The recent Vioxx decision spurred a lot of discussion in the blawgosphere (including a couple of posts by yours truly; e.g., here) about the ability of juries to handle scientific evidence. An empirical study of that very question was just published in the Journal of Legal Studies: Dale A. Nance and Scott B. Morris, Juror Understanding of DNA Evidence: An Empirical Assessment of Presentation Formats for Trace Evidence with a Relatively Small Random-Match Probability. As I read it, it's not good news for my position.
Here's the abstract:
In cases involving scientific evidence linking the accused to a crime (a "match"), expert testimony sometimes can provide a suitably reliable estimate of the chance of a coincidental match. Controversy attends the question whether, and in what form, to allow testimony reporting that probability. Further controversy concerns the implications of laboratory proficiency tests for the presentation of testimony about the chance of lab error. This large-scale empirical study, using members of an Illinois jury pool, confirms earlier research suggesting that, contrary to some predictions, jurors tend to undervalue forensic match evidence. Our results differ from most prior research, however, in showing that variation in the way the random-match probability is presented and explained can reduce the undervaluation, that it can do so without inducing significant inferential fallacies, and that incorporating information about comparatively large lab error rates, when it has any discernible effect, increases jurors' willingness to convict.
Here's the money quote from the conclusion:
There is now significant evidence that, when measured by Bayesian norms, juror assessments can be improved by providing appropriate instruction that offsets the otherwise extant, but reasonableindeed laudable tendency of jurors to discount the probative value of a scientific technique the results or implications of which they do not fully understand.
...our results tend to support a view of trial law as more urgently concerned with assisting the jury to understand the evidence, in this case a DNA match, than with shielding the jury from evidence, such as a calculated random-match probability, that some worry might lead them astray by triggering irrational mistakes or submissive credulity favoring the prosecution.
I have to admit that this study appears to support Beldar's analysis of the Vioxx case rather than mine. It looks like ability to present scientific evidence (at least of this sort) may well be a much bigger factor than the jury pool's intrinsic ability to assess such evidence.
