Implicit Bias
Skip to main content

Student Projects

For further study:

  • Ian Ayres & Joel Waldfogel, A Market Test for Race Discrimination in Bail Setting, 46 Stan. L. Rev. 987 (1994).

  • David C. Baldus et. al., Racial Discrimination and the Death Penalty in the Post-Furman Era: An Empirical and Legal Overview, with Recent Findings from Philadelphia, 83 Cornell L. Rev. 1638 (1998).

  • William Bowers, The Capital Jury Project: Rationale, Design, and Preview of Early Findings, 70 Ind. L. J. 1043, 1085-1101 (1995).

  • Theodore Eisenberg & Sheri Lynn Johnson, Implicit Racial Attitudes of Death Penalty Lawyers, 53 DePaul L. Rev. 1539 (2004).

  • Jerry Kang, Trojan Horses of Race, 118 Harv. L. Rev. 1489 (2003).

  • Justin D. Levinson, Huajian Cai & Danielle Young, Guilty by Implicit Bias: The Guilty–Not Guilty Implicit Association Test, 8 Ohio St. J. Crim. L. 187 (2010).

  • Justin D. Levinson & Danielle Young, Different Shades of Bias: Skin Tone, Implicit Racial Bias, and Judgments of Ambiguous Evidence, 112 W. Va. L. Rev. 307, 319–26 (2010).

  • Tara L. Mitchell et al., Racial Bias in Mock Juror Decision-Making: A Meta-Analytic Review of Defendant Treatment, 29 Law & Hum. Behav. 621, 627–28 (2005). 

  • David B. Mustard, Racial, Ethnic, and Gender Disparities in Sentencing: Evidence from the U.S. Federal Courts, 44 J. L. & Econ. 285 (2001).

  • Michele Benedetto Neitz, Socioeconomic Bias in the Judiciary, 61 Clev. St. L. Rev. 137 (2013).

  • Jeffrey J. Rachlinski et. al., Does Unconscious Racial Bias Affect Trial Judges?, 84 Notre Dame L. Rev. 1195 (2009).

  • Rice v. Collins, 546 U.S. 333, 126 S. Ct. 969, 163 L. Ed. 2d 824 (2006).

  • Anna Roberts, (Re)forming the Jury: Detection and Disinfection of Implicit Juror Bias, 44 Conn. L. Rev. 827 (2012).

  • Robert J. Smith, Justin D. Levinson, The Impact of Implicit Racial Bias on the Exercise of Prosecutorial Discretion, 35 Seattle U. L. Rev. 795 (2012).

  • Samuel R. Sommers & Phoebe C. Ellsworth, "Race Salience" in Juror Decision-Making:  Misconceptions, Clarifications, and Unanswered Questions, 27 Behav. Sci. & L. 599 (2009).

  • State v. Addison, 2008-945, 2013 WL 5960851 (N.H. Nov. 6, 2013).

Defining Implicit Bias

Compiled by Caisa Royer, Daniel Hido, and Michael Slotnick

Implicit bias is an unconscious, unintentional bias. Unlike explicit bias, which an individual is conscious and knowledgeable of, implicit bias exists when an individual does not have direct control or understanding of their perceptions and motivations. These biases are likely formed by schemas or associations in the brain that link two ideas together (i.e. a group of people with a trait), and these associations likely form through a combination of early experiences, affective experiences, and learned cultural biases.  Implicit biases allow individuals to categorize stimuli efficiently, and these classifications allow people to easily understand and interact with their world, especially in uncertain situations.

All people have implicit attitudes and stereotypes that can direct how they relate to their environment. These biases are pervasive and can either be positive or negative. Due to the unconscious nature of implicit bias, detecting and studying its effects can be challenging. However, research has shown that people have implicit attitudes towards many different topics, such as race, gender, age, disability, and sexual orientation. Furthermore, several general patterns of bias have repeatedly been shown in the research:

  1. Implicit bias measures can often be dissociated from explicit bias measures, meaning that known bias may be different from unconscious bias

  2. Socially dominant groups often have implicit bias against subordinate groups

  3. Individuals usually have a preference for members of a category to which they belong

Demonstrating Implicit Bias


Race and gender IAT tests:

There are a variety of methods for measuring implicit bias.  The foremost method is the Implicit Association Test (IAT).  The IAT takes many forms, but the most common test demonstrates implicit racial bias by having respondent's pair words with faces.  Similarly, because racial implicit bias in legal proceedings is the primary focus of this project, a typical "Race IAT" is what is described here.

