3.4 Personal probabilities

We mentioned in the previous section that probability was personal because individuals have different knowledge and beliefs. Some differences in beliefs might be more or less reasonable since they can be based upon more or less reliable information.

Probability assignments based upon more reliable and generally agreed information will be more objective. For example, the scientific evidence underlying the processing and analysis of full DNA profiles for single donors is well established and so experts are generally confident and in agreement about assigning probabilities in this circumstance.

Probability assignments which are based upon less reliable or less generally agreed information will be more subjective. For example, the transfer and persistence of DNA on surfaces is an active area of research for scientists and so probabilities involving these circumstances may be subjective. Subjective probabilities are not necessarily bad because they can still be the best assessment of the available knowledge. This might happen for example when there is very limited published scientific literature about a particular scenario. Even though there is limited published information, an expert may have relevant previous case experience in that scenario and so may be able to assign subjective probabilities. In other words, subjective probability assignments are not necessarily (and should not be) arbitrary.

For repeatable events like the coin toss we can check and update our probabilities using empirical data, e.g. repeated tosses of the coin. This can be seen as gaining a better understanding of the aleatory uncertainty about the distribution of the outcomes and eliminating epistemic uncertainty about whether the coin tossing process was being done fairly. With enough repetition, personal probabilities which were initially different between individuals can converge to the same value as more objective information becomes known. In criminal cases, the opportunity to repeat events and gather scientific data from the same case circumstances can be limited. This is because some events are one-off and difficult (if not impossible) to completely replicate in scientific experiments. In these situations, probabilities can still be informed by other available empirical data however, e.g. by observations from similar circumstances, and can also still be informed by expert knowledge. This knowledge should be disclosed and be available for audit by the court.