No abstract

individual's verdict was recorded, both before discussion and after its complchion. Each juror then snswered a questionnaire which, amongst other things, asked about personal responses to the trial. By repeating the procedure numerous times we were able to establish not just the reaction of one jury to one trial, but the reactions of a considerable number of jurors sitting with their fellows to hear the same trial. This is the unique advantage, whatever its other artifices, of the experimental approach. The verdicts thus obtained were related to five personal characteristics (sex, age, occupational status, educational attainment and previous experience of the legal system)6 and the results of each civtegory were subjected to the chi-square test (xa)." This statistic determines the probability thak dissimilarities between the classes being compared has arisen by chance. If that probability is shown to be less than 5 per cent. (P < 0.05), we make the assumption that a statistically significant relationship has been demonstrated. Where the probability is as low as 1 per cent., we treat the results as highly significant.k is not necessary to say much about the two trials that we took as the material for our experiments. The first was a case of Theft, in which, in the view of members of the project, the evidence against the accused was strong but far from watertight. The second, a Rape case, was more elaborake than the Theft case, in that (i) there were two defendants, " Harrison " and " Bryce " (we retain these names in analysing the results); and (ii) each could be convicted either of rape or attempted rape. The evidence against each differed in nature and strength. There was a strong case against Harrison, at least on the attempt charge, since he not only admitted that he had been with the alleged victim but also in cross-examination came close to agreeing that he had forced her to submit to sexual interference. There was corroborative evidence that the victim had been assaulted. The case against Bryce, which was substantially independent of that against Harrison, depended on the victim's uncorroborated accusation, to which he responded wikh a denial that he had done more than speak a few words to her.Thus we have to consider the sets of verdicts given against three defendants-the Theft case defendant, and Harrison and Bryce in the Rape case.7 The results are complicated by a further factor, produced by the method we used to obtain our jurors. In order to restrict expenditure while first using our technique, we conducted our initial experiments-twenty-eight juries trying the Theft casewith jurors who responded to invitations posted in commercial and 6 These characteristics were recorded in a questionnaire given to the jurors at the beginning of the experiment. 7 Our analysis is here confined for the most part to the final verdicts given after discussion. The initial verdicts were sought for the purpose of studying shifts of opinion during the course of discussion, a matter on which we intend to report separately. ...