulated by the cour2t5s.The nature of the
biases documented in thisArticle affirms the need to ree-xamine the role of
evidence showing disparate impact in equal protection claims.
The evidence presented in this Article also touches upon a second
current policy debate—the treatment of misdemeanors in thecriminal justice
process and their role in perpetuating racial disparities. Unlike other studies
that have focused on more serious felony crimes, thAisrticle stresses the
importance of racial disparities in misdemeanor cases, botphlea-in
bargaining and sentencing2.6 Although misdemeanors have traditionally
been overlooked due to the low-level nature of the offenses, sheer volume
of cases, and the shorter sentences involved, recent work has highlighted
their significant role in the criminalization of black male2s7. The fact that
the racial disparities inplea-bargaining and sentencing documented in this
Article are greater in cases involving misdemeanors adds further urgency to
the debate surrounding their regulation and decriminalization.28
The Article proceeds as follows. Part I summarizes the existing ie-v
dence on racial disparities in criminal case outcomes, highlighting the
mliited attention that has been devoted to the plea-bargaining process.29 Part II
provides background information on the criminal justice process in
Wisconsin and describes the dataset and the construction of the variables used in
the analyses.30 The results of these analyses are then presented in Part III.31
an individual’s race); Johnson, supra note 22, at 1027–28 (explaining “[a] burgeoning literature
documents the rise of the ‘aversive’ racist, a person whose ambivalent racial attitudes leads him or
her to deny his or her prejudice and express it indirectcloyv,ertly, and often unconsciously”);
Timothy D. Wilson et al.,A Model of Dual Attitudes, 107PSYCHOL. REV. 101, 102 (2000)
(explaining how individuals can harbor implicit biases distinct from their explicit attitudes)s;ee also
infra notes 43–44 and accompanying text.
24 See infra notes 44–45, 54–57 and accompanying text.
25 See Arthur H. Garrison, Disproportionate Incarceration of African Americans: What
History and the First Decade of Twenty-First Century Have Brought, 11 J. INST. JUST. & INT’L STUD.
87, 104 (2011) (“Each decision point of the criminal justice system: arrest by law enforcement;
arraignment, release, and pre-adjudicatory hearings; pre-trial jail and prison custody; adjudication
and sentencing; probation and community supervision; and parole decisions are all exercised with
various levels of discretion and subject to covert, overt, and unconscious biases.”); Johnson,supra
note 22, at 1019 (“The phenomenon of unconscious racism challenges both Powell’s fourteenth
and his eighth amendment analyses. The concept of purposeful discrimination, or at least its
terminology, does not mesh well with unconscious race discrimination.”).
26 See infra notes 145–146 and accompanying text.
27 See infra notes 177–186, 262–267 and accompanying text.
28 See infra notes 183–186 and accompanying text.
29 See infra notes 33–64 and accompanying text.
30 See infra notes 65–116 and accompanying text.
31 See infra notes 117–276 and accompanying text.
The policy implications of these results and avenues for future research are
discussed in the Conclusion.32
I. RACIAL DISPARITIES IN CRIMINAL CASE OUTCOMES
This Part discusses the existing empirical and theoretical work
examining racial disparities in the criminal justice processS.ection A begins with
an overview of the empirical literature that hsa identified significant racial
disparities in the imprisonment of white and black defendant,sfollowed by
a review of the theoretical work that has attempted to explain these
dispairties.33 Section B describes the critical role of prosecutorial discretionand
the plea-bargaining process in determining criminal case outcomes, hhi-g
lighting the limited attention that has been devoted in the empirical litae-r
ture to this particular area.34
A. Racial Disparities in Sentencing
Studies examining criminal case outcomes in federal courts have
identified substantial racial disparities in judg’esentencing decisions—
black defendants are incarcerated more frequently and receive longer
sentences than white defendants3.5 Although fewer studies have analyzed
sentencing disparities at the state level, the evidence suggests that racial
disparities in sentencing outcomes also exist at the state level.36
Various factors can explain these racial disparitiesF.or example, the
severity of the conviction offense and the defendant’s criminal history are
32 See infra notes 277–303 and accompanying text.
33 See infra notes 35–45 and accompanying text.
34 See infra notes 46–64 and accompanying text.
35 See, e.g., David B. Mustard,Racial, Ethnic, and Gender Disparities in Sentencing:
Evidence from the U.S. Federal Cour,ts 44 J.L. & ECON. 285, 306 (2001) (finding that black and
Hispanic defendants receive substantially longer sentences than white defendants and are also
more likely to be incarcerated); Max Schanzenbach, Racial and Sex Disparities in Prison
Sentences: The Effect of Distri-cLtevel Judicial Demographics, 34 J. LEGAL STUD. 57, 72–73 (2005)
(finding that black defendants in federal criminal cases receive sentences that are 2.9 months
higher than white defendants, a difference that represents 6% of the average sentence of 48.2
months); Darrel Steffensmeier& Stephen Demuth, Ethnicity and Sentencing Outcomes in U.S.
Federal Courts: Who Is Punished More Harshly?, 65 AM. SOC. REV. 705, 716 (2000) [hereinafter
Steffensmeier & Demuth, Outcomes] (documenting similar sentencing disparities).
36 See, e.g., David S. Abrams et al.,Do Judges Vary in Their Treatment of Race,?41 J.
LEGAL STUD. 347, 356 (2012) f(inding that black defendants in Chicago, Illinois receive longer
sentences and are 30% more likely to be incarcerated than white defendants); Darrel Steffensmeier
& Stephen Demuth, Ethnicity and Judges’ Sentencing Decisions: Hispanic-Black-White
Comparisons, 39 CRIMINOLOGY 145, 160 (2001) [hereinafter Steffensmeier & Demuth,Decisions]
(finding that white defendants in Pennsylvania are less likely to be incarcerated than black and
Hispanic defendants, and also receive shorter sentences).
associated with higher rates of incarceration and longer prison sentence3s7.
Thus, if black defendants have lengthier prior records and are convicted of
offenses that are more severe than white defendants, then one would expect
the former to receive, on average, longer sentences.38 Other defendant
characteristics, such as age39 and gender,40 which impact sentencing
determinations could also be correlated witha defendant’s race. Nevertheless, even
after controlling for these defendant and crime characteristics, a significant
black-white sentencing gap remains.41
Another explanation for the observed racial disparities in sentencing is
that judges, due to time and information constraints, may be using a
defendant’s race (an observable attribute) as a proxy for the defendant’s inherent
criminality (an unobservable attribute).42 Judges must make their sentencing
37 See Mustard, supra note 35, at 306; Steffensmeier & Demuth,Decisions, supra note 36, at
38 See Steffensmeier & Demuth, Outcomes, supra note 35, at 716 (finding that black
defendants are convicted of more severe offenses and have lengthier prior records than white
39 Existing work suggests that younger defendants receive harsher punishment than older
defendants. See, e.g., Mustard, supra note 35, at 309 (finding that younger defendants receive on
average higher sentences); Steffensmeier & Demuth, Decisions, supra note 36, at 161 (finding that
older defendants are less likely to be incarcerated and receive longer sentences than youngere-d
40 Prior studies have found that female defendants receive on average shorter sentences than
their male counterparts. See, e.g., Mustard, supra note 35, at 306 (finding that female defendants
are more likely to be assigned no prison term than male defendantsS)c;hanzenbach, supra note
35, at 84 (finding that female defendants on average receive sentences that are 5.4 months lower
than those received by males in federal criminal cases); Darrel Steffensmeier et aGl.,ender and
Imprisonment Decisions, 31 CRIMINOLOGY 411, 423 (1993) (finding that male defendants are
more likely to be incarcerated and receive lengthier sentences than female defendants in criminal
cases in Pennsylvania).
41 See supra notes 35–36 and accompanying text. Unobservable crime characteristics may
also be driving these sentencing disparities. The heinousness and other aspects of a crime, which
may be observed by a judge but not a researcher, may affect sentencing decisionsS. ee
Schanzenbach, supra note 35, at 63. However, for this to explain racial disparities in sentencing, one would
need the heinousness of a crime to correlate with the race of the criminal (i.e., black criminals
would need to be more likely to commit a given crime in a more heinous manner than white
42 See Celesta A. Albonetti, An Integration of Theories to Explain Judicial Discreti,on38
SOC. PROBS. 247, 250 (1991) [hereinafter Albonetti,Integration] (“Using defendant
charactersitics, circumstances of the crime, and case processing outcomes, judges assess the defendant’s
disposition toward future criminal activity. . . Discrimination and disparity in sentencing die-c
sions . . . may be the product of judicial attempts to achieve a ‘bounded rationality’ in sentencing
by relying on stereotypical images of which defendant is most likely to recidivate.”); Celesta A.
Albonetti, The Joint Conditioning Effect of Defendant’s Gender and Ethnicity on Length omf- I
prisonment Under the Federal Sentencing Guidelines for Drug Trafficking/Manufacturingf- O
fenders, 6 J. GENDER, RACE & JUST. 39, 42 (2002) (“From the uncertainty avoidance/causal t-a
tribution perspective, the defendant’s gender and ethnicity are salient to attributions of an enduring
predisposition to criminal activity and dangerousness. As such, these defendant characteristics
decision without full knowledge of the defendant’s likelihood of recidivt-a
ing and the danger that the defendant represents to the comumnity, factors
that arguably help determine the optimal sentence in a criminal ca4s3e.In
this setting, implicit biases can lead judges to make decisions that
systematically discriminate against defendants of a given race if they ascribe certain
characteristics to members of that group.44 If judges perceive black
defendants as being more dangerous and more likely to recidivate than whitee-d
fendants, then judges may (consciously or subconsciously) punisbhlack
defendants more severely than similar white defendants.45
B. The Critical Role of Prosecutors
Judges’ sentencing discretion is constrained in a number of ways.
Criminal statutes often prescribe a minimum and maximum sentence for a
particular crime, which are set by the legislature when enacting the law that
influence judicial sentencing decisions.”); Ian Ayres & Peter SiegelmanR,ace and Gender
Dsicrimination in Bargaining for a New C, ar85 AM. ECON. REV. 304, 317 (1995) (explaining
“‘[s]tatistical discrimination’ is based not on a psychological distaste for associating with blacks
or women, but rather on sellers’ use of observable variables (such as race or gender) to makne-i
ferences about a relevant but unobservable variable”); Sara Steen et aIlm.,ages of Danger and
Culpability: Racial Stereotyping, Case Processing, and Criminal Sentencin,g43 CRIMINOLOGY
435, 463 (2005) (arguing “because they lack complete information about individual cases, di-ec
sion makers form causal attributions for offending and assess dangerousness and culpability by
referencing stereotypes”). Theoretically, this model is closely related to the concept of statistical
discrimination. See Kenneth J. Arrow, What Has Economics to Say About Racial Discrimination?,
J. ECON. PERSP., Spring 1998, at 91, 96–97 (describing the statistical discrimination model).
43 See Albonetti, Integration, supra note 42, at 250 (explaining “uncertainty surrounding the
sentencing decision arises from an inability to predict accurately future criminal behavior”).