A Race IAT presents respondents with four stimuli:  white faces, black faces, "good" words, and "bad" words.  Respondents complete a "stereotype-congruent" trial, in which they sort white/good pairings from black/bad pairings.  Respondents then complete a "stereotype-incongruent" trial, in which they sort white/bad pairings from black/good pairings. 

Implicit racial bias is demonstrated by measuring the respondents’ "response latency," i.e., the difference in time it takes to complete the stereotype-congruent and the stereotype-incongruent trials.  Most white respondents demonstrate a "white preference" by completing the stereotype-congruent trial more quickly than the stereotype-incongruent trial.

Jury Case Study: State v. Addison (2013)

Implicit racial bias in the jury has been presented as an issue causing disparate sentencing for black and white defendants. One recent case in which the defense cited issues with implicit bias in the jury is State v. Addison. In the case, Michael K. Addison was convicted of capital murder after shooting a police officer and leading the police on a subsequent man hunt that lasted for 15 hours. This case was the first implementation of the death penalty in New Hampshire since the beginning of the current sentencing era in 1977.  Addison was a black man, and the police officer he killed was white.

Addison appealed the conviction and sentence. Among the 22 issues raised by the defense was the claim that race disparately impacts capital sentencing, especially in cases in which a black defendant kills a white victim. The defense asserted that "[e]xtensive research has demonstrated that in cases of an alleged murder of a white victim by a black defendant it is impossible to guarantee that race will not play a role in the decision of whether or not the defendant is sentenced to die." This argument is similar to the one made in McCleskey v. Kemp, in which the Supreme Court determined that the risk of racial discrimination in capital sentencing is not enough to overturn a guilty verdict, but rather explicit intent of discrimination is needed.

The state objected to the claim of racial discrimination and the court heard six days of expert opinion on the topic. As in in McCleskey v. Kemp, the court concluded that although the defense's witnesses may have proved that there is a risk of racial bias in capital sentencing, there was no proof that racial discrimination had in fact impacted the trial. The court stated: "[u]ndoubtedly racism, both implicit and explicit, exists, as [do] … many other types of prejudices," this "does not mean that persons who may be the subject of a particular bias should not be held accountable for their actions in the same way that others are." The court held that the social science evidence presented was "insufficient to establish [a] claim of purposeful racial discrimination under the State Equal Protection Clause." Addison's conviction and sentence was upheld, and he remains the only person on death row in New Hampshire.

Implicit Bias and Juries
Below is listed some of the evidence presented in the State v. Addison trial, as well as other research, that looks at the existence of implicit racial bias in juries.

  • Dr. Mahzarin Banaji, Ph.D. testified in State v. Addison about data from a national Race IAT database which included 1,000 participants in New Hampshire. According to the database, nationally 88% of white participants and 50% of black participants prefer white individuals over black individuals. These percentages were slightly higher for participants in New Hampshire. Dr. Banaji suggested it would be extremely hard for a black defendant to receive a fair and impartial trial in New Hampshire. However, while unpublished research links Race IAT scores with sentences imposed by mock jurors on black and white defendants, no published studies link IAT data to real jury decisions in capital trials. The court argued that without IAT data from real jurors to compare to sentencing outcomes in real cases, the IAT evidence was not sufficient to make a racial bias claim.
  • Additionally, a Guilty-Not Guilty IAT has been developed to further measure implicit views of race and innocence (Levinson, Cai, & Young, 2010). Data from the Guilty-Not Guilty IAT has shown a small implicit association between African Americans and guilty.
  • Many mock jury studies have been conducted to look at the influence of racial bias on a jury. Some of the following studies were presented in State v. Addison, though they were not found to be persuasive. Although the following studies did not measure implicit racial bias with the IAT, they all looked at the effects of race on sentencing decisions by mock jurors. The research shows a racial sentencing bias, which is often found to not be associated with traditional measures of overt racism. These studies suggest that an unconscious (or implicit) bias may exist.
    1. A meta-analysis of thirty-four juror verdict studies with defendant race manipulations found a significant in-group bias in sentencing decisions (Mitchell et al., 2005).
    2. In one study, groups of mock jurors were shown video recordings of a capital murder trials (Lynch, 2006). The race of the victim and defendant in the videos were altered so that different groups saw different race of victim/race of defendant combinations. The mock juries were found to be more likely to sentence black defendants to death than a white defendant, particularly if the black defendant had killed a white vicitm.
    3. Mock jurors were asked to read a trial summary about an interracial battery charge (Sommers & Ellsworth, 2009). The race of the defendant/victim were randomly assigned so that some participants read a case with a white defendant and others read a case with a black defendant. Furthermore, half of the participants also read a witness testimony about the defendant's minority status in an effort to create a "race salience" condition. When race was salient, the mock jurors were less likely to convict the black defendant, suggesting that being aware of racial content may activate motivation to appear nonprejudiced.
    4. Mock jurors were asked to view five photographs of a crime scene, including a surveillance camera photo of a masked gunman whose arms were visible (Levinson & Young, 2010) . The race of the defendant was altered so that half of the participants saw someone with dark skin and the other half saw with light skin. Depending on the skin color of the defendant, jurors evaluated the helpfulness of the evidence differently in determining the guilt of the defendant. Further, jurors were more likely to find a dark-skinned defendant guilty than a light skinned defendant. Assessment of guilt was not found to be correlated with scores on the Modern Racism Scale.

