Content uploaded by Tara Mitchell
Author content
All content in this area was uploaded by Tara Mitchell on Apr 07, 2015
Content may be subject to copyright.
Law and Human Behavior, Vol. 29, No. 6, December 2005 (
C
2005)
DOI: 10.1007/s10979-005-8122-9
Racial Bias in Mock Juror Decision-Making:
A Meta-Analytic Review of Defendant Treatment
Tara L. Mitchell,
1,5
Ryann M. Haw,
2
Jeffrey E. Pfeifer,
3
and Christian A. Meissner
4
Common wisdom seems to suggest that racial bias, defined as disparate treatment
of minority defendants, exists in jury decision-making, with Black defendants being
treated more harshly by jurors than White defendants. The empirical research, how-
ever, is inconsistent—some studies show racial bias while others do not. Two previous
meta-analyses have found conflicting results regarding the existence of racial bias in
juror decision-making (Mazzella & Feingold, 1994, Journal of Applied Social Psy-
chology, 24, 1315–1344; Sweeney & Haney, 1992, Behavioral Sciences and the Law,
10, 179–195). This research takes a meta-analytic approach to further investigate the
inconsistencies within the empirical literature on racial bias in juror decision-making
by defining racial bias as disparate treatment of racial out-groups (rather than focus-
ing upon the minority group alone). Our results suggest that a small, yet significant,
effect of racial bias in decision-making is present across studies, but that the effect
becomes more pronounced when certain moderators are considered. The state of the
research will be discussed in light of these findings.
KEY WORDS: racial bias; juror decision-making; meta-analysis.
It is evident from a review of the literature that minority groups experience a dis-
tinct disadvantage with regard to the American criminal justice system. For exam-
ple, according to a recent report by Human Rights Watch (2000), Black male drug
users are 13 times more likely than White male drug users to be sentenced to jail,
even though the estimated drug usage rates are equivalent for the two groups. Al-
though researchers tend to agree that these statistics clearly demonstrate the dis-
parate treatment that minority group individuals suffer within the legal system (see
e.g., Sommers & Ellsworth, 2001), there is far less agreement over the exact cause
1
Lock Haven University.
2
Big Bend Community College.
3
University of Regina.
4
University of Texas of El Paso.
5
To whom correspondence should be addressed at Department of Psychology, Lock Haven University,
Lock Haven, PA, 17745; e-mail: tmitchel@lhup.edu.
621
0147-7307/05/1200-0621/1
C
2005American Psychology-Law Society/Division 41 of the American Psychological Association
622 Mitchell, Haw, Pfeifer, and Meissner
for this disparity, especially given the numerous and varied decision points through-
out the system including: a police officer’s decision to arrest, a prosecutor’s decision
to charge, a prosecutor’s decision to offer a plea, a jury’s verdict, and the sentence
imposed by a judge.
A number of archival studies have illustrated that minority defendants (par-
ticularly Black defendants) are given longer sentences, and sentenced to the death
penalty more often, than White defendants (Austin & Allen, 2000; Mustard, 2001;
Williams & Holcomb, 2001). Austin and Allen (2000), for example, examined the
number of defendants committed to the Pennsylvania Department of Corrections
between 1991 and 1995 in relation to the number arrested between 1990 and 1994
(allowing for a 1-year time lag in arrest and commitment/sentence). They sought
to examine whether the disparate ratio of minorities to Whites in the Pennsylva-
nia prison system was due to differential commitment of crimes or to racial dis-
crimination within the justice system. Using arrest rate as a proxy for rate of crime
commission, the authors found that the arrest rate explained only 43% of the dis-
proportionality in commitment rate, and suggested that race was influencing the
likelihood that a defendant would be convicted and sentenced to the prison system.
Drug offenses were a large part of this disproportionality, however, as the arrest rate
explained 70% of the disproportionality in commitment rate when drug offense de-
cisions were excluded from the analysis. Mustard (2001) conducted a similar analysis
of 77,256 defendants sentenced in federal courts under the United States Sentenc-
ing Commission Guidelines. He found that Black and Hispanic defendants were
given longer sentences than White defendants, even after controlling for crime se-
riousness. The majority of these effects were confined, however, to those cases in
which judges departed from the federal guidelines—as might have been expected,
there was less discrepancy in cases where judges followed the guidelines. Baldus and
his colleagues (Baldus, Pulaski, & Woodworth, 1983; Baldus, Woodworth, Grosso,
& Christ, 2002; Baldus, Woodworth, Zuckerman, Weiner, & Broffitt, 1998) have
conducted several archival studies including thousands of capital cases within var-
ious locales, including Georgia, Nebraska, and Philadelphia. They have found that
defendant race influences death sentences, with Black defendants four times more
likely to receive the death penalty than White defendants (Baldus et al., 1998). They
have also found that defendant and victim race interact, such that Black defendants
are given much longer sentences than White defendants when the victim is White
(Baldus et al., 1983; Baldus et al., 2002). It is important to note, however, that a
handful of studies have not found a main effect for defendant race. Williams and
Holcomb (2001), for example, found that defendant race did not predict death sen-
tences in Ohio between 1981 and 1994, even when the sample of cases was restricted
to homicides with felony circumstances.
Citing the above significant findings as evidence, a number of researchers have
argued that it is clear that there is discriminatory treatment within the justice system
(e.g., Alexander & Gyamerah, 1997). Despite the variety of decision points in which
disparate treatment could occur, it may be argued that no other area has been empir-
ically investigated as often as that of jury decision making (Pfeifer, 1990). This sup-
position has led to a significant amount of empirical research on the decision-making
of “mock” jurors. Results from a number of these studies have indicated that Black
Racial Bias 623
defendants are treated significantly more harshly than White defendants, especially
in cases involving murder or rape (for a review of this research, see Sommers &
Ellsworth, 2001). In contrast, some studies have found a reverse effect, with White
defendants being treated more harshly than Black defendants (see Poulson, 1990),
leading to some confusion over the issue of race as it relates to mock juror decision-
making. Furthermore, a handful of studies and archival research have found that
defendant race did not influence juror decision-making (e.g., Pfeifer & Ogloff, 1991;
Williams & Holcomb, 2001).