44 See, e.g., id. at 249–50 (“[J]udges would attempt to manage uncertainty in thesentencing
decision by developing ‘patterned responses’ that are themselves the product of an attribution
process influenced by causal judgments . . These attributions provide a basis for arriving ata-r
tional decision in a domain of responsibility characterized by uncertainty.”;) Jeffrey J. Rachlinski
et al., Does Unconscious Racial Bias Affect Trial Judges?, 84NOTRE DAME L. REV. 1195, 1197
(2009) (arguing that judges “hold implicit racial biases” and that such biases “can influence their
judgment”). For an overview of the possible sources of implicit biases, see Jerry Kang et
alI.,mplicit Bias in the Courtroom, 59 UCLA L. REV. 1124, 1128–35 (2012); Anthony G. Greenwald &
Linda Hamilton Krieger, Implicit Bias: Scientific Foundations, 94 CALIF. L. REV. 945, 952–957
(2006); see also Christine Jolls & Cass R. Sunstein,The Law of Implicit Bias, 94 CALIF. L. REV.
969, 969–70 (2006) (providing examples of both explicit and implicit bias).
45 Judges may impose higher sentences on the defendants they perceive baesing riskier to
incapacitate them (i.e., prevent them from committing additional crimes) and to enhance the
deterrent effect of criminal sentencing.See Albonetti, Integration, supra note 42, at 258 (explaining
“[i]ncreases in sentence severity produced by the race variable support the causal attribution and
uncertainty avoidance hypothesis linking black defendants with attributions of a high risk of future
criminal behavior and judicial use of discretion as a means to deal with administrative concerns
for reducing such risk”); Baradaran,supra note 2, at 176–77 (explaining that criminal justice
actors often use group-based factors, rather than individual ones, to make bail, sentencing and parole
decisions, which may lead to the inappropriate consideration of race in those decisions).
criminalizes a given actitivity.Additionally, various states and the federal
government have sought to control the exercise of judicial discretion by
adopting sentencing guidelines, which generally provide a set of re-com
mended sentencing ranges which are determined by an offense score (a
function of the conviction crime) and the defendant ’s criminal history.46
The applicable sentencing guidelines range and statutory minimum and
maximum sentences are ultimately determined by theconviction crime.
Except in cases adjudicated in bench trial, the judge has no direct control over
the ultimate crime of conviction, which itself is, in the great majority of cases,
a result of the ple-abargaining agreement between the defendant’s counsel
and the prosecuting attorney.47 Not surprisingly, prosecutors are considered to
be the most influential players in the criminal justice process.48
Prosecutorial discretion in the -bpalregaaining process plsayan
important role in determining the conviction crime. Serious charges that were
initially filed against a defendant may be reduced to less serious onanesd
concurrent charges involving less serious crimes may be dropped altogether.49
As a result of this process, charges involving felony crimes may be reduced to
misdemeanors, or all charges carrying a possible incarceration term may be
dropped or reduced to charges that carry no possible jail or prison t5i0me.
Furthermore, prosecutors also enjoy ample discretion in idnegcidwhich
46 See, e.g., U.S. SENTENCING COMM’N, FEDERAL SENTENCING: THE BASICS 1–7 (Aug.
https//:www.ussc.gov/sites/default/files/pdf/research-and-publications/research-projects-andsurveys/miscellaneous/201510_fed-sentencing-basics.pdf [https://perma.cc/HZA3-HKH3]. Initially,
the federal sentencing guidelines were mandatory in the sense that judges wishing to depart from
the guidelines (i.e., impose a sentence above or below the guidelines range) had to justify their
decision orally or in writinSge.e Schanzenbach, supra note 35, at 60 (describing federal
sentencing guidelines). After the Supreme Court held that the federal sentencing guidelines were
unconstitutional, the sentencing guidelines began serving an advisory role, as one of the factors
judges consider in determining a sentence. See United States v. Booker, 543 U.S. 220, 245 (2005);
Joshua B. Fischman & Max M. Schanzenbach,Racial Disparities Under the Federal Sentencing
Guidelines: The Role of Judicial Discretion and Mandatory Minimu m,s9 J. EMPIRICAL LEGAL
STUD. 729, 733–34 (2012).
47 See U.S. SENTENCING COMM’N, supra note 46, at 5 (“Over 95 percent of federal
defednants convicted of a felony or Class A misdemeanor. . are adjudicated guilty based on a guilty
plea rather than a verdict at a trial.”).
48 See ANGELA J. DAVIS, ARBITRARY JUSTICE: THE POWER OF THE AMERICAN PROSECUTOR
5 (2007); M. Marit Rehavi & Sonja B. Starr, Racial Disparity in Federal Criminal Sentences, 122
J. POL. ECON. 1320, 1326 (2014) (explaining “[l]egal scholars, judges and practitioners broadly
agree that prosecutorial decisions play a dominant role in determining sentences”); Lauren O’Neill
Shermer & Brian D. Johnson,Criminal Prosecutions: Examining Prosecutorial Discretion and
Charge Reductions in U.S. Federal District Courts, 27 JUST. Q. 394, 398 (2010).
49 See Shermer & Johnson, supra note 48, at 395; Ronald F. Wright & Rodney L. Engen,
Charge Movement and Theories of Prosecutors , 91 MARQ. L. REV. 9, 9–10 (2007).
50 See Shermer & Johnson, supra note 48, at 395.
initial charges to file, a choice that defines the starting point of any- plea
In exercising their discretion, prosecutors may, like judges whenm-i
posing a sentence, be subject to implicit biases that adversely impact certain
types of defendant s.52 For example, prosecutors may consciously or
bsuconsciously perceive race as a proxy for recidivism or latent criminalit5y3.
With limited information available and facing time and resource constraints,
prosecutors may have to relyon such heuristics in their decisionmaking.54
And if that is the case, we would expect prosecutors to be on average more
lenient on white defendants relative to black defendants, for example by
agreeing to reduce the top charges faced by white defendants more often.55
Under this model of prosecutorial decision-making, race should play a
smaller role when prosecutors have access to other salient and easily
available proxies about a defendant’s dangerousness. That is, one would expect
the gap in charge reduction rates between white and black defendants to be
smaller when the defendants share a characteristic that is associated with
recidivism, such as a prior criminal record5.6 Black defendants thus receive
less “personalized” or “individualized” treatmet—ni.e., the differences in
charge reductions between black defendants with and without criminal
records will be smaller than the difference in charge reductions between white
defendants with and without criminal records.57
51 See id.
52 See id. at 402 (“Prosecutors, like other organizational actors, are faced with uncertainty that
may lead them to develop decisionm-aking schema that incorporate past practices and reflect the
subtle influences of social and cultural stereotypes in society.”).
53 See id. at 403 (“[P]rosecutors are likely to develop ‘perceptual shorthands’ that tie
attributions of dangerousness to the ascriptive characteristics of offenders and their victi m;s.”)
supra note 44–46 and accompanying text.
54 See Faigman et al., supra note 44, at 1141–42 (arguing that that prosecutors are likely to be
subject to implicit biases in the discharge of their duties given the fact that they have “wide
discretion” and have to make “quick decisions with little accountability”); Robert J. Smith & Justin D.
Levinson, The Impact of Implicit Racial Bias on the Exercise of Prosecutorial Discretion, 35SE
ATTLE U. L. REV. 795, 796–98 (2012) (noting that prosecutors use determinations about the
dnager posed by an individual to society when exercising their discretion).
55 See infra notes 260–267 and accompanying text.
56 See Steen et al., supra note 42, at 460–61 (finding that “the likelihood of incarceration is
virtually certain for bothblack and white offenders who fit the stereotype of a dangerous drug
offender, but in the less-serious categories . . . judges are less likely to incarcerate white offenders
than their black counterparts” and interpreting these findings “to mean that dseicoin-makers are
more likely to define low-level black offenders as a threat to public safety, and therefore deserving
of incarceration, than similarly situated white offenders”).
57 See id. at 461 (explaining “decision makers do not appear to make sharp distinctionse-b
tween the most ‘dangerous’ black offenders and most other black offenders. . . Thus, a principal
effect of minority status may be to produce less individualized, more homogenous decision
Surprisingly, empirical work analyzing the link between prosecutorial
discretion and racial disparitieiss limited, at least compared to work
analyzing judicial behavior and racial disparities in sentencin5g8. Moreover,
the conclusions reached by these studies have often been instceonnt.s5i9
Recent studies have focused on prosecutors’ initial charging decisions and the
importance of charges involving mandatory minimum sentences in ultimately
generating racial disparities in sentencing outcom60esS.tudies that have
examined disparities in charge reductions and dismissals have mostly focused
on a specific subset of crimes and have relied on a relatively low number of
observations.61 Although some of these studies find that the race of the
defendant has no effect on the exercise of prosecutorial discretion in reducing
58 See BESIKI KUTATELADZE ET AL., VERA INST. JUST., DO RACE AND ETHNICITY MATTER IN
PROSECUTION? A REVIEW OF EMPIRICAL STUDIES 1 (June 2012), https://www.vera.org/publications/
do-race-and-ethnicity-matter-in-prosecution-a-review-of-empirical-studies (follow the “PDF”
hyperlink) [https://perma.cc/NXZ7-S9K4] (explaining that “[r]elative to the attention that police
and the courts have received from researchers analyzing disproportionate minority contact with
the criminal justice system, there has been little study of prosecution”); Shermer & Johnson, supra
note 48, at 395 (noting the scant literature in the area)R;ehavi & Starr, supra note 48, at 1326
(noting that “prior empirical studies of racial and other demographic disparities in sentencing have
considered judicial sentencing decision only in isolation from the prosecutorial choices that ep-r
ceded them”); Wayne S. McKenzie, Dir., Prosecution& Racial Justice Program, Vera Inst.
Justice, Racial Disparities in the Criminal Justice System, Testimony Before the House Judiciary
Committee on Crime, Terrorism and Homeland Security 6 (Oct. 29, 2009) (transcript available on
the House of Representatives Judiciary Committee website) (explaining the Prosecution & Racial
Justice Program’s mission to collect and analyze data for prosecutors similar to already- well
established analysis of other actors in the justice system).