There are several changes that could be made to decrease the effect of implicit bias on sentencing decisions made by jurors. First, the IAT could be used during voir dire to eliminate potentially biased jurors. However this solution is often criticized for several reasons. Requiring all jurors take an IAT during voir dire could be time intensive and unrealistic. Furthermore, the IAT is not a perfect method for predicting future behavior, and some evidence suggests implicit biases can be overridden with education and motivation. Instead, some argue that the IAT could be used to serve to educate potential jurors about their implicit biases, rather than to eliminate biased jurors.

It has also been suggested that juror instructions should include a section on implicit bias in order to make jurors aware of how unconscious can play a role in decision making. Introducing a salience of racial bias may create motivation to override internal biases. Finally, exposure to counterstereotypical group members can decrease implicit biases. Therefore, further efforts to create diverse juries may naturally compensate for implicit biases.

Why Implicit Bias in Judges Matters

A recent study found that judges harbor the same kinds of implicit biases as others, and that these biases can affect their judgment and treatment of a defendant (Rachlinski et al, 2009).  A judge’s implicit bias toward a defendant can influence the judge’s decisions on setting bail and sentencing a defendant.  In bench trials, implicit bias may affect the judge’s factfinding, and even the determination of guilt or innocence. 

A judge’s implicit bias may influence the outcome of the trial in more subtle ways as well, such as rulings on objections and admissibility of evidence, and jury instructions.  At the appellate level, judges’ implicit bias may affect the determination of whether a defendant received a fair trial or sentence.

Implicit Bias in Setting Bail

Ian Ayres and Joel Waldfogel conducted a study that examined the effect of implicit bias on judges’ decisions in setting bail.  They concluded that judges routinely set bail at unjustifiably higher levels for black male defendants than for white male defendants.

Ayres and Waldfogel analyzed 1,118 arrests in New Haven, CT.  After controlling for eleven variables related to the severity of the offense, they found that judges on average set bail 35% higher for black male defendants than for their white male counterparts (Ayres & Waldfogel, 1994).

While this statistic is alarming in itself, it does not conclusively demonstrate an implicit bias in setting bail.  The purpose of bail is to deter a defendant from fleeing before trial.  Thus, judges decide at what level to set bail based on a defendant’s perceived flight risk.  It is therefore possible that there are other factors associated with race that indicate that a defendant presents a high flight risk.  For example, a black defendant may have fewer ties to the community, less stable employment, or lower socio-economic status (SES).  A judge basing the decision on setting bail on these factors would not necessarily indicate that the judge demonstrated implicit bias in making the decision.

To account for the influence of a defendant’s flight risk on the amount of bail, the researchers next analyzed the practice of bail bondsmen in the same sample.  Bail bondsmen conduct a business in which they post a defendant’s bond to the court to secure the defendant’s release.  The bondsmen charge the defendants a fee for doing so, usually a specified percentage of the amount of the bond. 

The bondsmen thus act as an insurer that the defendant will not flee while released.  Therefore, the rate that the bondsmen charge the defendants is typically based on the defendant’s perceived flight risk.  For example, if a defendant’s bail was set at $10,000, and the defendant presented a 10% flight risk, the bondsman would charge a fee of $1,000 for posting the bond.

The researchers found that, in the sample, bail bondsmen charged a 19% lower rate for black male defendants than their white male counterparts.  These data indicate that while judges set bail 35% higher for black male defendants than for their white male counterparts, this higher rate was unjustified by the defendants’ flight risk.  These data present evidence of possible implicit racial bias in judicial decision-making.

Implicit Bias in Sentencing

Numerous studies also demonstrate the effect of implicit bias on sentencing.  For example, black defendants are disproportionately subject to the death penalty (Mustard, 2001). 