In an attempt to summarize and clarify the racial bias literature, two meta-
analyses have previously been conducted (Mazzella & Feingold, 1994; Sweeney &
Haney, 1992). While both of these meta-analyses sought to synthesize the empirical
literature on racial bias in jury decision-making, each resulted in a different conclu-
sion regarding the state of the field. The Sweeney and Haney (1992) meta-analysis
included 14 studies, with 19 effect sizes, involving 2,836 participants. The analysis
focused on racial bias, defined as the disparate treatment of Black defendants, in
sentencing decisions made by White mock juror participants. The authors found a
small, but significant, effect of racial bias (d = .17) across studies, indicating that
the White participants were more likely to give Black defendants longer sentences
than White defendants. Sweeney and Haney (1992) also investigated the potential
influence of several moderator variables including: the year the study was published;
the region the study was conducted in (South v. not South); the type of crime (rape
v. not rape); the type of sample (student v. community); the method of conveying
defendant race (pictures v. words); whether the study specifically mentioned the
participants’ race; and whether the victim’s race was specified. Participant race and
victim race moderated the racial bias effect, such that studies specifying the race of
the participants and those specifying the race of the victim produced larger effect
sizes than those that did not specify these details.
A second meta-analysis conducted by Mazzella and Feingold (1994) involved
a much broader investigation of extralegal factors that could influence juror deci-
sions. In addition to race, these authors also examined physical attractiveness, so-
cioeconomic status, and gender of the defendant and victim. For purposes of the
present analysis, however, we will focus on the influence of race in the Mazzella and
Feingold meta-analysis. It is important to note that the implicit definition of racial
bias relied upon by the authors appears to have been disparate treatment of the
minority (Black) defendant. The analysis included 29 studies, with 63 effect sizes,
involving 6,709 participants. Unlike the Sweeney and Haney (1992) meta-analysis,
Mazzella and Feingold included studies that involved Black participant jurors and
an analysis of verdict decisions. Overall, the authors did not find a significant ef-
fect of racial bias on either judgments of guilt (d = .01, k = 21, N = 3, 486) or sen-
tencing decisions (d = .06, k = 27, N = 4, 045). Despite these non-significant effect
sizes, the authors found significant heterogeneity of effect sizes across studies indi-
cating that certain variables might be moderating the effect. Mazzella and Feingold’s
moderator analysis indicated that crime type played a significant role in decision-
making, such that Black defendants were given longer sentences for crimes of neg-
ligent homicide, while White defendants were given longer sentences for crimes
of fraud. Mazzella and Feingold also found that, although victim race influenced
624 Mitchell, Haw, Pfeifer, and Meissner
sentencing decisions (i.e., defendants were given longer sentences for crimes
against White victims than crimes against Black victims), it did not influence guilt
decisions.
In the decade following the publication of these two meta-analyses, researchers
have further investigated the existence of racial bias, as well as the degree to which
the effect may be moderated by certain variables that were not accounted for in pre-
vious research. For example, one prominent criticism of early jury decision-making
studies is that they were generally low in ecological validity, that is, such studies
often employed student populations (rather than community members), provided
continuous scale guilt responses (rather than dichotomous verdict choices), and
frequently failed to inform mock jurors of the relevant legal instructions prior to
rendering a verdict (Pfeifer, 1990). Some researchers have argued that mock jurors
will show racial bias only in ambiguous situations, and that such situations exist
when mock jurors are not instructed on relevant legal standards (see e.g., Pfeifer &
Bernstein, 2003; Pfeifer & Ogloff, 1991, 2003; Stephan & Stephan, 1986). Addition-
ally, continuous guilt measures are often used to maximize power in detecting an
effect, and some researchers have questioned whether a practically-significant effect
of racial bias might be observed when a dichotomous verdict decision is employed
(see Pfeifer, 1990). Taken together, such arguments lead to the hypothesis that
the racial bias effect seen in juror decision-making studies may not be evidenced
when more ecological methods are utilized. This supposition is an important one
within the context of the current analysis, given that previous meta-analyses did not
examine the possibility that certain legally relevant variables could moderate the
racial bias effect.
As noted above, the two previous meta-analyses appeared to define racial bias
as the disparate treatment of a minority defendant (typically a Black defendant).
As such, these studies restricted analysis to consideration of the disparate treatment
of a Black defendant by all participants, regardless of race or ethnicity. Sweeney
and Haney (1992), for example, excluded the responses of Black participants, when
they were separable from responses of non-Black participants, because they felt that
Black participants would not sentence same-race defendants more harshly. While
Mazzella and Feingold (1994) included the responses of Black participants, they
did not analyze the data separately based on the race of the participant. It may be
argued, however, that an alternative definition of racial bias may be more appro-
priate for understanding the treatment of defendants, especially if that definition
revolves around the disparate treatment of members of a racial out-group. Such a
definition allows for the possibility that Black participants will provide disparate
treatment towards White defendants just as White participants will provide dis-
parate treatment towards Black defendants. However, this definition can only be
applied when the responses from members of differing racial groups are analyzed
separately.
The current meta-analysis, therefore, attempts to resolve inconsistencies within
the racial bias literature by examining research on juror decision-making in which
the race of the defendant was manipulated. In contrast to previous meta-analyses,
racial bias is defined here as a juror’s disparate treatment of a defendant from
a racial out-group, when compared with a defendant of the juror’s own-race, in
Racial Bias 625
verdict and sentencing decisions. Additionally, this meta-analysis sought to ex-
amine the influence of several legally-relevant moderator variables on the racial
bias effect. It was expected that the meta-analysis would demonstrate a reliable
effect of racial bias on both verdict and sentencing decisions. It was also ex-
pected that the racial bias effect would be moderated by several key variables
related to the ecological validity of the studies. Of particular interest were vari-
ables that related to legal procedure, such as the presence of instructions and
the use of a continuous v. dichotomous guilt measure. In addition, several other
methodological variables were considered, including the race of the participant, the
type of sample (students v. community members), and the date and status of the
manuscript.
METHOD
Studies
Manuscripts were located through a number of sources, including: (a) searching
the PsychINFO, Sociological Abstracts, National Criminal Justice Reference Ser-
vices Abstracts, and Dissertation Abstracts computer databases using the search
terms “racial bias,” “race and defendant,” and “race and jury”; (b) cross-referencing
studies in the two previous meta-analyses (Mazzella & Feingold, 1994; Sweeney &
Haney, 1992); and (c) contacting researchers in the field who may have had knowl-
edge of unpublished literature. A total of 46 independent effect sizes from 34 stud-
ies, representing the responses of 7,397 participants, were located for the analysis of
racial bias on juror verdict decisions. Sixteen studies, representing 20 independent
effect sizes and 3,141 participants, were identified for the analysis of racial bias on
sentencing decisions.
