Hearings & Trials

The Setting: This research investigated whether a de-biasing jury instruction is effective at having jurors recognize their own implicit bias and overcome it during deliberation. The Sixth Amendment of the U.S. Constitution and Article I, Section 12 of the Utah Constitution guarantee the right to an impartial jury for citizens of Utah. The Salt Lake Justice Court (“SLJC”) used debiasing jury instructions as one mechanism to deliver on this right. The SLJC used debiasing jury instructions when the judge determined their presence necessary either sua sponte or upon request of one or both parties to an action. The instruction is intended to ensure that each juror remains aware that they may hold implicit biases, to identify any such biases, and to ensure those biases do not enter the decision-making process. Some research suggests implicit biases are unconscious. Other research suggests that they are conscious and, as such, labeling them as “implicit” is a misnomer.

The A2J Lab worked with SLJC to develop and launch an evaluation of debiasing jury instructions using a randomized control trail (“RCT”)

The Problem: The literature presents contrary conclusions about the prevalence of bias in juries and jurors individually, as well as the composition metrics that suggest greater bias will exist. In the face of such contradictions, it may be difficult to identify debiasing interventions. The mechanism tested in this evaluation was a debiasing jury instruction. Research is inconclusive on the usefulness of a debiasing instruction vis-a-vis its stated purpose – to debias. Our literature review revealed no field RCTs, and thus conclusions are necessarily limited. Evaluations of the effects of jury instructions thus far used mock juries.

The Questions: This research investigates whether a debiasing jury instruction causes jurors to recognize their own implicit bias and overcome it during deliberation. It is a first of its kind study utilizing an RCT on actual juries, rather than mock juries.

The Study: Our study hoped to build on the existing inconclusive research about the effect of jury debiasing. We measured bias with the Implicit Association Test (“IAT”).

During the study, after prospective jurors completed the standard jury questionnaire, SLJC asked them to take the race IAT. Prospective jurors did so using court-provided iPads preloaded with the hyperlink to the baseline IAT. Empaneled jurors retook the same Race IAT after deliberation, again on court-provided iPads. The research team obtained access to the race IAT via a contract with Project Implicit. The research team discarded baseline race IAT data for jurors not empaneled.

Random assignment occurred after empanelment but before the commencement of the trial. According to the assignment, trials either receive the debiasing instruction (“Instruction Present Group”) or not (“Instruction Not Present Group”).

After rendering the verdict and completing the post-trial Race IAT, SLJC provided jurors with a debriefing statement explaining their participation in the study and how to opt out if they so choose.

What We Learned: The study was delayed by COVID-19 pandemic court closures. After courts reopened for trials, the study ran for about one year. However, the volume of trials post-reopening decreased from the volume pre-pandemic. The A2J Lab recommended closing the study after it became clear that no realistic time period would allow accumulation of a number of trials sufficient for substantive inference.

There were no statistically significant differences in change in IAT scores, trial verdicts, or deliberation time based on group, and IAT scores did not change significantly from pre- to post-trial. IAT scores bore no association with juror gender, race, or Juror Attitude Questionnaire responses. Providing jurors with instructions regarding implicit bias thus did not appear to affect their deliberations or the verdicts they ultimately deliver and does not appear to affect their IAT performance. However, the sample size was too small to generate casual evidence, and as such, these conclusions are really only examples of what an adequately powered study might produce as well as hypotheses for further research.

The study was under-enrolled and, ultimately because of that, the distribution of the instructions heavily skewed. Therefore, the research team makes no assignment of causation to findings but rather hypothesizes or suggests suspicions. The first of those suspicions is that the instructions alone may not debias a jury. The IAT scores pre- and post- show no change. However, in addition to the under-enrollment caveat, there may be others. First, the demographic analysis shows that the juries in this study were generally highly educated and mostly white. So, even if causal, results likely do not generalize to more diverse juries. Second, there may be a question about whether the IAT is the best measurement of bias. All of this taken together, additional research is needed, continuing with a randomized design, with a larger, more diverse sample and multiple methods of measuring bias, including using the IAT.

Ultimately, even if the instruction simply does not fulfill its intended mission – to debias the jury – it may serve another useful purpose – to instill trust in the judicial system by serving as a values statement and a commitment to fairness.

