This thesis considers a game in which imperfectly informed professional judges and lay judges in a mixed tribunal vote to select out of two alternatives: acquit or convict, when there are natural unequal competencies between the two class of judges. In Feddersen and Pesendorfer (1998), jurors are assumed to be homogeneous. This thesis extends their model with minimal enhancements to examine the optimality of different voting rules in a mixed tribunal under different behavioural assumptions of homogeneous professional judges and homogeneous lay judges. Voters in the mixed tribunal simultaneously deliberate together to decide verdict and punishment where applicable, however, they may be subject to different voting rules and votes are counted in relation to their respective voting rules.
This thesis shows that in a large size mixed tribunal and assuming strategic behaviours, requiring lay judges only to reach a majority vote and professional judges to agree unanimously actually lowers the probability of both type of errors than having a unanimity rule for the mixed tribunal as a whole and also that the relative incompetencies of lay judges becomes largely irrelevant in reducing both type of errors in a large sized mixed tribunal under the majority rule. In particular, the level of threshold of reasonable doubt of voters in the mixed tribunal has significant impact on the result of different voting rules.
Overall the contribution of this thesis to the literature is to show how different tribunal systems or behaviour of judges affect the probability of convicting the innocent and acquitting the guilty. Findings from this thesis lend support to the presence of lay judges in mixed court legal systems.