Inclusion/Exclusion Criteria
To be included in the sample, each study had to meet the following criteria:
(1) the study had to involve an experimental manipulation of the race of the de-
fendant; (2) the study had to contain enough information to define racial bias as
the disparate treatment of a defendant from a racial out-group, such that results
from multi-race participant samples were presented separately for each race; and
(3) the study had to assess guilt or sentencing in the context of a mock juror simula-
tion. Every effort was made to calculate an exact effect size. The authors of studies
that did not report sufficient statistical information for the calculation of an effect
size were contacted in order to obtain that information. Those studies that reported
non-significant results (p >.05) for which we could not calculate an effect size for
verdict or sentencing were given an effect size of .00 (Dovidio et al., 1997; Foley &
Chamblin, 1982; Foley & Pigott, 2002; Hymes et al., 1993; McGowan & King, 1982;
McGuire & Bermant, 1977; Rector & Bagby, 1995; Rector et al., 1993; Sargent &
Bradfield, 2004). This approach was used to establish a conservative estimate of
the effect of racial bias, although it should be noted that such a procedure weak-
ens the overall effect size (Pigott, 1994). Researchers are encouraged to provide
626 Mitchell, Haw, Pfeifer, and Meissner
more complete statistics such that insertion of zero effect sizes in future syntheses is
unnecessary.
Coded Variables
Moderator variables were identified based upon three categories of informa-
tion, namely: variables of theoretical interest; variables that may be sources of
distortion, bias, or artifact (e.g., methodological variables); and extrinsic study char-
acteristics (see Lipsey, 1994). Due to the fact that many of our methodological
variables were also of theoretical interest, we combined those two categories of in-
formation. The methodological/theoretical variables of interest included (a) race of
participant (White v. Black participants); (b) type of sample (students v. commu-
nity members); (c) jury instructions (present v. absent); and (d) type of dependent
measure (continuous v. dichotomous scale). Each of these variables had some le-
gal or theoretical relevance. Jury instructions and type of dependent measure, for
example, are important elements of the legal system that are not consistently em-
ployed in the same manner within mock juror materials. Type of sample is also a
possible concern, as the responses of the typical college student employed in many
mock juror studies may not generalize to the responses of jurors in real cases who
represent a wider range of demographic characteristics (Bornstein, 1999). Race of
the participant also has a strong theoretical and legal relevance, as much of social
psychological literature points to an ingroup bias, in which individuals favor mem-
bers of their own group over others, regardless of whether they are a member of the
minority or majority group (e.g., Wilder & Shapiro, 1991). Legally, race of the juror
can no longer be used as a basis for removing someone from juror service and, as
such, the number of minority members on juries appears to be increasing, particu-
larly as many states show a concerted effort to increase the number of minority jury
members (Joint State Government Commission, 2003). It is therefore important
to determine if minority members show the same racial bias as majority members.
We had originally planned on including additional legally and theoretically relevant
variables, such as race of the victim and crime type, but were unable to do so due to
small sample sizes for these variables. In addition to the methodological/theoretical
variables of interest, several extrinsic study characteristics were included such as (e)
date of the study and (f) publication status (published v. unpublished).
Measure of Effect Size
We chose an r-based correlation coefficient as our effect size, following the
recommendations of Rosenthal (2000) when dealing with samples that include both
continuous and dichotomous data. The effect size was recorded based upon an own-
race bias in juror decision-making, such that positive effect sizes demonstrate sup-
port for the hypothesis of disparate treatment of other-race defendants. Calculation
of effect sizes involved the ratio of the exact test statistic (taken from the study) to
the exact test statistic summed with the degrees of freedom for continuous data or
the ratio of the exact test statistic to the sample size for dichotomous data (see Eqs. 1
and 2 below, respectively). All r
s were then converted to Fisher’s z
to control for
Racial Bias 627
skewness in estimating the true population parameter (see Eq. 3 below).
r = SQRT((t
2
)/(t
2
+ df)) (1)
r = SQRT(χ
2
/N)(2)
z
= 0.5 × log
e
[(1 + r)/(1 − r)] (3)
Finally, to better compare our results with that of the two previous meta-analyses
(Mazzella & Feingold, 1994; Sweeney & Haney, 1992), aggregate effect sizes were
subsequently converted to Cohen’s d effect size (see Eq. 4 below):
d = (2 × r)/(1 − r
2
)(4)
RESULTS
Racial Bias in Juror Verdicts
Effect sizes were calculated with the hypothesis that mock jurors would exhibit
an in-group bias in decision-making, such that individuals would be more lenient on
defendants of their own racial group than defendants of another racial group. Thus,
a positive effect size indicated disparate treatment of racial out-group members.
In testing the racial bias hypothesis in juror verdicts, 34 studies (k = 46) met the
criteria for inclusion, representing the responses of 7,397 participants. A weighted
effect size analysis indicated a small, yet significant, effect of racial bias, z
= .046,
d = .092, Z = 3.93, p <.001, N
fs
= 212, demonstrating that participants were more
likely to render guilt judgments for other-race defendants than for defendants of
their own race. The homogeneity test statistic for this sample was significant, Q =
279.28, p <.001, indicating that the variance of the sample exceeded that expected
on the basis of sampling error, and suggesting that a moderator analysis could be
performed. Appendix A provides a listing of study characteristics and effect sizes
for the sample of studies.
Six moderator variables were identified for the present sample, including race
of the participant, sample type, dependent measure scale, presence v. absence of
jury instructions, publication status, and date of study. A fixed-effects regression
model was used to determine if these potential moderator variables influenced the
racial bias effect. The frequency of moderators across studies and results of this
analysis are displayed in Table 1. Significant moderator effects were observed for
race of participant, type of dependant measure, presence v. absence of instructions,
and date of study. The direction of effects indicated that racial bias was more pro-
nounced in the following conditions: for Black participants; when a continuous mea-
sure of guilt was utilized; when jury instructions were not provided; and in studies
conducted or published in the 1970s.