Research Team:
D. James Greiner, Faculty Director, Access to Justice Lab; S. William Green Professor of Public Law, Harvard Law School
Renee L. Danser, Associate Director of Research and Strategic Partnerships, Access to Justice Lab
Salt Lake Justice Court (UT)

Resources:
Remote Testimonial Fact Finding, Danser RL, Greiner DJ, Guo E, Koulton E., In: Engstrom DF, ed. Legal Tech and the Future of Civil Justice. Cambridge University Press; 2023:93-111.

The Setting: The intervention offers counsel at first appearance for all offenses or not, selected by random assignment, in two counties in Texas.  If, by random assignment, counsel is present at first appearance, counsel will engage in initial client contact prior to the formal court appearance. During this time the attorney will engage in typical attorney-client behavior, studying case details and interviewing the defendant prior to representing the defendant during first appearance.  If a defendant is randomly assigned to the status quo, they proceed with no counsel at their first appearance.

The Problem: Despite the potential for pretrial incarceration, most states do not supply an attorney to advocate for a defendant’s release at this “first appearance.”[1] There is a presumption of innocence pretrial, and that presumption requires special circumstances for detention. Yet, without the assistance of counsel, defendants are ill-equipped to challenge the prospect of their own detention.[2] Thus, when considered alongside the idea that the first appearance amounts to a “trial-like confrontation,”[3] a growing chorus of legal scholarship contends that the first appearance is crucial to the integrity of an individual’s defense.[4] While the United States Supreme Court has not yet recognized a constitutional right to CAFA, specifically,[5] there is a growing sentiment that such counsel should be considered among Sixth Amendment guarantees. Following Gideon v. Wainwright[6] and its progeny indicating, inter alia, that the right to counsel extends to criminal proceedings in which a defendant faces a loss of freedom,[7] legal scholars argue that the potential for pre-trial detention is just such a loss of freedom. The theory runs that there is a risk of pre-trial detention, a loss of freedom, at the first appearance and thus this confrontation fits squarely within the directive of Gideon and Argersinger. Although systematic, rigorous analysis of the effects of CAFA is rare, some evidence regarding these programs exists.[8] While these extant studies are valuable, there is little causal evidence regarding the effects of CAFA. 

The Questions:

1. Does CAFA have an impact on bail and pretrial release conditions and decisions? Specifically, do defendants with CAFA exhibit higher rates of personal recognizance bonds and lower bond amounts than defendants without?  

2. For those who are released pre-trial, do defendants with CAFA exhibit lower FTA rates than those without such representation? 

3. Does CAFA impact case disposition, sentencing, or pre- or post-trial recidivism? 

4. Does CAFA have an effect on arrestees’ attitudes toward the legal system?  

5. Finally, what is the cost-benefit ratio for jurisdictions interested in piloting this program? 

The Study: The availability of counsel was randomized by days conditional over weeks. In other words, the days for which counsel was present during a given seven-day period was randomly varied across weeks – this protocol was necessary to ensure that systematic variance in criminological or judicial behaviors associated with certain weekdays was randomly distributed. This resulted in treating roughly half of all new cases during the enrollment period.  

What We Learned: Our findings show that the defense counsel’s impact on defendant outcomes is mixed. In both jurisdictions, defense counsel lowers the bond amount a defendant must pay to be released, but defense counsel only affects the type of the bond in one county (changing it from financial to non-financial). Attorneys do not make own (or personal) recognizance release more likely, and are successful in affecting the days to release in only one jurisdiction, with those effects on the mild side. Meanwhile, bond conditions increased in both number and severity in one jurisdiction but not in the other. Across both jurisdictions, there is no impact on recidivism, failure to appear (only measured in one jurisdiction), and disposition outcomes for defendants.

Overall, we conclude that attorneys had an impact on bond amounts, but depending on the jurisdiction may or may not impact bond type and release conditions and outcomes for defendants. These findings provide evidence that the impact of CAFA is (i) more limited than suggested in previous studies with less credible designs, and (ii) contingent on jurisdictional factors that are as yet poorly understood. We see our study as a step toward a better understanding of the role defense counsel plays at magistration, but given the contrary results we observe, more research is needed.

Research Team: The A2J Lab partnered with the Public Policy Research Institute at Texas A&M University to conduct the evaluation which occurred in two jurisdictions in Texas.

Resources:
George Naugal, Bethany Patterson, Renee Danser, D. James Greiner, “The Causal Impact of Counsel at First Appearance: Evidence from Two Randomized Control Trials,” IZA Institute of Labor Economics, Discussion Paper No. 17712, February 2025