Additionally, in one of the most comprehensive studies of judicial sentencing, David Mustard examined 77,236 federal offenders sentenced under the Sentencing Reform Act of 1984 (SRA).  The SRA was intended to eliminate sentencing disparities based on race, ethnicity, gender, and income.  However, after controlling for offense type, severity, criminal history, and other criminological variables, blacks, Hispanics, and other racial minorities received 12% longer sentences than white defendants (Mustard, 2001).

Mustard also examined the effect of various SES-related variables on sentencing, such as income and education, factors that are highly correlated with race.  Although these factors are not supposed to have an effect on sentencing, Mustard found that lower income and lower levels of education were significantly correlated with longer sentences.  Other studies, through the use of IAT, confirmed the presence of what Ninth Circuit Chief Judge Kozinski referred to as the "unconscious cultural elitism" of judges (Neitz, 2013).

Hope is Not Lost

Professor Rachlinski’s study, discussed supra, administered a Race IAT to 133 judges from three different jurisdictions.  The sample consisted of eighty-five white judges and thirty-three black judges.  While the black judges showed no clear racial preference, seventy-four (87.1%) of the white judges demonstrated a white preference. 

The researchers then administered a vignette depicting a criminal defendant whose race was depicted as either white or black.  The researchers measured the judges’ likelihood of convicting the defendant based on the evidence in the vignette, their confidence in their verdict, and their beliefs about the future recidivism of the defendant.
Interestingly, there was no significant effect of the white judges’ white preference on the IAT on their treatment of the defendant.  That is, white judges who exhibited a white preference on the IAT were just as likely to convict the white defendant as the black defendant.  Their confidence in their decision and views about the defendant’s future recidivism were also not influenced by race.

A likely explanation for these results is, as the authors hypothesized, that the judges caught on to what the researchers were measuring, and thus made a conscious effort to avoid letting the race of the defendant influence their decision.  These results are encouraging, because they demonstrate that judges are able to overcome implicit bias in their decision-making.

Judges frequently participate in continuing legal education programs, which could be expanded to train the judges to be aware of the presence of implicit bias and how to overcome it in their decision-making.  This could be as easy as administering an IAT.  Additionally, Rachlinski et al. suggest "auditing" programs, which would examine the way in which implicit bias had influenced past judicial behavior in order to put judges on alert in their future decisions.

Further, multiple-judge courts, particularly those with a diverse array of judges, can help to alleviate the problem.  One possible solution could be to expand the use of multi-judge courts at the trial level.  This solution may prove difficult to implement, however, because it would require a drastic restructuring of the trial courts, and it is unclear whether there would be a sufficient number of judges available to meet the demand.

However, most appellate courts are already made up of multiple judges.  However, these courts review many trial judge decisions, particularly the type of factfinding that is often influenced by implicit bias, under an abuse of discretion standard.  A potential solution would be to subject such trial judge findings to a more stringent standard of review, perhaps even de novo review. 

Implicitly Biased Attorneys

In Rice v. Collins (2006), the prosecutor struck a young black woman from a jury panel. The defendant objected that the peremptory strike was based upon the juror’s race. "As race-neutral explanations for striking Juror 16, the prosecutor said that Juror 16 had rolled her eyes in response to a question from the court; that Juror 16 was young and might be too tolerant of a drug crime; and that Juror 16 was single and lacked ties to the community. A further, more troubling part of the prosecutor's unorganized explanation was her reference to Juror 16's gender." The trial court judge did not observe the juror’s eye rolling and demeanor that the prosecutor described, but accepted the peremptory strike over the defendant’s objection.

The Supreme Court applied the test for defendant challenges of peremptory strikes from Batson v. Kentucky. The Batson challenge test is a three-step inquiry. The defendant must show that the challenge was prima facie on the basis of race. If so, the prosecutor must provide a plausible race-neutral reason for striking the juror. Then the defendant must prove the prosecutor purposefully discriminated against the juror on the basis of race while accounting for the prosecutor’s race-neutral reasons. The court held that the defendant did not meet the Batson test, and affirmed the defendant’s conviction.

In concurrence, Justice Breyer noted that the court applied the Batson test correctly, but criticized the test. One aspect of Justice Breyer’s criticism is that peremptory challenges are based off instinct, not reason. Therefore, prosecutors cannot adequately perform the second step of the Batson inquiry because prosecutors cannot adequately explain their instincts. Another aspect of Justice Breyer’s criticism is that "sometimes, no one, not even the lawyer herself, can be certain whether a decision to exercise a peremptory challenge rests upon an impermissible racial, religious, gender-based, or ethnic stereotype." Therefore, it seems that Justice Breyer advocates changing the test for peremptory challenges based on the risk of prosecutors’ implicit bias.