59 See KUTATELADZE ET AL., supra note 58, at 7 (“While a review of the 34 studies discussed
here suggests that defendants’ and victims’ race affect prosecutorial decisions, the findings are
complex and somewhat difficult to interpret. Overall, research finds that the effect of race and
ethnicity on prosecutorial decision making is inconsistent, and it is not always blacks or Latinos
and Latinas who are treated more punitively.”).
60 See, e.g., Fischman & Schanzenbach,supra note 46, at 757 (finding that racial disparities
may be exacerbated by the existence of statutory minimum sentences); Rehavi & Starr, supra note
48, at 1335–36, 1344 (finding that 12.4% of black arrestees face an initial charge with a
mandaotry minimum while 7.5% of the white arrestees do and that prosecutors’ decision to bring a charge
carrying a mandatory minimum sentence explains more than half of the b-wlahcikte sentence
disparities not explained by other case or defendant characteristics). Some earlier studies did not
find these racial disparities in initial charging decisions by prosecutors. See KUTATELADZE ET AL.,
supra note 58, at 7–11 (summarizing existing studies examining racial disparities in prosecutors’
initial charging decisions); Jeffrey T. Ulmer et al.,Prosecutorial Discretion and the Imposition of
Mandatory Minimum Sentences, 44 J. RES. CRIME & DELINQ. 427, 440–46 (2007) (finding no
differences in prosecutors’ decision to bring charges with a mandatory minimum sentence against
black and white drug offenders in Pennsylvania).
61 See KUTATELADZE ET AL., supra note 58, at 12–14 (summarizing existing studies
examining racial disparities in dismissals and charge reductions); Shermer & Johnsons,upra note 48, at
400 (describing the methodological limitations of existing studies examining racial disparities in
dismissals and charge reductions).
or dismissing charges62 or that prosecutorial discretion may be used in a
manner favorable to black offenders6,3 others conclude that there are racial
disparities in the plea-bargaining process disfavoring black defendants.64
II. SENTENCING IN WISCONSIN AND DESCRIPTION OF THE DATA
Section A of this Part presents an overview of the legal framework
governing the criminal justice process in Wisconsin, focusing on the
discretion afforded to district attorneys and judges, followed by a brief description
of recent incarceration trends in the state and its prison
populatio6n5.Sec62 See, e.g., Celesta A. Albonetti, Charge Reduction: An Analysis of Prosecutorial Discretion in
Burglary and Robbery Cases, 8 J. QUANTITATIVE CRIMINOLOGY 317, 323, 328 (1992) (analyzing
400 burglary and robbery cases in Jacksonville, Florida, and finding no evidence of racial or gender
disparities in prosecutors’ decision to reduce initial charges); Rodney Kingsnorth et al.A,dult Sexual
Assault: The Role of Racial/Ethnic Composition in Prosecution and Sentencing, 2J6. CRIM. JUST.
359, 362–365 (1998) (finding no racial disparities in the prosecution and sentencing of cases in a
sample of 365 sexual assaults in Sacramento County, California); Shermer & Johnson, supra note 48,
at 420–21 (finding race of defendant does not affect likelihood of receiving a reduction in charges);
EMILY OWENS ET AL., EXAMINING RACIAL DISPARITIES IN CRIMINAL CASE OUTCOMES AMONG
INDIGENT DEFENDANTS IN SAN FRANCISCO 9 (2017), http://sfpublicdefender.org/wp-content/
uploads/sites/2/2017/06/quattronefullreport.pdf [http://perma.cc/PY5W-2GDM] (finding that felony
charges filed against white defendants represented by the San Franciso Public Defender’s Office
were more likely to be downgraded (31%) than felony charges filed against black defendants
represented by that same office, but htat such disparities are not statistically significant after
controlling for criminal history and booking charges).
63 See, e.g., Malcolm D. Holmes et al., Determinants of Charge Reductions and Final
Dispositions in Cases of Burglary and Robbery, 24 J. RES. CRIME & DELINQ. 233, 242–45 (1987)
(finding that black defendants accused of burglary and robbery offenses in Delaware County,
Pennslyvania are more likely to receive a charge reduction); Cassia Spohn et alT.,he Impact of Ethnicity
and Gender of Defendants on the Decision to Reject or Dismiss Felony Charge,s 25
CRIMINOLOGY 175, 183–86 (1987) (finding that black-on-white sexual assaults were more likely to be sd-i
missed by prosecutors than white-on-black assaults in a sample of 321 sexual assaults in a
Michigan county); John Wooldredge & Amy ThistlethwaiteB,ilevel Disparities in Court Dispositions
for Intimate Assault, 42 CRIMINOLOGY 417, 437–39 (2004) (finding that black offenders are less
likely to be charged and fully prosecuted relative to white offenders in a study of 2,948 male
arrests for misdemeanor intimate assaults in Cincinnati, Ohio).
64 See, e.g., BESIKI KUTATELADZE ET AL., VERA INST. JUSTICE, RACE AND PROSECUTION IN
MANHATTAN 6 (July 2014), https://www.vera.org/publications/race-and-prosecution-in-manhattan
(follow Research Summary “PDF” hyperlink) [https/:/perma.cc/8NX7-UVTE] (analyzing a
sample of misdemeanor and felony drug cases in New York City and findinthgat “[a]lthough some
evidence emerged that black defendants were less likely to receive an offer of a lower charge than
were similarly situated white defendants, this difference was not statistically significant due to a
relatively small sample size”); Besiki Luka Kutateladze et al., Opening Pandora’s Box: How Does
Defendant Race Influence Plea Bargaining?, 33JUST. Q. 398, 414 (2016) (finding in a sample of
misdemeanor marijuana cases in New York County that black defendants are less likely than
white defendants to be offered a charge reduction); Spohn et al.,supra note 63, at 183–86 (finding
that Hispanic and black males are more likely to be fully prosecuted in a sample of cases from Los
65 See infra notes 72–101 and accompanying text.
tion B describes the dataset and the construction of the variables used in the
empirical analyses presented later in the Article.66
Before proceeding further, it is worth explaining the choice of Wisconsin
as the jurisdiction in which to conduct this study. The most important reason
is the nature of the available data. Wisconsin courts maintain records
thatniclude a comprehensive set of information for each criminaalction, allowing
us to follow a case from the initial filing of charges, through the dismissal or
reduction of charges, and up to adjudication and sentencin6g7. Similar
databases maintained by other states or the federal government are more limited
in their scope, often including only cases which resulted in a conviction and
containing only sentencing information68. Another important consideration
for choosing Wisconsin is its geographical location. Existing evidence
suggests that racial disparities in dismissals are greater in southern states.69
A. Criminal Justice in Wisconsin
This section provides a brief overview of thecriminal justice process
in Wisconsin, which informs the construction of the variables and design of
the analyses described later. The first subsection describes the criminal
justice process in Wisconsin—from the time charges are initially filed in a case
by the prosecutor to the point where the judge imposes a senten7c0eT.he
second subsection provides general information on Wisconsin’s prison
population and existing racial disparities in incarceration rates.71
1. The Criminal Justice Process
Circuit courts have jurisdiction to hear and determine all criminal
cations and proceedings in Wisconsin7.2 With a few exceptions, each county
in the state has its own circuit court, with the number of branches (or
judges)73 varying from circuit to circuit.74 Circuit court judges are elected at the
66 See infra notes 102–116 and accompanying text.
67 For an overview of the data see infra notes 112–116 and accompanying text.
68 See Shermer & Johnson, supra note 48, at 396 (sumarizing existing literature on the topic);
id. at 420 (“The limited empirical attention devoted to prosecutorial discretion is largely the result
of data limitations. Whereas data on judicial sentencing decisions are now readily available,
records on prosecutorial charging behavior remain elusive.”).
69 See Travis W. Franklin, Community Influence on Prosecutorial Dismissals: A Multilevel
Analysis of Case- and County-Level Factors, 38 J. CRIM. JUST. 693, 699–700 (2010).
70 See infra notes 72–90 and accompanying text.
71 See infra notes 91–101 and accompanying text.
72 See WIS. STAT. § 753.03 (2018).
73 See id. § 753.061(
74 See id. § 753.06. Dane County, which has its own circuit court, has seventeen branches (or
judges). See id. § 753.06(5)(a).
circuit court level for a term of six years.75 Similarly, with a few exceptions,
each county in the state also encompasses a prosecutorial unit and elects a
district attorney who serves a term of four years.76
Charging and plea-bargaining decisions are made by the local district
attorney’s office.77 Prosecutorial discretion in this respect is quite broad so
long as the charges are supported by probable cause7.8 Although judges do
review plea bargains, in practice few agreements are rejected7.9 As in other
jurisdictions, the minority of cases which are not resolved via a
pleaagreement are adjudicated at a bench or jury tria8l0. Once a defendant has been
convicted of a particular crime, Wisconsin law provides the presiding judge
ample discretion in choosing the appropriate sentence.81
Wisconsin does not have a set of sentencinugideglines providing a
mandatory or advisory sentencing range based on the crime committed by the
defendant and the defendant’s criminal histor8y2. Also, few crimes carry a
mandatory minimum sentence8.3 Most crimes are classified into classes of
felonies or misdemeanors, with each class carrying its own maximum
penalty.84 Currently, there are nine classes of felony crime8s5 and three classes of
misdemeanors.86 Certain statutory provisions allow for sentencing
enhanecments which increase the possible maximum sentence a judge may impose,
75 See id. § 753.01.
76 See id. § 978.01.
77 See Michael M. O’Hear, Sentencing Policies and Practices in Wisconsin,OXFORD
HANDBOOKS ONLINE 7 (Mar. 2016) (on file with author).
