Jury Decision Making: Unanimity, Race, and Creating Better Juries
In a US federal court, a jury that votes 10-to-2 in favor of conviction in a criminal trial would be considered an undecided, or hung, jury. The case would be declared a mistrial and may be put aside to be tried again later. The same case, with the same jury and vote, tried at the state level in Oregon in 2019 would have led to a guilty verdict. However, the US Supreme Court recently ruled that jury unanimity in criminal trials is the constitutional right of the defendant, effectively putting an end to the non-unanimous jury law in Oregon, which had allowed 11-1 or 10-2 convictions since 1934 [1].
From a networks perspective, we can sum up juries as simply a group of voters who each vote with the aim to reach the “best” group decision, which is ultimately that the defendant should be convicted if they are guilty, but acquitted if they are innocent. We’ve learned about two Nash Equilibria for voting in a jury: (1) ignore your signal and vote to acquit and (2) always vote to convict when you have a signal that the defendant is guilty, and vote to convict with probability q between 0 and 1 when you believe the defendant is innocent. Focusing on the second equilibrium, as the first is an extreme, we find that even as the jury size grows to infinity, there is still a positive probability of an incorrect decision because individual jurors will over-correct too strongly for the possibility of being wrong. On the other hand, a non-unanimous f-majority model, like what was previously being used in Oregon, is shown to lead to less randomized correction and a probability of an incorrect verdict converging to 0 as the jury size grows to infinity. [Networks textbook]
The analysis above seems to point to two ideals for a jury: they should be large and they should be non-unanimous. Why, then, was there so much uproar from legal professionals over the non-unanimous system used in Oregon and why do federal courts all require unanimous juries? One possibility is that the theoretical model does not hold up well when real world complexities are introduced. Especially in a state like Oregon with a large majority white population, juries are likely to consist mostly, or even fully, of white people. This means when the defendant is a minority, they do not get a jury of their peers. This can have dangerous effects, as demonstrated by a case in 1993 in which Jerome Morgan, an African American teenager, was convicted of a crime by a 10-2 vote and served 20 years in jail before the ruling was overturned [3]. The two jurors who initially voted against conviction were the only Black individuals on the jury. For this reason, among others, unanimity is still regarded widely as a safeguard to prevent erroneous convictions.
While an f-majority jury model maybe not be adopted into the court system, there are still ways we can create juries with better decision making processes and, hopefully, have a greater probability of accurate outcomes. One important practice, as we touched on earlier, is creating racially diverse juries. A study on racially heterogeneous juries found that diverse juries were 10% less likely to assume a Black defendant as guilty than an all-white jury, even before deliberations [4]. Diverse juries don’t only benefit because of the contributions of minority members, but also because the very nature of having a diverse jury causes white jurors to be more aware and careful of their biases when evaluating evidence. In all kinds of decision making, and thus in a jury setting, groups with diverse backgrounds arrive at more vigorously considered decisions that take multiple perspectives into account. Courts can also create better juries by informing their jurors of how to spot logical fallacies during deliberations. As individuals, we all use heuristics in our thinking. Groups tend to lean into these heuristics even more and amplify errors. If jurors are prepared before hand to recognize logical errors and group decision making pitfalls, such as polarization or cascade effects, they can challenge other jurors when they arise or simply be more intentional in how they themselves are thinking.
Overall, there seems to be a disconnect between what we learn from this course about which jury model is better and what is actually practiced in the real world. We can use calculations to illustrate how a unanimous voting system may produce more errors than a majority voting system, but when real world influences interact with the model, the numbers may not always hold up or be convincing enough. While, at the end of the day, each juror makes their own decision on how to vote and behave, there are things that can be done by the court system beforehand to prepare the jury for a fair and logical deliberation process.
Main Article:
[1] https://www.opb.org/news/article/us-supreme-court-ruling-oregon-nonunanimous-jury-verdicts/
Other References:
[2] https://www.nytimes.com/2020/02/23/us/oregon-court-case-verdicts.html
[4] https://sparq.stanford.edu/solutions/diverse-juries-make-better-decisions