Probabilistic Analysis of Identification Evidence

A brief review of psychological studies of probability judgements is said to support a prima facie case that formal methods of analysis could correct biases present in intuitive judgments in legal cases. A simple probability model for using matching identification evidence (like blood types) to find posterior probabilities of identity between a suspect and a responsible party in a legal case is illustrated by the use of Steinhaus’ data from 1515 paternity cases in Poland. Various extensions of and issues surrounding this basic model are then introduced: a potential third party in a ‘whodunit’, probability of guilt as well as identity, classifications of cases, personal probability estimates of priors on identity, frameups, observation and source error in the evidence, selection of the suspect by the evidence, double-counting of the evidence.