78 See id.
79 See id.
but these are advisory in nature and not ultimately binding.87 If a defendant is
convicted of more than one count that carries a potential sentence, the judge
may impose sentences for each count to be served concurrently or
consecutively.88 Except for crimes involving a Class A felony, the sentencing judge
also has the discretion to impose probation instead of a sentence carrying an
incarceration term.89 Sentences imposed by judges in Wisconsin are
definitive, as defendants are required to serve the full term of their sentences
wihtout the possibility of parole or early release due to good behavior.90
2. Prison Population and Incarceration Rates
Historically, Wisconsin had imprisonment rates that were lower on
average than those of otherstates.91 Nevertheless, starting in the late 1970s,
Wsiconsin’s imprisonment rate began to converge to the national average, a trend
that has been linked to increases in the state’s violent crimes rates and the
associated increase in incarceration rates andlength of sentences.92 This
increase in the prison population has been accompanied by a shift in the
mcoposition of inmates from property offenders to violent and sexual offenders.93
These high imprisonment rates go hand in hand with significant racials-di
parities, as African Americans have consistently been overrepresented in
Wisconsin’s prison population.94 As of 2010, African Americans
representedforty-three percent of the prison population, a figure that dwarfs the group’s
seven percent share of the totla population.95 In fact, Wisconsin has one of the
country’s highest incarceration rates for African American males,96 as well as
one of the highest race-based differentials in incarceration rates.97
These striking racial disparities in imprisonment rates aremostly
driven by Milwaukee County, which has a higher imprisonment rate than the
other counties in the state, as well as the highest proportion of African
Americans.98 According to a 2015 report, Milwaukee was home to one-sixth
of the state population but accounted for about a quarter of prison adsm-i
sions.99 Such rates coupled with the fact that seventy percent of Wisconsin’s
African American population lives in MilwaukeeCounty explains a good
share of the state’s racial disparities in imprisonment rates1.00 Notably, over
fifty percent of Milwaukee’s black males between thirty and fifty years old
are serving, or have at some point served, time in a state prison.101
B. Description of the Data
The first subsection provides an overview of the dataset used to
cnoduct the analyses presented later in thAerticle.102 The second subsection
describes how the different variables were constructed.103
1. Overview of the Database
The data comes from the public records of the Wisconsin Circuit
Courts, which are availableelectronically at the Wisconsin Circuit Court
Access website.104 This site provides the case information entered into the
Consolidated Court Automation Programs (“CCAP”) case management
system by court staff in the Wisconsin Circuit Courts where the files are
ed.105 The focus of this study is those cases labeled as criminal
misdemeanors or criminal felonies,106 which involve offenses committed after
December 31, 1999107 and adjudicated before December 31, 2006.108
The analyses presented later in theArticle restrict the sample to cases
filed and adjudicated in Dane County. Focusing on a single county is
advantageous to the extent that it helps maintain a number of factors relating to
law enforcement and thecriminal justice process (such as district attorney
office and judges) constant. A natural choice would be Milwaukee County,
the largest county of the state by population. Nevertheless, its high
incarceration rates and demographical make-up does not make it an ideal candidate
as it is not representative of thestate.109 Dane County, which includes the
capital city of Madison, is the second most populous county in the state
fater Milwaukee County110 and has a demographic make-up that reflects that
of the state as a whole.111 In addition, the coding of different variables in the
datasets appeared to be more consistent in Dane County relative to
Cases which were transferred to another county or jurisdiction before
adjudication or that deal with the extradition of a defendant to another state
105 Case Search, WIS. COURT SYS., https://www.wicourts.gov/casesearch.htm [http://perma.
cc/L7NE-GGWV]. Some counties started using CCAP during 1991, but Dane county began using
the system in January 1995. See When Wisconsin Counties Began Using CCAP, WIS. COURT SYS.
CIRCUIT COURT ACCESS, https://wcca.wicourts.gov/countyOnCCAP.xsl
[http://perma.cc/9L9DT6RC]. Several counties, including Dane, have loaded their historical cases to the system, making
them available on the Wisconsin Circuit Court Access system.See Help, WIS. COURT SYS. CIRCUIT
COURT ACCESS, https://wcca.wicourts.gov/help.xsl [http://perma.cc/B3NF-ADRB].
106 Frequently Asked Question,s WIS. COURT SYS. CIRCUIT COURT ACCESS, https://wcca.
wicourts.gov/faq.xsl#Faq15 [http://perma.cc/3BSE-PUJ4] (Frequently Asked Question #15: “Could
You Describe What the Different Case Types Mean?”).
107 This restriction is imposed so that only cases adjudicated under the Truth in Sentencing
Law are included. See supra note 90 and accompanying text.
108 The Wisconsin Circuit Court provided all available information in the CCAP system for
all cases resolved on or before December 31, 2006.
109 See supra notes 98–101 and accompanying text.
110 As of 2010, Dane county had a population of 488,073 and Milwaukee county had a popau-l
tion of 947,735. See David Egan-Robertson, Population and Race Data, Wisconsin County
Subdivisions (Minor Civil Divisions), Census 2000 and 2010 Compari,soWnsIS. DEP’T ADMIN. (Mar.
2011), https://doa.wi.gov/DIR/Pop_Race_MCD_2000_2010.pdf [http://perma.cc/SM62-Y4TD]. Dane
County also has the second largest number of cases in the dataset. Of the 704,190 cases in the
dataset involving commissions—on or after January 1, 2000—of felony or misdemeanor offenses,
116,541 were filed in Milwaukee County, while 51,028 were filed in Dane County.
111 According to the 2010 census, 86.2% of the Wisconsin population was white. That number
is considerably lower in Milwaukee, where only 60.6% of the population was white according to
census data. See Egan-Robertson, supra note 110. In Dane county, on the other hand, 84.7% of the
population was white. See id. The other major racial groups in Dane county are Hispanics (5.9%
of the population), African Americans (5.2% of the population), and Asians (4.7% of the popua-l
tion). See id.
nies are ultimately convicted of a misdemeanor (or a lesser crime22)6. The
results confirm that the significant racial disparities in these alternative
outcomes discussed earlier are present even after controlling for varioues- d
fendant and crime characteristics.227 White defendants charged with
misdemeanors are more likely than black defendants to be ultimately convicted of
crimes carrying no potential incarceration hoavre their charges s-di
missed.228 Similarly, white defendants charged with felonies are more likely
than black defendants to be convicted of a misdemeanor or lesser charge2.29
As noted earlier, these disparities are mainly driven by cases involving
edfendants with no prior convictions—if we focus on cases involving
defendants with prior convictions, racial disparities are substantially lower.230
226 These outcome variables are constructed as follows: (a) an indicator variable equal to 1 if
all charges carrying a potential jail or prison sentence were dropped or reduced to charges carrying
no potential incarceration (misdemeanors cases) and (b) an indicator variable equal to 1 if all
feolny charges were dropped or reduced to misdemeanor (or lesser) charges (felony cases).
227 See supra notes 205–208 and accompanying text.
228 See infra note 231 and accompanying text (Table 7, column (
229 These disparities are substantially greater in cases where a defendant is charged with less
severe felonies (i.e., those carrying potential sentences of five years or lessS)e.e infra note 231
and accompanying text (Table 7, columns (6)–(7)).
230 See infra note 231 and accompanying text (Table 7, columns (2)–(3) (misdemeanors) and
Table 7, columns (4)–(5) (felonies)).
231 This Table is permanently available ahttp://www.bc.edu/content/dam/bc1/schools/law/
pdf/law-review-content/BCLR/59-4/berdejo-graphics.pdf [https://perma.cc/ECS5-GP79]. Note:
Robust standard errors in brackets (* significant at 10%; ** significant at 5%; *** significant at
1%). This table presents the marginal effects from the series of probit models described nextT.he
outcome variable in columns (1–)(3) is an indicator variable equal to 1 if all charges carrying a
potential sentence were dropped, amended to a charge carrying no sentence, or dismissed. The
outcome variable in columns (4)–(7) is an indicator variable equal to 1 if all initial felony charges
were dropped, amended to misdemeanor charges, or dismissed. Column (
) includes all
misdemeanor cases, while columns (2) and (3) include misdemeanor cases in which the defendant had
no prior conviction and in which the defendant had at least one prior conviction, respectively.
Columns (4) and (5) include all felony cases in which the defendant had no prior conviction and in
which the defendant had a prior conviction, respectively. Columns (6) and (7) divide felony csaes
based on whether the principal charge carried a potential sentence of 5 years or less or more than 5
years. For a description of the explanatory variables of interest seeinfra note 304 and
accompanying text (Appendix 1: Table 1). All regressions include a set of crime class fixed effects, a set of
crime type fixed effects and a set of year fixed effects.
4. The Role of Prosecuting and Defense Attorneys
The results presented thus far indicate that white defendants are more
likely than black defendants to see their initial charges reduced as part of
the plea-bargaining process2.32 This allows white defendants initially
charged with a felony to avoid a felony conviction more often than black
defendants.233 It also allows white defendants originally charged with a
misdemeanor to avoid being convicted of a crime that carries a potential
imprisonment sentence at a rate higher than black defendant2s3.4 These
results are robust to the inclusion of variables controlling for various
defendant and crime characteristics.235
a. Differences in Prosecuting Attorneys
One possible explanation relates to the identity of the prosecuting
attorney making the decision whetheror not to reduce the principal charge in
a given case. If prosecutors differ in their willingness to agree to a charge
reduction and cases are not allocated randomly across prosecutors, one may
be concerned that the racial disparities documented earlier are the result of
the assignment of cases involving black defendants to prosecutors that are
less likely to agree to reduce charges. If this is true, then the coefficient on
the black indicator variable would be capturing some of these systematic
differences across prosecutors, thus artificially inflating the magnitude of
the observed racial disparities.236
To address these concerns, we can use the identity of the prosecuting
attorney, which the CCAP database provides for all but 211 of the 47,601
cases considered in the analyses presented earlier in this section. Th-irty
three prosecutors appear in at leasotne hundred cases (and in a total of
47,012 cases). Using this information, we can construct thirty-five indicator
variables: (i) one indicator variable for each of thethirty-three prosecutors
that appear in at least one hundred cases; (ii) one indicator variable for
cases involving prosecutors that appeared less than100 times in the dataset;
and (iii) one indicator variable for cases in which the identity of the proe-s
cutor was not available. To verify whether systematic diffecreesn in the
identity of the prosecutor are driving the racial disparities in plea-bargaining
232 See supra notes 165–172 and accompanying text.
233 See supra notes 187–195 and accompanying text.
234 See supra notes 173–186 and accompanying text.
235 See supra notes 210–230 and accompanying text.
236 In addition, cases assigned to the same prosecutor are not necessarily independent of one
another, which can result in artificially deflated standard errors that make results appear
statistically significant when they might not be.
outcomes documented earlier, the specifications presented in Table 5 (which
237 and Table 6 (which divided cases
divided cases along charge severity)
along defendant criminal history)238 can be re-estimated including this set of
thirty-five prosecuting attorney fixed-effects. The results for these
specifications are presented in Table 8.239
237 See supra note 215 and accompanying text (Table 5); see also supra notes 210–221 and
238 See supra note 225 and accompanying text (Table 6); see also supra notes 222–224 and
239 To correct for the fact that cases managed by the same prosecutor are not independent
from one another, standard errors are clustered at the prosecutor level.
