Selection effects in forensic science

Abstract

In this report we consider the following question: does a forensic expert need to know exactly how the evidential material was selected? We set up a few simple models of situations in which the way evidence is selected may influence its value in court. Although reality is far from a probabilistic model, and one should be very careful when applying theoretical results to real life situations, we believe that the results in our models indicate how the selection of evidence affects its value. We conclude that selection effects in forensic science can be quite important, and that from a statistical point of view, improvements can be made to court room practice.