Robert J. Smith & Justin D. Levinson (2012) provided several practical examples of how different implicit associations can affect prosecutors’ work. One example is that implicit bias testing, especially IAT testing, shows an association between black people and aggression, and IAT testing shows an association between white people and harmless objects. This may lead prosecutors to consider self-defense claims differently based upon the race of the victim. For example, if a victim is black, a prosecutor may be implicitly biased and therefore more likely to believe that a white suspect was reasonable for believing that the victim was reaching for a gun which was actually a cell phone.

Smith and Levinson note that prosecutors may also be implicitly biased in deciding whether to charge a defendant under eighteen years old as a juvenile or as an adult. "Sandra Graham and Brian Lowery, for example, found that when [juvenile probation officers and police officers] were subliminally primed with words related to the category black, they judged an adolescent's behavior as more dispositional, of greater culpability, and more likely to lead to recidivism." Therefore, prosecutors may have similar implicit biases and decide that criminal actions by black seventeen year olds are more likely to be dispositional than the same actions by white seventeen year olds, which prosecutors are more likely to believe is based on their youth.

A continuing theme is that the race of the victim is just as important if not more important than the race of the suspect in terms of prosecutors’ implicit biases. "Alessio Avenanti (2010) "used a method called transcranial magnetic stimulation (TMS) to measure corticospinal activity level in participants who viewed short video clips of a needle entering into the hand of either a lightskinned or dark-skinned person . . . Avenanti found that region-specific brain activity levels were higher when Caucasian-Italian participants viewed the clip of a light-skinned participant experiencing pain than when they saw a clip of a dark-skinned target being subjected to pain." Prosecutors may implicitly empathize with white victims, and therefore be prone to act more harshly while prosecuting the defendant to gain retribution for the victim.

Another important association from the Smith and Levinson article that may affect prosecutors is the implicit association between black people and "lack of respect for law enforcement." Therefore, this bias may implicitly affect prosecutors’ jury selection methods. According to Smith and Levinson,

If a prosecutor questions a prospective black juror, the simple act of even talking to that person might activate . . . negative implicit attitudes, causing the prosecutor to think or feel negative thoughts about the juror. The prosecutor might project this negativity through body language and gestures, which could, in turn, cause jurors to avoid eye contact, provide awkward or forced answers that make the juror appear less intelligent, or simply fidget and look nervous. Thus, even accurate raceneutral [sic] behavior descriptions might stem from racialized assessments (albeit, without conscious thought) of the characteristics of individual jurors.

Perhaps juror 16 in Rice v. Collins did not actually roll her eyes as the prosecutor claimed but, rather, involuntarily responded to the prosecutor's negative body language.

One area of the law that has been subject to extensive racial implicit bias research is on use of the death penalty. David C. Baldus et. al., (1998), survey evidence on racially discriminatory use of the death penalty since the landmark case of Furman v. Georgia, and find that the most useful complication of this data was done by the General Office of Accounting in 1990. The GOA evaluated twenty-eight empirical death penalty studies during the 1970s and 80s. The GOA found that "[t]he evidence for the race of victim influence was stronger for the earlier stages of the judicial process (e.g., prosecutorial decision to charge defendant with a capital offense, decision to proceed to trial rather than plea bargain) than in later stages. This was because the earlier stages were comprised of larger samples allowing for more rigorous analyses." While prosecutors may not be more subject to implicit bias than other actors in death penalty cases, their actions affect a case more often because they act earlier in the case, before the case may end before completion.

Finally, prosecutors are not the only type of attorney whose implicit bias may affect death penalty cases. Theodore Eisenberg & Sheri Lynn Johnson (2004) study the administration of the IAT test on death penalty defense attorneys and law students. The results of the study showed that death penalty defense attorneys show the same level of implicit bias as law students. White and Asian men had the largest implicit preference for white, white and Asian women had a somewhat smaller implicit preference for white, and black subjects had a significantly smaller implicit preference for black.

These articles indicate a few remedies that attorneys might take to try to reduce implicit bias. Eisenberg and Johnson suggest that capital defense attorneys should consider whether they have implicit racial biases themselves and try to ensure that it does not affect how they react during a case. Smith and Levinson suggest that prosecutors’ offices train their prosecutors about implicit biases. Smith and Levinson also suggest a more creative approach; prosecutors’ offices should hire a more racially diverse group of prosecutors. The authors note that studies have shown that students exposed to counter-stereotypical teachers such as women engineering professors have reduced implicit bias. Furthermore, jury research shows that group decision-making exceeds individual decision-making. The authors’ conclusion is that a more diverse group of prosecutors that work together on cases may be able to reduce their collective implicit bias.