628 Mitchell, Haw, Pfeifer, and Meissner
Table 1. Moderator Analysis for Verdict Decisions
Moderator Zk Level z
d
Date of publication 5.25
∗
6 1970s 0.197 0.404
7 1980s 0.031 0.062
13 1990s −0.032 −0.064
20 2000s 0.029 0.058
Publication status 0.76
40 Published 0.047 0.094
6 Unpublished 0.025 0.050
Sample 1.87
7 Community −0.008 −0.016
37 Student 0.053 0.106
Instructions 2.53
∗
26 No Instructions 0.078 0.157
18 Instructions 0.016 0.032
Race of participant 5.92
∗
10 Black 0.208 0.428
36 White 0.014 0.028
Dependent scale 3.14
∗
18 Continuous 0.075 0.151
28 Dichotomous 0.017 0.034
∗
p ≤ .01.
Racial Bias in Sentencing Judgments
In examining the effect of racial bias on sentencing judgments, 16 studies
(k = 20) met the criteria for inclusion representing the responses of 3,141 partic-
ipants. A weighted effect size analysis indicated a significant effect of racial bias,
z
= .092, d = .185, Z = 5.10, p <.001, N
fs
= 25, demonstrating that participants
were more likely to render longer sentences for other-race defendants. However,
given the small number of null studies required to invalidate the significant effect
of racial bias in sentencing decisions, this effect should be viewed with caution. The
homogeneity test statistic for this sample was significant, Q = 122.81, p <.01, indi-
cating that the variance of the sample exceeded that expected on the basis of sam-
pling error, and suggesting that a moderator analysis could be performed. Appendix
B provides a listing of study characteristics and effect sizes for the sample of studies.
A fixed-effects regression model was used to determine if the potential modera-
tor variables discussed previously influenced the racial bias effect in sentencing judg-
ments. The moderator variables and sample characteristics for this analysis included
race of the participant, sample type, publication status, and date of study. We did
not include instructions as a moderator in this analysis due to the fact that all stud-
ies employed standard jury instructions (e.g., burden of proof, case law) rather than
sentencing guidelines. The frequency of moderators and results of a fixed-effects
regression analysis are displayed in Table 2. Significant moderator effects were ob-
served for race of participant, sample type, and publication status. The direction
of effects indicated that racial bias was more pronounced in the following condi-
tions: for Black participants; when community members were participants; and in
published studies.
Racial Bias 629
Table 2. Moderator Analyses for Sentencing Decisions
Moderator ZkLevel z
d
Date of publication 1.46
2 1970s 0.075 0.151
3 1980s 0.091 0.183
7 1990s 0.027 0.054
8 2000s 0.241 0.501
Publication status 2.64
∗
18 Published 0.103 0.207
2 Unpublished −0.009 −0.018
Sample 3.22
∗
5 Community 0.192 0.394
15 Student 0.059 0.118
Race of participant 3.38
∗
3 Black 0.339 0.731
17 White 0.048 0.096
∗
p ≤ .01.
DISCUSSION
The results of this meta-analysis underscore the complexity of racial bias in ju-
ror decision-making. Specifically, the analysis of research on this issue indicates a
small, but significant, effect for racial bias in both verdict (d = .092) and sentenc-
ing (d = .185) decisions. It is worth noting that, although small, these effects were
somewhat larger than those observed in the two previous meta-analyses (Mazzella
& Feingold, 1994; Sweeney & Haney, 1992). They are, however, consistent with
the Sweeney and Haney (1992) findings regarding sentencing, rather than those of
Mazzella and Feingold (1994) in which no racial bias effect was observed. The fail
safe N for verdict underscores the reliability of the effect, with over 200 null studies
required to eliminate the racial bias effect. Despite the larger effect size, however,
the fail safe N for sentencing decisions was only 25, suggesting that the effect may
not be robust. Further research involving sentencing decisions appears warranted.
Results also indicated that several moderator variables significantly influenced
the size of the racial bias effect. The moderator analysis demonstrated that racial
bias in juror verdict decisions was more prominent in Black participants than in
White participants, when a continuous measure of guilt was employed (as opposed
to a dichotomous “guilty” v. “not guilty” measure), when jury instructions were not
provided to jurors prior to a verdict decision, and in studies conducted or published
in the 1970s. Consistent with verdict decisions, racial bias in sentencing decisions
were also influenced by the race of the participant, such that Black participants
showed larger effects than White participants. It is interesting to note that our mod-
erator analysis resulted in a significant effect of participant race similar to that of
Sweeney and Haney (1992), in which specification of the participants’ race resulted
in a larger effect size. Our definition of racial bias, along with the significant effect of
participant race, however, suggests that the size of the effect found in Sweeney and
Haney (1992) may be due to the fact those “unspecified” participants were show-
ing racial bias towards the White defendants. Thus, the racial bias shown by White
630 Mitchell, Haw, Pfeifer, and Meissner
participants was “weakened” by the racial bias shown by other participants, rather
than a lack of bias overall.
The race of participant moderator effect is one that deserves careful consid-
eration. On the one hand, only nine of the samples involved Black participants
and seven of those nine studies failed to provide instructions and involved contin-
uous guilt measures (conditions that appear to promote the racial bias effect). On
the other hand, there is evidence from the social psychological literature that the
self-concept of ethnic minorities may be more dependent upon perceptions of race
than the self-concept of the ethnic majority. For example, Clark (1985) reported
that Black participants’ self-concept was related to their ratings of Blacks, whereas
the same relationship was not found with White participants. Grier and Deshpande
(2001) further reported that Black participants placed a higher importance on race
than did White participants, and that this racial importance predicted consumer de-
cisions. It is possible that group identity may be stronger among minorities than
among majority race individuals, making minority jurors more likely to exhibit an
own-race bias. Future research should examine this as a possible mediating variable
for the racial bias effect.
Practical Significance of Findings
The fact that use of instructions and type of dependent measure influenced
the size of the racial bias effect in verdict decisions also deserves some discussion.
More specifically, these results underscore the importance of considering the prac-
tical significance of the racial bias effect and the limitations of the current research
literature before drawing conclusions regarding the effect of racial bias in everyday
juror decision-making. One question to consider when reviewing the present find-
ings is whether the effect of racial bias persists when ecologically valid conditions
are considered. Examination of the moderating variables in the current study in-
dicates that racial bias does indeed decrease when ecologically relevant procedures
are used. More specifically, our results indicated that when a dichotomous guilt scale
was used or when standard jury instructions were presented, the effect of racial bias
was non-significant (z
= .017, d = .034, Z = .93, and z
= .016, d = .032, Z = .83,
respectively). Furthermore, studies that used both a dichotomous guilt scale and
standard jury instructions also demonstrated a non-significant effect of racial bias
(d = .040, Z = .83). These findings suggest that the racial bias effect in judgments of
guilt may be less pronounced when procedures match those in the real world, as has
been suggested by previous researchers (e.g., Pfeifer & Ogloff, 1991).
