Bayesian Networks for Probabilistic Inference and Decision Analysis in Forensic Science, 2nd Edition

6.3 Scenarios with more than one trace: ‘Two stain-one offender’ cases

The Bayesian networks discussed so far in this chapter concentrate on the evaluation of single traces or stains given crime level propositions. This reserved comment for further concepts, such as relevance and innocent acquisition (Section 6.1). In addition, attention has been drawn to different levels of detail, or components, that may be considered when describing forensic results (Section 6.2). The discussion of such cases, involving a single offender, can be extended to scenarios that involve more than one stain. Imagine, for example, the scenario described by Stoney (1994), where two bloodstains, with profiles and of , are recovered at the scene of a crime. It is not known which, if either, of the two stains are relevant. A suspect is found who is of profile . The result of interest thus is, here, a correspondence between the suspect's characteristics and those of one of the two stains.

A formal approach to this scenario is based on the definition of the following probabilities:

If is true (i.e. the suspect ...