240 This Table is permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/law/
pdf/law-review-content/BCLR/59-4/berdejo-graphics.pdf [https://perma.cc/ECS5-GP79]. Note:
Robust standard errors clustered at the prosecuting attorney level in brackets (* significant at 10%;
** significant at 5%; *** significant at 1%). This table presents the amrginal effects from probit
models in which the outcome variable is an indicator variable equal to 1 if the initial top charge in
a case was dropped, amended to a lesser charge or dismissed. Column (
) includes all
misdemeanor cases. Column (2) includes felony cases in which the principal charge carried a potential
sentence of 5 years or less, while column (3) includes felony cases in which the principal charge
carried a potential sentence of more than 5 years. Columns (4) and (5) include all cases in which
the defendant had no prior convictions and in which the defendant had at least one prior cocn-vi
tion, respectively. For a description of the explanatory variables of interest seeinfra note 304 and
accompanying text (Appendix 1: Table 1). All regressions include a set of crime class fixed
effects, a set of crime type fixed effects, a set of year fixed effects and a set of prosecuting attorney
fixed effects (see supra notes 237–239 and accompanying text).
In misdemeanor cases, black defendants are %8.3less likely than
white defendants to receive a charge reduction, which is virtually identical
to the difference documented in the specifications excluding the indicator
variables (8.2%).241 In cases involving low-level felonies, black defendants
are 10.8% less likely than white defendants to receive a charge reduction,
nearly identical to the difference documented earlier in earlier analyses
(10.9%).242 In cases involving high level felonies, black defendants are
2.1% less likely than white defendants to receive a charge reduction, a rate
differential similar to that documented earlier (1%.8).243 Results are also
consistent when dividing cases based on the criminal history of the
defendant. Black defendants with no prior convictions are 14.3%less likely than
white defendants to receive a charge reduction, which is identical to the
earlier estimate.244 Finally, including these prosecutor fixed effects confirms
that differences in charge reduction rates between white and black
defednants that have at least one prior conviction are smaller (1.9%).245
b. Differences in Attorney Quality
Another possible explanation for the racial disparities -in
bargaining outcomes documented earlier is that there are systematic
diffreences in the quality of the legal representation ofblack and white
defendants. If more qualified and experienced attorneys negotiate superior terms
241 See infra note 253 and accompanying text (Table 9, column (
)); supra note 215 and
accompanying text (Table 5, column (2)).
242 See infra note 253 and accompanying text (Table 9, column (2)); supra note 215 and
accompanying text (Table 5, column (4)).
243 See infra note 253 and accompanying text (Table 9, column (3)); supra note 215 and
accompanying text (Table 5, column (5)).
244 See infra note 253 and accompanying text (Table 9, column (4)); supra note 225 and
accompanying text (Table 6, column (
245 See infra note 253 and accompanying text (Table 9, column (5)); supra note 225 and
accompanying text (Table 6, column (2)).
for their clients during the pl-ebaargaining process and white defendants
have on average better legal representation than black defendants, then one
would expect white defendants to have a higher rate of charge reductions
than black defendants. In that case, the coefficient on the black indicator
variable would be capturing differences in attorney quality, an omitted
The quality of legal representation for black defendants could
different from that of white defendants if the formerlargely have to rely on
public defenders or legal defense funds and the lattegrenerally are able to
afford private defense attorneys2.47 For this to be true, onewould need to
assume that privately retained attorneys outperform attorneys from
public defenders’s office or local legal defense fundNse.vertheless, the
existing evidence does not necessarily support this assumption. Although
more experienced counsel seem to obtain better sentencing outcomes for
their clients,248 studies that have examined how public defenders perform
relative to private counsel have reached inconsistent conclusions.249
The dataset allows us to partially rule out the possibility that
systematic differences in legal representation between white and black defendants
are driving the observed disparities. The name of the defense attorney is
246 If a relevant factor is omitted from a multivariate regression analysis, the econometric model
may compensate for this omission by giving more weight (i.e., ascribing a higher explanatory value)
to a variable that is included in the analysis and correlated with the missing variable. This is known
as the omitted variable bias problem. For a discussion of this issue WseeOOLDRIDGE, supra note
142, at 96–97.
247 See Schanzenbach, supra note 35 at 63 (explaining “both income and assets are deter
minants of quality of legal counsel and hence will play a role in sentencing, offense level
determination, and probability of a downward departure”).
248 See David S. Abrams & Albert H. YoonT,he Luck of the Draw: Using Random Case
Assignment to Investigate Attorney Ability, 74 U. CHI. L. REV. 1145, 1172–73 (2007) (finding that
defendants represented by more experienced public defenders are less likely to plead to the most
serious charge, are less likely to be incarcerated, and are more likely to receive lower sentences).
249 See Nadine Frederique et al., What Is the State of Empirical Research on Indigent Defense
Nationwide? A Brief Overview and Suggestions for Future Research, 78 ALB. L. REV. 1317, 1327
(2015) (“Several studies found that private attorneys are able to obtain better sentencing outcomes
for their clients when compared to public counsel. However, a competing line of research noted
that public defender systems are able to achieve very similar outcomes to private attorneys.”).
Compare Morris B. Hoffman et al.,An Empirical Study of Public DefenderEffectiveness: Self
Selection by the “Marginally Indigent,”3 OHIO ST. J. CRIM. L. 223, 230 (2005) (finding that
“public defenders achieved worse sentence outcomes for their clients than private defense cno-u
sel”), with Abrams & Yoon, supra note 248, at 1172–73 (finding that experienced public
defenders are more likely to achieve better results for their clients), Michael A. Roach,Indigent Defense
Counsel, Attorney Quality, and Defendant Outcome,s 16 AM. L. & ECON. REV. 577, 578 (2014)
(finding that “assigned counsel generate significantly less favorable defendant outcomes than
public defenders”), and Radha Iyengar, An Analysis of the Performance ofFederal Indigent
Defense Counsel 28 (Nat’l Bureau of Econ. Research, Working Paper No. 13187, 2007) (finding that
public defenders outperform assigned private attorneys in federal criminal cases).
available for 42,951 of the 47,601 cases considered in the analyses in this
section.250 Ninety defense attorneys appear in at least one hundred cases in
the database and in the aggregate, comprise 36,270total observations.
Using this information, one can construct a series of (i)ninety indicator
variables for each of theseninety defense attorneys; (ii) one indicator variable
for the 6,681 cases involving attorneys that appear less than one hundred
times; and (iii) one indicator variable for the 5,417 cases for which no
defense attorney information was available. To verify whetherthe identity of
the defense attorney is driving the racial disparities in ple-abargaining
outcomes documented earlier, we can r-eestimate the specifications presented
in Table 5 (which divided cases along charge seve25r1itya)nd Table 6
(which divided cases along defendant criminal history),252 but including the
set of ninety-two defense attorney indicator variables described above. The
results for these specifications are presented in Table 9 below.
250 For cases involving more than one defense attorney, I selected the attorney who was active
(i.e., had not withdrawn) as of the date in which the case was resolved. When more than one
attorney was active as of this date, I selected the attorney based on alphabetical ordering. In 2,281 of
these 42,951 cases, only the attorney’s organization was available (the Legal Defense Program in
1,850 cases and State Public Defender in 431 cases). I treated these organizations as the attorney
251 See supra note 215 and accompanying text (Table 5); see also supra notes 210–221 and
252 See supra note 225 and accompanying text (Table 6); see also supra notes 222–224 and
253 This Table is permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/law/
pdf/law-review-content/BCLR/59-4/berdejo-graphics.pdf [https://perma.cc/ECS5-GP79]. Note:
Robust standard errors in brackets (* significant at 10%; ** significant at 5%; *** significant at
1%). This table presents the marginal effects from probit models in which the outcome variable is
an indicator variable equal to 1 if the initial top charge in a case was dropped, amended to a lesser
charge or dismissed. Column (
) includes all misdemeanor cases. Column (2) includes felony
cases in which the principal charge carried a potential sentence of 5 years or less, while column
(3) includes felony cases in which the principal charge carried a potential sentence of more than 5
years. Columns (4) and (5) include all cases in which the defendant had no prior convictions and
in which the defendant had at least one prioorncviction, respectively. For a description of the
explanatory variables of interest see infra note 304 and accompanying text (Appendix 1: Table 1).
All regressions include a set of crime class fixed effects, a set of crime type fixed effects, a set of
year fixed effects and a set of defense attorney fixed effects (see supra notes 250–252 and
In misdemeanor cases, black defendants are %5.3less likely than
white defendants to receive a charge reduction2.54 Although this difference
is slightly lower than the difference when excluding the indicator variables
(8.3%),255 it is still substantial, representing 13.35% of the average charge
reduction rate in the sample. In cases involving lo-lwevel felonies, black
defendants are 8.7% less likely than white defendants to receive a charge
reduction, a difference representing 19.33% of the average rate in the
sample.256 Consistent with the analyses presented earlier, there are no
substantial differences in charge reductionrates between black and white
defednants initially charged with high level felonies.257
The results highlighting the mediating role of defendants’ criminal
history in determining disparities in plea-bargaining outcomes are also robust
to the inclusion of defense attorney fixed effects. Black defendants with no
prior convictions are 9.9% less likely than white defendants to receive a
charge reduction, a difference representing 16.5%6 of the average charge
reduction rate in the sample.258 Additionally, including these attorney fixed
effects confirms our earlier conclusion that there are no substantial dirff-e
ences in charge reduction rates between white and black defendants that
have at least one prior conviction.259
5. Race as a Proxy for Criminality
A theory that seeks to explain the racial disparities in ple-abargaining
outcomes documented earlier needs to account not just for differences in the
average charge reduction rates for black and white defendants, but also for
the cross-sectional variation in racial disaprities documented earlier. For
example, if race appears to matter inplea-bargaining because it serves as a
proxy for some omitted variable, then the inclusion of suchvariable would
need to explain (i) why black defendants fare worse than white defendanst
when they have no prior conviction,s but fare as well as white defendants
254 See supra note 253 and accompanying text (Table 9, column (
255 See supra note 215 and accompanying text (Table 5, column (2)).
256 See supra note 253 and accompanying text (Table 9, column (2)). This difference is
slightly smaller than that documented earlier when the specifications did not include attorney fixed
effects (i.e., 8.4%). See supra note 215 and accompanying text (Table 5, column (4)).
257 See supra note 253 and accompanying text (Table 9, columns (3))s;upra note 215 and
accompanying text (Table 5, column (5)).