Despite the above finding, additional empirical research is needed before one
may conclude with absolute certainty that racial bias is eliminated with these eco-
logically relevant procedures, as it is unclear how closely these conditions actually
reflect the real world. For example, the studies in this sample provided limited in-
formation regarding the instructions employed, frequently referring to them only
as “jury instructions.” We interpreted this to indicate instructions on burden of
proof, reasonable doubt, and some form of case law. It is unlikely, however, that
each study included all of the language provided in standard case law instructions.
Racial Bias 631
Future researchers should be sure to clarify the instructions used and the manner
in which they were generated (standard pattern, modified standard pattern, etc.).
Also, although actual jurors never provide guilt judgments on a continuous scale,
they may be provided with multiple verdict options (first degree murder, second
degree murder, manslaughter, etc.). Future research should focus on these vari-
ations and more ecologically relevant judgment options in the context of racial
bias.
Additionally, it is possible that other variables may increase the likelihood of
racial bias. For example, the current analysis indicates that community members
tended to demonstrate more racial bias than student samples for sentencing deci-
sions (but not guilt decisions). This finding, however, should be interpreted with
caution given that few jurisdictions allow jurors to recommend sentences. The one
exception to this involves capital cases, in which jurors make the determination of
life or death sentences. It is important to note, however, that only one of the studies
included in the current meta-analysis utilized a death penalty decision as a depen-
dent measure (i.e., Dunn, Willis-Esqueda, & Schopp, 2004).
Furthermore, it is important to note that all of the effect sizes calculated for this
meta-analysis examined racial bias in jurors’ decisions (or individual-level analysis)
given the limited number of studies using juries (or group-level analysis). There-
fore, we cannot conclude from the current findings whether racial bias is produced
when jurors make decisions as a group. It is possible that the process of delibera-
tion could increase or decrease racial bias. The limited research that has been con-
ducted, for example, suggests that the degree of racial bias may be influenced by
the race of the foreperson, with a Black foreperson resulting in less guilt being at-
tributed to Black defendants (see Foley & Pigott, 2002). Further research appears
warranted.
Another issue of practical significance regards the manner in which racial bias
is defined. The current meta-analysis revised previous definitions to include the dis-
parate treatment of a racial out-group member in order to examine whether racial
bias might be present in racial minority juror decisions as well as racial majority
juror decisions. As noted above, the current synthesis found that participant race
moderated the effect of racial bias such that Black participants were more likely
to display racial bias than were White participants. We believe that the examina-
tion of multiple racial groups is important for understanding racial bias and should
not be overlooked. A growing number of minorities are likely to participate in the
legal system as jurors or victims. One in eight people living in the United States
(13.3%), for example, are Hispanic, while Blacks make up 12.3% of the population
(Yax, 2002). Even with the possibility that minorities are underrepresented on jury
rolls, their increasing representation in the population should lead to increases in
juror service. As researchers, we increase the ecological validity of our analysis by
examining the responses of multiple racial groups, especially those with large rep-
resentations in our population. While the current literature base primarily involves
Black and White racial groups, future researchers should expand their exploration
of participant race or ethnicity to consider other racial groups, particularly Hispan-
ics (see Stephan & Stephan, 1986).
632 Mitchell, Haw, Pfeifer, and Meissner
Additional Moderator Variables
In considering moderators of the racial bias effect, there were several other
variables that we would have liked to explore. For example, Sweeny and Haney
(1992) included crime type (i.e., rape v. other) and victim race (i.e., victim race spec-
ified v. not specified) as moderators in their analysis, finding that while crime type
did not serve as a moderator, specification of victim race did moderate racial bias.
Mazzella and Feingold (1994) also found a significant effect of socioeconomic status
(SES) on juror verdicts and punishment decisions (although they did not examine
SES as a moderator of the racial bias effect). These variables—SES, crime type, and
victim race—would appear to be quite relevant moderators to the racial bias effect.
Unfortunately, our sample of studies precluded our ability to parse the data to ex-
amine these moderators. Given the importance of these variables (particularly vic-
tim race) in archival studies, the existing literature should be augmented. Williams
and Holcomb (2001), for example, found that victim race—not defendant race—
influenced death sentences in Ohio. The largest percentage of defendants receiving
the death penalty involved Black defendants accused of killing a White victim, fol-
lowed by White defendants accused of killing a White victim. This is consistent with
the Baldus et al. (1983, 2002) findings, in which victim and defendant race interacted
to suggest that Black defendants were most disadvantaged when they were accused
of crimes against White victims. Mustard (2001) further found that defendant in-
come influenced departures from the federal sentencing guidelines, with low-income
defendants resulting in harsher sentences. The importance of these variables, as well
as their potential for complex interactions, has also been seen in previous empirical
research (e.g., Foley & Chamblin, 1982; Gordon, 1990; Wuensch et al., 2002). Fu-
ture research should examine the variables of victim race, socioeconomic status,
and crime type within the context of racial bias.
Other Decision Points
Given the results of this meta-analysis, it is important to once again consider
the myriad of decision “steps” leading a person to incarceration. For the purposes
of this analysis, we will focus on the commission of a crime, the police officer’s de-
cision to arrest, the prosecutor’s decision to charge/offer a plea, the jury’s verdict,
and the judge’s sentence. Although we did find a racial bias effect in White juror
decisions, it appears too small to explain the large discrepancy in Black and White
incarceration rates (e.g., Austin & Allen, 2000). Recent research has shown that
Blacks and Whites commit some offences at equal rates (e.g., narcotics violations)
(Human Rights Watch, 2000). This suggests that some of the discrepancy in incar-
ceration must occur at other stages in the criminal justice process (i.e., in the decision
to arrest, charge/offer of a plea, and/or sentence). We know of no studies that look
at these three steps from an experimental paradigm. There is some experimental
evidence, however, suggesting that race may influence decisions to shoot a suspect.
