ABSTRACT
Courts have historically avoided informing jurors about their nullification power (i.e. the power to return a not-guilty verdict when their conscience demands it but the law directs otherwise), fearing that such knowledge would prompt disregard for the law and reliance on attitudes and emotions rather than evidence. We investigated jurors’ inclination to nullify the law in a morally ambiguous case of physician-assisted suicide, testing the impact of euthanasia attitudes on case judgments as well as moderators and mediators of that effect. Mock jurors with pro-euthanasia attitudes were overall less likely to vote guilty than anti-euthanasia jurors, especially when they were given jury instructions informing them of jurors’ power to nullify. Nullification instructions also exacerbated the effect of jurors’ attitudes on anger, disgust, and moral outrage toward the defendant – emotions that mediated the effect of attitudes on verdicts. We also tested the impact of incidentally induced anger on jurors’ reliance on their attitudes rather than the law, given anger’s propensity to increase certainty and heuristic processing. Anger enhanced mock jurors’ reliance on their attitudes under certain conditions. Theoretical and practical implications for understanding juror decision-making are discussed.
Acknowledgement
We thank Amani Elayan and Taylor Borgmann for invaluable research assistance; Alison Perona for legal consultation; and Jessica Salerno, Cynthia Najdowski, Dan Cervone, and Linda Skitka for feedback and advice.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. Some participants in the standard condition (27%, or 42%) reported that they had heard instructions to rely on their conscience. These participants were not excluded, because their answer cannot be attributed to lack of attention – they did not miss any information. There is every reason to believe that actual jurors would think the same thing, given that the actual jury instructions were used, therefore excluding them would make the study less ecologically valid. Even so, we conducted the main regression analyses a second time, using the manipulation check variable instead of the objective manipulation of the instructions, thus allowing participants’ subjective answers to the manipulation check question replace the objective manipulation. That is, the 27 jurors in the standard condition who reported they had received nullification instructions were coded as being in the nullification condition. Both analyses yielded similar results, therefore the results section presents only the analyses using the actual manipulation.
2. Another potential methodological concern could be that the similarity of response scales across all proposed mediators prompted generic response bias – participants’ simply giving the same ratings regardless of the variables measured. This was not the case in our study: We found different patterns of results for positive and negative emotional reactions to the defendant. Further, we performed the same analyses, post hoc only to address this issue, on two additional items that measured perceptions of the defendant with similar scales, but did not assess emotional reactions (‘Dr. Wood is a sympathetic physician who cares about his patients’; ‘Dr. Wood is a proud man who likes to play God with his patients’ lives’). Perceiving the defendant as a sympathetic doctor was negatively related to guilty verdicts (B = −0.52, Wald = 5.13, OR = .60, p = .024) and degree of guilt (B = −2.06, 95% CI [−3.45, −0.64], t(128) = −2.88, p = .005). Perceiving the defendant as a proud man was positively related to guilty verdicts (B = 0.32, Wald = 5.79, OR = 1.38, p = .016) and degree of guilt (B = −1.32, 95% CI [0.36, 2.28], t(128) = 2.73, p = .007). Yet unlike similarly measured emotional reactions to the defendant, these perceptions of the defendant did not mediate the effect of attitudes on verdicts. Jurors’ euthanasia attitudes did not significantly predict perceptions of the defendant as sympathetic, B = −0.15, 95% CI [−0.7, 0.18], t(124) = −0.90, p = .37, or proud, B = 0.13, 95% CI [−0.09, 0.34], t(124) = 1.18, p = .24. The other main and interaction effects involving incidental anger and nullification instructions were also not significant. Thus, although the measurement scales were similar, participants’ responses to these items measuring different constructs were not. Had participants merely responded to all questions in the same way because of measurement bias, we would have observed the same pattern for these separate items.