258 See supra note 253 and accompanying text (Table 9, column (4)). The magnitude of this
coefficient is slightly lower than our earlier estimate, 15.27% which represented 28.6% of the
average outcomes in the sample. See supra note 225 and accompanying text (Table 6, column (
259 See supra note 253 and accompanying text T(able 9, column (5); supra note 225 and
accompanying text (Table 6, column (2)).
when they have a prior convicti o,2n60 and (ii) why black defendants fare
worse than white defendants when they commit less severe crimes, but are
treated more equally when they commit more severe crimes. 261
An explanation consistent with these two patterns is that the race of the
defendant is subconsciously used by prosecutors as a proxy for a
defendant’s recidivism potential and dangerousnes2s6.2 Even though prosecutors
may have more information than judges about the specifics of the crime,
prosecutors cannot easily observe a defendant’s true type (for example,
likelihood to recidivate and latent criminality) and may not have the time or
resources to assess this thoroughly on acase-by-case basis before making
charging and plea-bargaining decision.263 The use of race as a heuristic
would explain the cross-sectional differences in racial disparities identified
For example, defendants with a prior conviction have already proved
their likelihood to recidivate, rendering heuristics less useful in the risk
sasessment of that defendant.264 This could explain why disparities arise in the
sample of defendants with no prior convictions (where race is perhaps being
used as a proxy for recidivism) but not in the sample of defendants with
prior convictions (where such heuristics are less usefu2l6)5. Similarly, the
severity of the crime can affect the value of race as a proxy for a dedf-en
ant’s latent criminality and dangerousness. One who commits a serious
felony crime has already proved thatthey are dangerous. Thus, in the sample
of cases involving more severe crimes, race would be less valuable as a
proxy for a person’s latent criminality. Nevertheless, in cases involving less
severe crimes, it is harder to tell ex-ante how dangerous a person might
ultimately be. Under these circumstances,prosecutors could be using a
defendant’s race as a proxy for an individual’s latent criminality.266 Consistent
with this explanation, racial disparities are higher for cases involving less
severe crimes, but virtually non-existent for those cases involving the most
260 See supra notes 197–209 and accompanying text.
261 See supra notes 165–195 and accompanying text.
262 This connection of race and criminality may be conscious (i.e., statistical discrimination)
or subconscious (i.e., implicit bias). See supra notes 42–45, 52–54 and accompanying text.
263 See supra notes 52–54 and accompanying text.
264 See supra notes 52–57 and accompanying text.
265 See supra notes 197–209 and accompanying text.
266 See supra notes 56–57 and accompanying text.
267 See supra notes 165–195 and accompanying text.
C. Racial Disparities in Initial Charging
The analyses presented thus far have taken the prosecutors’ initial
charging decisions at face value, without considering possible discriminat
ory decisions against the defendant. As noted earlier, the decision whether to
bring initial charges against an offender and which specific charges to file
(given the particular characteristics of the arrest offense) is generally at the
discretion of the prosecuting attorney.268 One concern is that there may exist
racial biases at this stage, a fact that would complicate the interpretation of
racial disparities in the ensuing plea-bargaining process.269 For example, it
could be that prosecutors “over-charge” black defendants relative to white
defendants—that is, for a given arrest offense, black defendants are charged
with more serious crimes than white defendants.270 If this is the case, then
the magnitude of racial disparities in plea-bargaining documented in the
Article underestimates actual racial disparities.271
More problematic is the possibility that prosecutors “-ochvaerge”
white defendants relative to black defendants. If this is the case, then the
racial disparities in plea-bargaining (favoring white defendants)
documented above would merely be correcting (i.e., reversing) racial disparities in the
preceding initial charging process (which favored black defendants). The
data lacks the variables necessary to estimate the existence and magnitude
of such racial disparities in initial charging (this would require, for example,
arrest data at the individual level). Nevertheless, examining general patterns
in the initial charges faced by black and white defendants reassures us that it
is unlikely that white defendants are being initially “overcharged” by
Of the 48,357 observations in the final dataset, 25,629 (or 55.07%)
nivolve white defendants whereas 18,471 (or 38.2%) involve black
defendants. This 16.87% difference appears large but says little by itself. One way
to provide context to these figures is to look at the demographic
characteris268 See supra notes 77–79 and accompanying text.
269 To complicate matters further, prosecutors’ initial charging decisions may be influenced
by decisions made by policeofficers at the initial booking stage.See, e.g., OWENS ET AL., supra
note 62, at 2, 7 (arguing that racial disparities in sentencing are explainbedy differences at the
initial booking stage, wherein black defendants are booked for more severe charges than white
defendants, and not by disparate plea-bargaining outcomes).
270 Existing empirical evidence suggests that this may in fact be the caseS.ee supra note 60
and accompanying text.
271 In other words, a white defendant would be more likely to receive a charge reduction than
a similar black defendant who committed the same offense even though the white defendant was
initially charged with a less serious crime than the black defendant.
272 In addition, existing empirical evidence from other jurisdictions suggests that this is
unlikely to be the case. See supra note 60 and accompanying text.
tics of Dane County. In the 2010 census, 84.7% of the population was
categorized as white, with black individuals making up 5.2% of the pao-pul
tion.273 Black individuals thus appear to be overrepresented in this universe
of individuals charged with a crim—e their share of the population in the
database is over seven times their share of the general population in the
Looking at the distribution of black and white defendants in felony and
misdemeanor cases provides additional context. Of the 15,729 cases with at
least an initial felony charge, 51.43% involve white defendawnthsereas
41.51% involve black defendants, a difference of 9.9.2%Of the 32,627
misdemeanor cases, 56.82% involve white defendantswhereas 36.6%
involve black defendants, a difference of 20.2%2.275 Thus, the population of
individuals charged with a misdemeanor or felony crime, black individuals
appear to be significantly overrepresented in the felony subset relative to
The results presented in thisArticle document the existence of racial
disparities in the ple-baargaining stage of thecriminal justice proces.s
White defendants are more likely than black defendants to receive a
redcution in their principal initial charg2e7.7 As a result, white defendants who
face initial felony charges are more likely than black defendants to end up
being convicted of misdemeanors rather than more serious crimes2.78
Similarly, white defendants initially charged with misdemeanors are more likely
than black defendants to be convicted for crimes carrying no possible
incarceration or not being convicted at all.279 These disparities in plea-bargaining
appear to be driven by cases in which defendants have no prior cco-nvi
273 See Quick Facts—Dane County, Wisconsin: 2010 Demographic ProfileU, .S. CENSUS
BUREAU, https://factfinder.census.gov/ (search in Community Facts search bar for “Dane County,
Wisconsin”; then follow “2010 Demographic Profile” hyperlink).
274 To the extent that some of the black defendants prosecuted in Dane County resideee-ls
where, this metric may overestimate the relative representation of black individuals in the sample.
275 Another way of looking at this particular disparity is to consider the proportion of
defendants who are initially charged with a felony. For white defendants this rate is 30.38%, while for
black defendants the rate is 35.35%. The difference between thesreates—4.97%—is statistically
significant. That is, black defendants are 16.36% more likely than white defendants to be charged
initially with at least one felony.
276 In addition, in these felony cases, the average statutory maximum sentence for the
prinicpal charge in cases involving black defendants is 147 months, 13% higher than that of casens- i
volving white defendants, 130 months.
277 See supra notes 168–169 and accompanying text.
278 See supra notes 187–190 and accompanying text.
279 See supra notes 173–175 and accompanying text.
tions280 and cases which involve less serious offens2e8s1. These patterns
suggest that in “low information” cases, a defendant’s race may be used as a
proxy for their likelihood to recidivate and latent criminality.282
These results inform the evaluation of various current policy debates.
Efforts to mitigate racial disparities in sentencing and incarceration rates
should consider disparities in the plea-bargaining process and initial
charging decisions. Proposals aimed at restricting prosecutorial discretion
bynicreasing judicial discretion, for example via the elimination of statutory
minimum sentences, would seem to remedy these disparities2.83 The results
presented in thisArticle also highlight how racial disparities might run
deeper in misdemeanor cases, adding an empirical dimension to the nc-o
cerns of those scholars who have called for the decriminalization of
misedmeanors and for increased scrutiny of the misdemeanor adjudication op-r
cess.284 More generally, the possible implicit nature of the biases driving
these racial disparities lends support to those who have argued that the
weight afforded to evidence showing disparate impact in equal protection
claims should be re-examined.285
The results andconclusions presented in thisArticle are subject to
some caveats. As with most empirical work in this area, it is difficult to
establish a causal link between a defendant’s race and a criminal caset- ou
come (in this case, charge reductions). One concern isthat the models
estimated above may not be accounting for certain crime and defendant
characteristics that play a role in determining ple-abargaining outcomes and that
are also correlated with the race of the defendan2t8.6 Failing to control for
such unobservable variables could beskewing the results.287 One variable
for which the earlier analyses do not control is whether a defendant was
280 See supra notes 197–208 and accompanying text.
281 See supra notes 192–194 and accompanying text.
282 See supra notes 262–267 and accompanying text.
283 See, e.g., Fischman & Schanzenbach,supra note 46, at 757–61; Rehavi & Starr,supra
note 48, at 1349–51; Berman & Protass, supra note 159, at A17.
284 See supra notes 177–186 and accompanying text.
285 See supra notes 21–25 and accompanying text.
286 See, e.g., supra note 246 and accompanying text.
287 For example, consider two male individuals, one white and one black, of similar age and
with no prior criminal history, who are separately charged with first degree assault. If the black
defendant committed the assault in a more heinous manner than the white defendant (a fact
observable to the prosecutor, but unobservable to the researcher), it would not be surprising to
observe that the white defendant received a charge reduction while the black defendant did not.
However, for such unobservable variable to account for the results presented in this Article one
would need such variable (e.g., committing a given crime in a more heinous manner) to bes-sy
tematically correlated with defendants’ race.
granted bail.288 If black defendants in the sample are less likely to be
granted bail than white defendants,289 then some of the observed racial disparities
in charge reduction rates could be attributed to the fact that defendants
fcaing pre-trial detention are more likely to plead guilty to their initial top
charge.290 Similarly, it could be that prosecutors are relying on defenndta
characteristics other than race—but that are correlated with race—to assess
the risk posed by a defendant.291 More generally, one must consider the
possibility that the results are driven not by prosecutors showing more mercy
for white defendants, but bydefense attorneys being less zealous (whether
consciously or subconsciously) in their representation of black
defendants.292 Finally, the dataset includes cases from 200–02006 and the results
may not necessarily be representative of Dane County’s currecnrtiminal
justice process.293 In 2010, the Wisconsin governor appointed Dane
County’s first African American district attorney2.94 And starting in 2016, Dane
288 See Stephanos Bibas, Plea Bargaining Outside the Shadow ofTrial, 117 HARV. L. REV.
2463, 2493 (2004) (arguing that prter-ial detention “hampers a defendant’s ability to mount a
defense,” thus “plac[ing] a high premium on quick plea bargains in small cases”).