Plant and Peruche (2005), for example, had law enforcement officers view simula-
tions of Black and White males holding either a neutral object (i.e., wallet) or a gun.
The law enforcement participants were asked to indicate whether they would shoot
Racial Bias 633
or not shoot the person depicted. Plant and Peruche found that race influenced the
likelihood of shooting in the early trials, with Black males being shot more often.
There is also archival evidence suggesting that race does influence a judge’s sen-
tence. As discussed previously, Mustard (2001) found that race was a factor when
federal judges’ departed from the sentencing guidelines. Future research should be-
gin to consider the possibility that racial bias may occur at a variety of points within
the legal system.
CONCLUSIONS
Based upon the current meta-analysis, it appears that the effect of racial bias in
juror decision-making is small, yet reliable. Furthermore, we found support for the
conclusion that ecologically valid procedures (such as jury instructions and verdict
scale properties) appear to reduce the effect of racial bias in verdict decisions (cf.
Pfeifer, 1990). One caveat to these findings is the simulated nature of the studies.
Although researchers attempt to design studies that can generalize to the real world,
it is unclear if these findings reflect the behavior of actual jurors (see Bornstein,
1999). Although any evidence of racial bias in the legal system is discouraging, it is
worth noting that these results are consistent with several archival studies of actual
juries. Although it is impossible to ever fully compare simulated studies with real
trial outcomes, the similarities across archival and empirical research suggest that
these results may be generalizable.
This meta-analysis was also the first to consider the role of participant race,
and to move beyond the inference that racial bias may only be displayed by White
(or majority race) jurors. Rather, the current analysis proposed an inter-group defi-
nition of racial bias and found the opposite effect—namely, that Black participants
were more likely to demonstrate racial bias in both verdict and sentencing decisions.
Future research is warranted in examining the inter-group processes governing the
racial bias effect in juror decision-making, particularly in the areas of victim race and
the corollaries between socioeconomic status, crime stereotypes, and race of the de-
fendant. Further attention should also be given to potential psychological processes
that may moderate the racial bias effect, such as strength of jurors’ racial group iden-
tity. Research in these areas will advance our understanding of both the theoretical
basis of the effect and the ecological validity of conclusions that are drawn. Overall,
the racial bias literature has advanced considerably in the last decade and we hope
to see continued revision and progression in future research.
REFERENCES
∗
Abwender, D. A., & Hough, K. (2001). Interactive effects of characteristics of defendant and mock juror
on U. S. participants’ judgment and sentencing recommendations. Journal of Social Psychology, 141,
603–615.
Alexander, R. Jr., & Gyamerah, J. (1997). Differential punishing of African Americans and Whites who
possess drugs: A just policy or a continuation of the past? Journal of Black Studies, 28, 97–111.
*Asterisks denote studies included in meta-analysis.
634 Mitchell, Haw, Pfeifer, and Meissner
Austin, R. L., & Allen, M. D. (2000). Racial disparities in arrest rates as an explanation of racial disparity
in commitment to Pennsylvania’s prisons. Journal of Research in Crime and Delinquency, 37, 200–
220.
∗
Bagby, R. M., Parker, J. D., Rector, N. A., & Kalemba, V. (1994). Racial prejudice in the Canadian legal
system. Law and Human Behavior, 18, 339–350.
Baldus, D. C., Pulaski, C., & Woodworth, G. (1983). Comparative review of death sentences: An empir-
ical study of the Georgia experience. Journal of Criminal Law and Criminology, 74, 661–754.
Baldus, D. C., Woodworth, G., Zuckerman, D., Weiner, N. A., & Broffitt, B. (1998). Race discrimination
and the death penalty in the post-Furman era: An empirical and legal analysis with recent findings
from Philadelphia. Cornell Law Review, 83, 1638–1770.
Baldus, D. C., Woodworth, G., Grosso, C. M., & Christ, A. M. (2002). Arbitrariness and discrimination in
the administration of the death penalty: A legal and empirical analysis of the Nebraska experience
(1973–1999). Nebraska Law Review, 81, 486–775.
Bornstein, B. H. (1999). The ecological validity of jury simulations: Is the jury still out? Law and Human
Behavior, 23, 75–91
∗
Bottoms, B. L., Davis, S. L., & Epstein, M. A. (2004). Effects of victim and defendant race on jurors’
decisions in child sexual abuse cases. Journal of Applied Social Psychology, 34, 1–33.
∗
Brown, K. P. (2001). Racial bias in criminal sentencing and verdict: The role of different evaluation stan-
dards. Unpublished Doctoral Dissertation: University of Alabama.
Clark, M. L. (1985). Social stereotypes and self-concept in Black and White college students. The Journal
of Social Psychology, 125, 753–760.
∗
Dovidio, J. F., Smith, J. K., Donnella, A. G., & Gaertmer, S. L. (1997). Racial attitudes and the death
penalty. Journal of Applied Social Psychology, 27, 1468–1487.
∗
Dunn, K. F., Willis-Esqueda, C., & Schopp, R. F. (2004). Effects of victim race, perpetrator race, and jury
instructions on experimental mock capital sentencing decisions. Unpublished manuscript.
∗
Fein, S., Morgan, S. J., Norton, M. I.,
∗
Sommers, S. R. (1997). Hype and suspicion: The effects of pretrial
publicity, race, and suspicion on jurors’ verdicts. Journal of Social Issues, 53, 487–502.
∗
Field, H. S. (1979). Rape trials and jurors’ decisions: A psycholegal analysis of the effects of victim,
defendant, and case characteristics. Law and Human Behavior, 3, 261–284.
∗
Foley, L. A., & Chamblin, M. H. (1982). The effect of race and personality on mock jurors’ decisions.
Journal of Psychology, 112, 47–51.
Foley, L. A., & Pigott, M. A. (2002). Race, self presentation and reverse discrimination in jury decisions.
American Journal of Forensic Psychology, 20, 37–52.
Gordon, R. A. (1990). Attributions for blue-collar and white-collar crime: The effect of subject and de-
fendant race on simulated juror decisions. Journal of Applied Social Psychology, 20, 971–983.
∗
Gordon, R. A. (1993). The effect of strong versus weak evidence on the assessment of race stereotypic
and race nonstereotypic crimes. Journal of Applied Social Psychology, 23, 734–749.
