Leniency Bias

Evidence for the Leniency Bias

The initial and strongest evidence for this bias comes from a number of jury simulation studies. Robert MacCoun and Norbert Kerr meta-analytically combined the results of 13 such studies and reported that (a) on average, acquittal was about four times as likely as conviction for mock juries that began deliberation evenly split (e.g., 6 G vs. 6 NG); (b) an initial twothirds majority favoring acquittal was more likely to ultimately prevail (94% of the time, on average) than a two-thirds majority favoring conviction (67% of the time); and (c) the stronger the evidence against a defendant, the weaker was this bias. On the other hand, a handful of surveys of ex-jurors from actual criminal jury trials (e.g., by Dennis Devine and his colleagues) have suggested either no such asymmetry or even the reverse effect (i.e., a harshness bias), but at present it remains unclear whether or not actual criminal juries do exhibit a leniency bias. This is because there are a number of potentially important methodological ambiguities clouding the comparisons of the mock versus actual juries. For example, the surveys of actual jurors all appear to have treated jurors who say that they are undecided at the first jury vote as advocates for acquittal, which is likely to overestimate the true size of the pro-acquittal faction in the jury. In summary, there is good evidence of a leniency bias in mock juries where [Page 454]estimates of proand anticonviction faction sizes are based on direct assessment of mock jurors' predeliberation verdict preferences in relatively close cases. There is currently no strong evidence of such a bias (and some evidence to the contrary) where these estimates are based on ex-jurors' retrospective recollections of the number of proconviction votes at their actual jury's first ballot, in convenience samples of diverse cases.

Explaining the Leniency Bias

One explanation for the leniency bias is the existence of a prodefendant norm among jurors. Research on group decision making and polarization suggests that one effect of group deliberation is to increase commitment to shared norms. The more consistent evidence for a leniency bias among mock jurors, who are usually college students, than among actual jurors could be interpreted as reflecting different norms in the student and nonstudent populations. A direct comparison of the leniency bias for a student versus a nonstudent sample has shown that nonstudents exhibit a somewhat weaker leniency effect, but the difference was not statistically significant.

Another, related explanation is based on the common law's aversion to false conviction. Such values are reflected in several elements of the law, including the presumption of innocence, the prosecution's burden of proof, and particularly the reasonabledoubt standard of proof. The law prescribes that juries must give a criminal defendant the benefit of any reasonable doubt. This should give advocates of acquittal an advantage over advocates of conviction during jury deliberation. For example, jurors favoring acquittal need only raise a single reasonable doubt in the minds of proconviction jurors, whereas jurors arguing for conviction must refute all reasonable doubts in the minds of pro-acquittal jurors. This explanation predicts that there should be no leniency bias when jurors apply a standard of proof that is not slanted to favor the defendant, such as the “preponderance of evidence” standard used in civil trials; this prediction has been confirmed experimentally. A model presented by Norbert Kerr, Robert MacCoun, and Geoffrey Kramer generalizes the asymmetry effect, demonstrating how any shared local norm can create disproportionate influence for one side of an issue.

...