289 See Traci Schlesinger, Racial and Ethnic Disparity ni Pretrial Criminal Processing, 22
JUST. Q. 170, 181–83 (2005) (finding that black defendants are 25% more likely to be denied bail
than white defendants and that black defendants are 87% more likely than white defendants to be
incarcerated prior to trial). Such disparities in bail determinations are not necessarily reflecting the
risk posed by individual defendants.See David Arnold et al.,Racial Bias in Bail Decisions23
(Nat’l Bureau of Econ. Working Paper No. 23421, 2017) (finding that “marginally released white
defendants are 19.8 percentage points more likely to be rearrested prior to disposition than rm-a
ginally released black defendants . . . , consistent with racial bias against blacks”).
290 See Megan Stevenson, Distortion of Justice: How the Inabiitly to Pay Bail Affects Case
Outcomes 17–18 (Jan. 12, 2017) (unpublished manuscripth),ttps://ssrn.com/abstract=2777615
[http://perma.cc/UV42-EW5S] (finding that pre-trial detention is associated with a 13% increase
in the likelihood of being convicted, due mostly to increased guilty pleas by defendants).
291 A defendant’s socio-economic status (e.g., income or education) could be one such
characteristic. However, the fact that we observe racial disparities even after controlling for the identity
of the defense attorney (arguably a proxy for a defendant’s soceioco-nomic status) reassures us
that this particular omitted variable is not driving the results.See supra notes 246–259 and
292 See L. Song Richardson & Phillip Atiba GoffI,mplicit Racial Bias in Public Defender
Triage, 122 YALE L.J. 2626, 2636 (2013) (arguing that implicit biases can affect how public
defenders evaluate evidence relating to a defendant based on the defendant’s race, “potentially
causing [a public defender] to unintentionally interpret information as more probative of guilt”); Molly
J. Walker Wilson, Defense Attorney Bias and the Rush to the Plea, 6K5AN. L. REV. 271, 283
(2016) (“Differences in socio-economic background, education, and race can influence even
dilgent defense attorneys powerfully, and without the attorney being conscious of them.”).
293 See supra notes 165–195 and accompanying text.
294 On 2010, the Wisconsin governor appointed Ismael Ozanne as Dane County’s district
attorney. Mr. Ozanne, who was subsequently elected and re-lected in 2012 and 2016, has implemented
internal policies seeking to address racial discrimination. See Ed Treleven, Ismael Ozanne Re-elected
as DA over Prosecutor Bob Jambo,isWIS. ST. J. (Aug. 10, 2016), http//:host.madison.com/wsj/
County has provided implicit bias training for judges, prosecutors and
public defenders.295 It will be interesting to see whether recent changes in
policies and personnel training have been effective in addressing the disparities
Setting these caveats aside, there is no reason why the exercise of
prosecutorial discretion and the plea-bargaining process should not be
subject to the same rigorous empirical scrutiny that has been applied to judicial
decision making and the sentencing process. It would be interesting to see
whether the patterns in racial disparities documented in thisArticle are
replicated in other jurisdictions. Empirical work should also examine more
closely any disparities in plea-bargaining outcomes along gender lines and
explore the intersection of gender and race in this regard.296 Future work can
also focus more closely on the role of the prosecutor by examining, for
example, which attributes and characteristics of prosecuting attorneys (if any)
explain differences in plea-bargaining outcomes297 or documenting the
ex7b015c97-1a1b-53e0-b0fd-b65428a8e203.html [https//:perma.cc/26LN-RH6H]; Ed Treleven,
Race for Dane DA Has Experienced Underling Sniping at the IncumbenWt,IS. ST. J. (Aug. 9,
295 Dane County recently adopted a number of recommendations made by workgroups charged
with evaluating policies to improve the criminal justice system. Among these recommendations was
the implementation of an ongoing “Implicit Bias, Equity and Inclusion, Diversity and Poverty
traniing” for judges, prosecutors and public defenders, among others.See DANE CTY. BD. SUPERVISORS,
INVESTIGATING SOLUTIONS TO RACIAL DISPARITIES AND MENTAL HEALTH CHALLENGES IN THE
DANE COUNTY JAIL AND THROUGHOUT DANE COUNTY’S CRIMINAL JUSTICE SYSTEM: WORK
GROUP RECOMMENDATIONS 9 (Sept. 2015). These implicit bias trainings are already taking place.
See Chris Rochester & Tyler Brandt, Dane County Spends $50,000 to Send Courthouse Staff to
“Implicit Bias” Trainin,g MACIVER INST. (Aug. 8, 2016), http//sw:ww.maciverinstitute.com/2016/
296 For a summary of studies that have documented gender disparities in sentencing outcomes
see supra note 40 and accompanying text. Recent studies have examined gender disparities in the
pre-sentencing stages of federal criminal cases.See, e.g., Shermer & Johnson,supra note 48, at
413–15 (examining differences in charge reductions); Sonja B. Starr,Estimating Gender
Disparities in Federal Crimnial Cases, 17 AM. L. & ECON. REV. 127, 154 (2015) (stressing the
m“iportance of pre-sentencing stages in shaping sentence disparities”). In a separate study I document
gender disparities in plea bargaining using the same dataset employed in this ArticleS.ee Carlos
Berdejo, Gender Disparities in Plea Bargaining, 94 IND. L.J. (forthcoming 2019).
297 Studies analyzing differences in judges’ sentencing behavior based on judges’
demographic characteristics have been inconclusive. While some of these studies have found
differences between male and female judges and between minority and white judges, other studies
have found that the race and gender of judges have little influence on their sentencing behavior.
See Abrams et al., supra note 36 at 372–74 (finding that black judges are associated with longer
sentences but lower incarceration rates); Claire S.H. Lim et al.D,o Judges’ Characteristics
Matter? Ethnicity, Gender, and Partisanship in Texas State Trial Cour,ts18 AM. L. & ECON. REV.
302, 305 (2016) (finding that demographic characteristics of judges have little effect on sentence
istence of significant heterogeneity across individual prosiencgut
Conducting this type of empirical work requires detailed data that
provides information on pre-sentencing decisions made by prosecutors. One of
the reasons why so much empirical work has focused on judges’ sentencing
decisions is the fact that public entities at the state and federal levels collect
and maintain comprehensive data on sentencing.299 There needs to be a
similar level of transparency with respect to decisions taken by actors in the
pre-sentencing stages of the criminal justice process. Collecting and
maintaining such data certainly presents a more complex and challenging
endeavor than collecting data on sentencing decisions, both in terms of the
number of observations and variables.Nevertheless, the Wisconsin Circuit
Courts’ CCAP shows that this can be achieved3.00 And a number of district
attorney offices across the nation have tried to collect and use data to
intrenally identify and address instances of racial bias in the exercise of
prosecutorial discretion.301 The availability of this type of information would also
allow courts and other external groups to scrutinize prosecutorial decisio-n
making, an area that traditionally has been less open to the public than
snetencing.302 Decision-makers exercising discretion in a transparentcriminal
length); Schanzenbach, supra note 35, at 73 (documenting similar findings); Darrell Steffensmeier
& Chester L. Britt, Judges’ Race and Judicial Decision Making: Do Black Judges Sentence Df-i
ferently?, 82 SOC. SCI. Q. 749, 757–58 (2001) (finding that black judges are more likely to
incracerate offenders than white judges); Darrell Steffensmeier & Chris Hebert,Women and Men
Policymakers: Does the Judge’s Gender Affect the Sentencing of Criminal Defenda,nt7s7? SOC.
FORCES 1163, 1174–75 (1999) (finding that female judges are more likely to incarcerate offenders
and sentence the imprisoned to slightly longer sentences than male judges).
298 Recent studies on judges’ sentencing behavior have focused on individual judges, finding
substantial heterogeneity in average incarceration rates and sentencing length across judgeSs.ee
Abrams et al., supra note 36, at 367–68 (finding that judges’ decisions show significant
heterogeneity in all sentencing measures, including incarceration, average sentence length, and average
sentence length conditional on receiving a nonzero jail sentence); Lim et asl.u,pra note 297, at
305 (finding substantial heterogeneity in sentencing harshness across judges).
299 See supra notes 58, 68 and accompanying text.
300 See supra notes 104–106 and accompanying text.
301 See WAYNE MCKENZIE ET AL., VERA INST. JUST., PROSECUTION AND RACIAL JUSTICE:
USING DATA TO ADVANCE FAIRNESS IN CRIMINAL PROSECUTION 7 (Mar. 2009), https://www.
vera.org/publications/prosecution-and-racial-justice-using-data-to-advance-fairness-in-criminalprosecution (follow “PDF” hyperlink) [https://perma.cc/42QN-T9AV].
302 See KUTATELADZE ET AL., supra note 64, at 1 (“While prosecutorial discretion is often
guided by internal policies, external regulation or oversight of this discretion is quite limited.”);
MCKENZIE ET AL., supra note 301, at 1 (“Unlike officials in law enforcement and the judiciary,
who have come under varying degrees of oversight in recent years, prosecutors act with little
otuside scrutiny or governance.”).
303 See KUTATELADZE ET AL., supra note 64, at 9 (“Prosecutors, as powerful actors in the
criminal justice system, are empowered to adopt measures that promise to significantly promote
equity for all people throughout all stages of the criminal justice continuum. Doing so will require
a commitment to accountability and transparency.”).
Table 1—Description of Main Explanatory Variables 304
Indicator variable equal to 1 if the defendant was African American.
Indicator variable equal to 1 if the defendant was not African American or
Indicator variable equal to 1 if the defendant was a woman.
Equal to the age of the defendant as of the initial filing and calculated as the year
of the case filing minus the defendant’s year of birth.
Indicator variable equal to 1 if defendant had one prior conviction.
Indicator variable equal to 1 if defendant had two or more prior convictions.
Indicator variable equal to 1 if defendant was convicted of at least two chargse. In
specifications restricted to felonies, it is equal to 1 if there are alteast two felony
convictions. In specifications restricted to misdemeanors, it’s equal to 1 if there are
at least two misdemeanor convictions.
Indicator variable equal to 1 if defendant was initially cahrged with at least two
crimes. In specifications restricted to felonies, it is equal to 1 if there aer at least
two felony charges. In specifications restricted to misdemeanors, it’s equal to 1 if
there are at least two misdemeanor charges.
