This section relies on aspects of the joint evaluation of scientific findings as outlined in the previous sections. A sequential procedure is followed in order to illustrate the rationale behind the proposed Bayesian networks. First, network structures are elicited for reasoning separately about each trace item. Then, in a second step, ways are examined in order to combine network fragments logically.
Consider the following scenario, adapted from Stockton and Day (2001, p. 3) with ‘credit card’ in the original changed to ‘chequebook and accompanying bankcard’:
Two people, Mr. Adams and Miss Brown, have been arrested in a shop trying to pass a stolen cheque. Mr. Adams admits to writing the signature on the cheque, but claims he has only just acquired the chequebook and certainly has not written out any more cheques from the book. Miss Brown claims to know nothing of the whole affair. On investigation it is found that sixteen further transactions have taken place using the chequebook and an accompanying bankcard, which was reported stolen a few days ago. The owner, Mr. Constantine, claims the card must have been stolen in the post and his signature does not look anything like the signature on the card. Handwriting samples, including the signatures in the name of Mr. Constantine, are obtained from both Mr. Adams and Miss Brown.
Different analyses have been proposed by Stockton and ...