∗
Gordon, R. A., & Anderson, K. S. (1995). Perceptions of race-stereotypic and race-nonstereotypic
crimes: The impact of response-time instructions on attributions and judgments. Basic and Applied
Social Psychology, 16, 455–470.
∗
Gordon, R. A., Bindrim, T. A., McNicholas, M. L., & Walden, T. L. (1988). Perceptions of blue-collar
and white-collar crime: The effect of defendant race on simulated juror decisions. Journal of Social
Psychology, 128, 191–197.
Grier, S. A., & Deshpande, R. (2001). Social dimensions of consumer distinctiveness: The influence of
social status on group identity and advertising persuasion. Journal of Marketing Research, 38, 216–
224.
∗
Hill, E. L., & Pfeifer, J. E. (1992). Nullification instructions and juror guilt ratings: An examination of
modern racism. Contemporary Social Psychology, 16, 6–10.
Human Rights Watch. (2000). Punishment and prejudice: Racial disparities in the war on drugs. Human
Rights Watch Report: United States, 12(2), G1202.
∗
Hymes, R. W., Leinart, M., Rowe, S., & Rogers, W. (1993). Acquaintance rape: The effect of race of
defendant and race of victim on white juror decisions. The Journal of Social Psychology, 133, 627–
634.
Joint State Government Commission. (2003). Minority representation in the jury selection process in
Pennsylvania. Retrieved April 19, 2005, from http://jsg.legis.state.pa.us/JURY.PDF.
∗
Jones, C. S., & Kaplan, M. F. (2003). The effects of racially stereotypical crimes on juror decision-making
and information-processing strategies. Basic and Applied Social Psychology, 25, 1–13.
∗
Klein, K., & Creech, B. (1982). Race, rape, and bias: Distortion of prior odds and meaning changes.
Basic and Applied Social Psychology, 3, 21–33.
Racial Bias 635
Lipsey, M. W. (1994). Identifying potentially interesting variables and analysis opportunities. In H.
Cooper & L. V. Hedges’s (Eds.), The Handbook of Research Synthesis (pp. 111–124). New York:
Russell Sage Foundation.
∗
Marcus-Newhall, A., Blake, L. P., & Baumann, J. (2002). Perceptions of hate crime perpetrators and
victims as influenced by race, political orientation, and peer group. American Behavioral Scientist,
46, 108–135.
Mazzella, R., & Feingold, A. (1994). The effects of physical attractiveness, race, socioeconomic status,
and gender of defendant and victims on judgments of mock jurors: A meta-analysis. Journal of
Applied Social Psychology, 24, 1315–1344.
∗
McGlynn, R. P., Megas, J. C., & Benson, D. H. (1976). Sex and race as factors affecting the attribution
of insanity in a murder trial. Journal of Psychology, 93, 93–99.
∗
McGowen, R., & King, G. D. (1982). Effects of authoritarian, anti-authoritarian, and egalitarian legal
attitudes on mock juror and jury decisions. Psychological Reports, 51, 1067–1074.
∗
McGuire, M. V., & Bermant, G. (1977). Individual and group decisions in response to a mock trial: A
methodological note. Journal of Applied Social Psychology, 7, 220–226.
Mustard, D. B. (2001). Racial, ethnic, and gender disparities in sentencing: Evidence from the U.S. fed-
eral courts. Journal of Law and Economics, 64, 285–314.
∗
Nemeth, C., & Sosis, R. H. (1973). A simulated jury study: Characteristics of the defendant and the
jurors. Journal of Social Psychology, 90, 221–229.
Pfeifer, J. E. (1990). Reviewing the empirical evidence on jury racism: Findings of discrimination or
discriminatory findings? Nebraska Law Review, 69, 230–250.
∗
Pfeifer, J. E., & Bernstein, D. J. (2003). Expressions of modern racism in judgments of others: The
role of task and target specificity on attributions of guilt. Social Behavior and Personality, 31, 749–
766.
∗
Pfeifer, J. E., & Ogloff, J. R. P. (1991). Ambiguity and guilt determinations: A modern racism perspec-
tive. Journal of Applied Social Psychology, 21, 1713–1725.
Pfeifer, J. E. & Ogloff, J. R. P. (2003). Mock juror ratings of guilt in Canada: Modern racism and ethnic
heritage. Social Behavior and Personality, 31, 301–312.
Pigott, T. D. (1994). Methods for handling missing data in research synthesis. In H. Cooper & L. V.
Hedges’s (Eds.), The Handbook of Research Synthesis (pp. 163–175). New York: Russell Sage Foun-
dation.
Plant, E. A., & Peruche, B. M. (2005). The consequences of race for police officers’ responses to criminal
suspects. Psychological Science, 16, 180–183.
∗
Poulson, R. L. (1990). Mock juror attribution of criminal responsibility: Effects of race and the
guilty but mentally ill (GBMI) verdict option. Journal of Applied Social Psychology, 20, 1596–
1611.
∗
Rector, N. A., & Bagby, R. M. (1995). Criminal sentence recommendations in a simulated rape trial:
Examining juror prejudice in Canada. Behavioral Sciences and the Law, 13, 113–121.
∗
Rector, N. A., Bagby, R. M., & Nicholson, R. (1993). The effect of prejudice and judicial ambiguity on
defendant guilt ratings. Journal of Social Psychology, 133, 651–659.
∗
Rickman, L. E. (1988). Juror and defendant characteristics, “crime stereotype,” and jurors’ guilty verdicts.
Dissertation: Howard University.
Rosenthal, R. (2000). Effect sizes in behavioral and biomedical research: Estimation and interpreta-
tion. In L. Bickman’s (Ed), Validity and Social Experimentation: Donald Campbell’s Legacy (Vol. 1,
pp. 121–140).
∗
Sargent, M. J., & Bradfield, A. L. (2004). Race and information processing in criminal trials: Does the
defendant’s race affect how the facts are evaluated? Personality and Social Psychology Bulletin, 30,
995–1008.
∗
Shaw, J. I., & Skolnick, P. (1995). Effects of prohibitive and informative judicial instructions on jury
decisionmaking. Social Behavior and Personality, 23, 319–326.
∗
Sommers, S. R., & Ellsworth, P. C. (2000). Race in the courtroom: Perceptions of guilt and dispositional
attributions. Personality and Social Psychology Bulletin, 26, 1367–1379.
