ABSTRACT
The forensic scientist is concerned with the value of his evidence. The meaning of probability is discussed in Chapter 1. There are sound mathematical reasons why the function of the two probabilities, the probability of the evidence given guilt and the probability of the evidence given innocence, that best measures the value of evidence depends only on the ratio of these probabilities. These reasons, for those who wish to study them, are given in the Appendix of Chapter 3. The function of these probabilities which is most sensible to use is the logarithm of the ratio, since this ensures that various independent pieces of evidence contribute in an additive way. (Note that this is quite different from the logarithm of the odds referred to in section 1.8). It is this expression, the logarithm of the ratio of the two probabilities, that is given the name ‘weight of evidence’ by Professor Good. It should be emphasized that weights of evidence are not probabilities. Thus they are not restricted to lie between 0 and 1; they may be negative and they may have magnitudes considerably in excess of 1. Confusion has arisen in the past because probability itself, that is the probability of guilt given the evidence (together with background information), has been used as a measure of weight of evidence. The word ‘weight’ suggests that two independent pieces of evidence can be added when both pieces of evidence are taken into account. However, this would be impossible if, for example, each piece of evidence, was conclusive by itself: the sum would be 2 and hence not a probability.