Indicator variable equal to 1 if any of the charges in a case was adjudicated ian
304 This Table is permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/law/
Table 2—Crime Type Categories 305
Other Drug Offenses (Wis. Stat. §§ 961.41(4),
OWI/PAC (Wis. Stat. § 346.63)
OWI/PAC w/Child (Wis. Stat. § 346.63)
Reckless/Negligent Injuries (Wis. Stat. §§
Bail Jumping (Wis. Stat. § 946.49)
Crimes Against Animals (Wis. Stat. § § 951)
Receiving Stolen Property (Wis. Stat. § 943.34)
Criminal Damage to Property (Wis. Stat. § 943.01)
Resisting Arrest (Wis. Stat. § 946.41)
Disorderly Conduct (Wis. Stat. § 947.01)
Robbery (Wis. Stat. § 943.32)
Drug Manufacture/Deliver (Wis. Stat. § 961.41(
Sex Crimes (Wis. Stat. §§ 944.15-.32)
Drug Possession (Wis. Stat. § 961.41(3g))
Drug Possession w/ Intent (Wis. Stat. §§ 961.41(1m), (2))
Endangering Safety (Wis. Stat. § 941.30)
Escape (Wis. Stat. § 946.42)
Fail to Report to Jail (Wis. Stat. § 946.425)
Fleeing Officer (Wis. Stat. § 346.04)
Sex Registry Violation (Wis. Stat. § 301.45)
Sexual Assault Child - Repeated (Wis. Stat.
Sexual Assault Child (Wis. Stat. § 948.02)
Sexual Assault (Wis. Stat. § 940.225)
Stalking (Wis. Stat. § 940.32)
Theft - Credit Card (Wis. Stat. § 943.41)
Forgery/Fraudulent Writing (Wis. Stat. §§ 943.37-40)
Theft - Identity (Wis. Stat. §§ 943.201-.203)
Fraud on Merchants (Non-Retail) (Wis. Stat. § 943.21)
Theft - Retail (Wis. Stat. § 943.50)
Hit and Run (Wis. Stat. §§ 346.67-.69)
Homicide (Wis. Stat. §§ 940.01-.10)
Injury by use of Vehicle (Wis. Stat. §940.25)
Theft (Wis. Stat. § 943.20)
Threats to Injure/Accuse of Crime (Wis. Stat.
Threats/Harassment (Wis. Stat. §§ 947.012-.013)
Intimidate Witness/Victim (Wis. Stat. §§ 940.42-.45)
Trespass (Wis. Stat. §§ 943.125-.145)
Kidnap/Hostage (Wis. Stat. §§ 940.30-.31)
Violation of TRO (Wis. Stat. §§ 813.12 - .128)
Operate Vehicle w/out Consent (Wis. Stat. § 943.23)
Weapons/Explosives (Wis. Stat. §§ 941.20-.315)
Other Crimes Against Children (Wis. Stat. §§ 948.07-.13)
Worthless Checks (Wis. Stat. § 943.24)
Other Crimes Against Children (Wis. Stat. §§ 948.30-.62)
305 This Table is permanently available aht ttp://www.bc.edu/content/dam/bc1/schools/law/
Table 3—Charge Reduction Rates & Crime Severity
(with Alternative Crime Controls) 306
306 This Table is permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/law/
pdf/law-review-content/BCLR/59-4/berdejo-graphics.pdf [https://perma.cc/ECS5-GP79]. Note:
Robust standard errors in brackets (* significant at 10%; ** significant at 5%; *** significant at
1%). This table presents the marginal effects from probit models in which the outcome variable is
an indicator variable equal to 1 if the initial top charge in a case was dropped, amended to a lesser
charge or dismissed. Column (
) includes all cases, column (2) includes all misdemeanor cases
and column (3) include all felony cases. Columns (4) and (5) divide felony cases based on whether
the principal charge carried a potential sentence of (i) 5 years or less or (ii) more than 5 years,
respectively. For a description of the explanatory variables of interest seeinfra note 304 and
accompanying text (Appendix 1: Table 1). All regressions include a set of year fixed effects and
crime class-type interaction fixed effects (see supra note 213).
Table 4—Charge Reduction Rates & Criminal History
(with Alternative Crime Controls) 307
307 This Table is permanently available at http://www.bc.edu/content/dam/bc1/schools/lawp/df/
law-review-content/BCLR/59-4/berdejo-graphics.pdf [https/:/perma.cc/ECS5-GP79]. Note: Robust
standard errors in brackets (* significant at 10%; ** significant at 5%; *** significant at 1%). This
table presents the marginal effects from the series of probit models described nexTt.he outcome
variable in columns (
)–(6) is an indicator variable equal to 1 if the initial top charge in a case was
dropped, amended to a lesser charge or dismissed. Columns (1–)(2) includes all cases, while
columns (3)–(4) and (5)–(6) include all misdemeanor and felony cases, respectively. Columns (
and (5) restrict the sample to cases in which the defendant had no prior convictions. Columns (2),
(4) and (6) restrict the sample to cases in which the defendant had at least one prior conviction.
For a description of the explanatory variables of interest see infra note 304 and accompanying text
(Appendix 1: Table 1). All regressions include a set of year fixed effects and crime c-ltayspse
interaction fixed effects (see supra note 213).
1. The Criminal Justice Process....................................................................................... 1201 2 . Prison Population and Incarceration Rates .................................................................. 1203 1. Overview of the Database............................................................................................ 1204 2. Construction of the Variables ...................................................................................... 1206 1 . Charge Reductions and Crime Severity....................................................................... 1214 2 . Charge Reductions and Defendant's Criminal History................................................ 1220 3. Controlling for Defendant and Crime Characteristics.................................................. 1224 4. The Role of Prosecuting and Defense Attorneys ......................................................... 1231 5. Race as a Proxy for Criminality................................................................................... 1237 C. Racial Disparities in Initial Charging ............................................................................. 1239
80 See WIS . STAT. § 972 .02 ( 2018 ).
81 See infra notes 82-89 and accompanying text.
82 See O'Hear , supra note 77, at 1; supra note 46 and accompanying text.
83 See O'Hear , supra note 77, at 8 (noting that few crimes carry a mandatory minimum and that the most important of these are Class A felonies which carry a mandatorylife sentence).
84 See id. ; infra notes 85-86 and accompanying text.
85 The classes of felony crimes are as follows, with maximum sentences in parenthesis: A (Life), B (60 years), C (40 years), D (25 years) , E (15 years) , F (12 . 5 years), G (10 years), H (6 years), and I (3.5 years) . WIS. STAT. § 939 .50. Prior to 2002 , there were six classes of felonies: A (Life ), B (60 years) , BC (30years) , C (15 years) , D (10 years) , and E y( e5ars ). WIS. STAT. § 939.50 ( 1999 ). Three new classes of felony crimes were added as part of a broader reform to the criminal system in 2002 . See 2001 WIS . ACT 109 §§ 545 - 559 ( July 26 , 2002 ) a(mending WIS . STAT. § 939 .50). Prior to this reform, a number of drug offenses, among others, had mmaxuim sentences that were set by statute and not keyed to a class .See, e.g., WIS. STAT. § 961 . 41 (setting sentence lengths for certain controlled substance offenses).
86 The three classes of misdemeanor crimes are the following, with maximum prison sentences in parenthesis: A (9 months), B (90 days), and C30( days) . See WIS. STAT. § 939.51 ( 2018 ). More generally, misdemeanors are defined as crimes that are not punishable by imprisonment in a state prison . See id. § 939 . 60 .
87 See O'Hear , supra note 77, at 7-8. For example, if a violent felony was committed in a designated school zone, the applicable maximum sentence is increased by five ye . aSrsee WIS. STAT. § 939.632 ( 2 ).
88 See O'Hear , supra note 77, at 7.
89 See WIS . STAT. § 939 . 50 ( 3 ). A judge choosing to impose probation has wide discretion in structuring a probationary scheme . See O'Hear, supra note 77 , at 8. The judge may withhold sentencing altogether, or impose a sentence of incarceration and stay itfor the duration of probation . See WIS. STAT. § 973.09 ( 1 ) (a). In cases where sentencing is withheld, a defendant that violates probation requirements has to return to court to be sentenced by a judgSe.e Community Corrections-General Information, WIS . DEP'T CORRS ., https://doc.wi.gov/Pages/AboutDOC/Community Corrections/GeneralInformation.aspx#ppes [http://perma.cc/HAB4-DJ8V]. If a sentence was miposed but stayed, a defendant that violates a conditionin its probation is automatically incarcerated for the term of the original sentence . Id.
90 See Michael O'Hear , Good Conduct Time for Prisoners: Why ( and How) Wisconsin Should Provide Credits Toward Early Release , 98MARQ . L. REV. 487 , 496 ( 2014 ) ; O'Hear , supra note 77, at 8. This applies to all felony offenses committed on or after December 31 , 1999 . See O'Hear , supra note 77, at 16. Additionally, although appellate courtsreview judges' sentencing decisions, the process is not very rigorous and seldom leads to the overturning of a sentenSce.e Michael O'Hear, Appellate Review of Sentence Explanations: Learning from the Wisconsin and Federal Experiences, 93 MARQ. L. REV. 751 , 762 ( 2009 ) ; O'Hear , supra note 77, at 9-11.
91 See O'Hear , supra note 77, at 2.
92 See Michael M. O'Hear ,Mass Incarceration in Three Midwestern States: Origins and Trends , 47 VAL. U. L. REV. 709 , 741 ( 2013 ) ; O'Hear , supra note 77, at 2-3.
93 See O'Hear , supra note 77, at 6.
94 See id . at 7.
95 See NELLIS , supra note 4 , at 16; CHRISTINA D. CARMICHAEL , WIS. LEGISLATIVE FISCAL BUREAU, ADULT CORRECTIONS PROGRAM 12 ( 2011 ), http://docs.legis.wisconsin.gov/misc/lfb/ informational_papers/january_2011/0057_adult_corrections_program_informational_paper_57.pdf [https://perma.cc/ADZ5-Y996].
96 See JOHN PAWASARAT & LOIS M. QUINN , UNIV. WIS. MILWAUKEE, WISCONSIN'S MASS INCARCERATION OF AFRICAN AMERICAN MALES: WORKFORCE CHALLENGES FOR 2013 , at 8 ( 2013 ), http://www4.uwm.edu/eti/2013/BlackImprisonment.pdf [http://perma.cc/BE2N-79BH].
97 See NELLIS , supra note 4 , at 8.
98 See O'Hear , supra note 77, at 6.
99 See id.
100 See id.
101 See Pawasarat & Quinn, supra note 96, at 12.
102 See infra notes 104-111 and accompanying text.
103 See infra notes 112-116 and accompanying text.
104 WIS. COURT SYS. CIRCUIT COURT ACCESS , https://wcca.wicourts.gov/index.xsl. Battery - Special (Wis. Stat. §§ 940 .201, 203 ) Battery (Wis. Stat . §§ 940 . 19 - 20 ) Burglary (Wis . Stat. § 943 .10) Child Abuse (Wis . Stat. § 948 .03) Child Neglect/Fail to Support (Wis . Stat. §§ 948 . 21 -23)