∗
Sommers, S. R., & Ellsworth, P. C. (2001). White juror bias: An investigation of prejudice against
Black defendants in the American courtroom. Psychology, Public Policy, and Law, 7, 201–
229.
Stephan, C., & Stephan, W. G. (1986). Habla Ingles? The effects of language translation on simulated
juror decisions. Journal of Applied Social Psychology, 16, 577–589.
∗
Sunnafrank, M., & Fontes, N. E. (1983). General and crime related racial stereotypes and influence on
juridic decisions. Cornell Journal of Social Relations, 17, 1–15.
636 Mitchell, Haw, Pfeifer, and Meissner
Sweeney, L. T., & Haney, C. (1992). The influence of race on sentencing: A meta-analytic review of
experimental studies. Behavioral Sciences and the Law, 10, 179–195.
Ugwuegbu, D. C. (1973). Is justice color-blind? The racial factor in jury attribution of criminal responsi-
bility. Doctoral Dissertation: Kent State University.
∗
Wayne, J. R. H. (1998). The effects of cross-race and same-gender interactions on juror decisions in a
sexual harassment case. Unpublished dissertation. University of Georgia.
Wilder, D. A., & Shapiro, P. (1991). Facilitation of outgroup stereotypes by enhanced ingroup identity.
Journal of Experimental Social Psychology, 27, 431–452.
Williams, M. R., & Holcomb, J. E. (2001). Racial disparity and death sentences in Ohio. Journal of
Criminal Justice, 209, 207–218.
∗
Wuensch, K. L., Campbell, M. W., Kesler, F. C., & Moore, C. H. (2002). Racial bias in decisions made
by mock jurors evaluating a case of sexual harassment. Journal of Social Psychology, 142, 587–600.
Yax, L. K. (2002). The population profile of the United States: 2000 (Internet Release). Retrieved August
6, 2004 from http://www.census.gov/population/www/pop-profile/profile2000.html.
APPENDIX A: STUDY CHARACTERISTICS AND EFFECT SIZES FOR VERDICT DECISIONS
Experiment
Study No. N Status
a
Sample
b
Instruction
c
Race
d
DV
e
z
f
Abwender & Hough (2001) 63 P S N W C 0.08
89 P S N B C 0.23
Bagby et al. (1994) 243 P S M W D 0.17
Bottoms et al. (2004) 2 228 P S Y W D 0.03
Brown (2001) 261 U S N W C 0.03
86 U S N B C 0.07
Fein et al. (1997) 79 P S Y W D 0.28
Foley & Chamblin (1982) 171 P S N W C 0.00
20 P S N B C 0.00
Foley & Pigott (2002) 133 P B N W C 0.00
44 P B N B C 0.00
Gordon (1993) 144 P S N W C 0.05
Hill & Pfeifer (1992) 125 P S Y W C 0.23
Hymes et al. (1993) 78 P S N W C 0.00
Jones & Kaplan (2003) 360 P S Y W D 0.02
Klein & Creech (1982) 2 133 P S Y W D 0.02
Marcus-Newhall et al. (2002) 1 133 P S N W C 0.16
2 104 P C N W C 0.25
McGlynn et al. (1976) 208 P S Y W D 0.08
McGowen & King (1982) 360 P S Y W C 0.00
McGuire & Bermant (1977) 226 P C Y W D 0.00
Pfeifer & Bernstein (2003) 236 U S Y W C −0.04
Pfeifer & Ogloff (1991) 61 P S Y W C −0.02
57 P S Y W D 0.11
61 P S N W C 0.39
65 P S N W D 0.22
Poulson (1990) 94 P S N W D 0.33
Rector et al. (1993) 245 P S M W D 0.00
Rickman (1988) 239 U S N W D 0.12
248 U S N B D 0.07
Sargent & Bradfield (2004) 1 240 P C N W C 0.16
2 147 P C N W C 0.00
Shaw & Skolnick (1995) 316 P S Y W D 0.11
Sommers & Ellsworth (2000) 1 33 P S Y W C 0.13
1 29 P S Y B C 0.43
2 156 P C Y W C 0.17
2 55 P C Y B C 0.19
Sommers & Ellsworth (2001) 196 P C Y W D 0.09
Racial Bias 637
APPENDIX A: Continued
Experiment
Study No. N Status
a
Sample
b
Instruction
c
Race
d
DV
e
z
f
Sunnafrank & Fontes (1983) 2 50 P S Y W D 0.13
Ugwuegbu (1973) 244 P S N W C 0.21
186 P S N B C 0.37
244 P S N W C 0.22
186 P S N B C 0.34
Wayne (1998) 408 U S N W C 0.01
Wuensch et al. (2002) 1 161 P S N W D 0.10
2 152 P S N B D 0.26
a
Publication status: P: published; U: unpublished.
b
Sample type: S: student; C: community; B: both.
c
Instructions: Y: yes; N: no; M: manipulated.
d
Race of participant: B: Black; W: White.
e
Dependant variable: C: continuous; D: dichotomous.
f
Own-group bias is indicated by positive scores.
APPENDIX B: STUDY CHARACTERISTICS AND EFFECT SIZES FOR SENTENCING
DECISIONS
Study Experiment No. N Status
a
Sample
b
Race
c
z
d
Dovidio et al. (1997) 104 P S W 0.00
Dunn et al. (2004) 320 U S W 0.00
Field (1979) 843 P C W −0.13
Gordon (1993) 144 P S W 0.07
Gordon et al. (1988) 28 P S W 0.00
28 P S B 0.00
Gordon & Anderson (1995) 216 P S W −0.05
Hymes et al. (1993) 78 P S W 0.00
Klein & Creech (1982) 2 133 P S W −0.27
Marcus-Newhall et al. (2002) 1 103 P S W 0.04
2 104 P C W 0.07
Nemeth & Sosis (1973) 80 P S W −0.02
Rector & Bagby (1995) 42 P S W 0.00
41 P S W 0.00
Sommers & Ellsworth (2000) 1 33 P S W −0.19
129PSB−0.48
2 156 P C W −0.50
255PCB−0.72
Wayne (1998) 408 U S W 0.18
a
Publication status: P: published; U: unpublished.
b
Sample type: S: student; C: community; B: both.
c
Race of participant: B: Black; W: White.
d
Own-group bias is indicated by positive scores.
A preview of this full-text is provided by American Psychological Association.
Content available from Law and Human Behavior
This content is subject to copyright. Terms and conditions apply.
