ABSTRACT
Here is another relevant question for assessing the two procedures: how do they compare with regard to their truth-tracking potential? Following the idea of Condorcel's jury theorem, this question is addressed by Rabinowicz and Pettit in an appendix to Pettit 2001, and it is studied further and in greater generality in Bovens and Rabinowicz 2004.4
Bovens and Rabinowicz investigate a model in which a group of an odd number of n voters faces a situation of the type introduced at the outset. The ex ante probability of P being true is q, and so is the ex ante probability of Q being true. Each voter is assumed to be right about P with the probability p, and he has the same chance of correctly assessing the truth value of Q. All probabilities involved are independent.
