When: 20 January, 2021, kl. 13-14 Where: This seminar is given online. E-mail Dan Hedlin if you want to attend.
Abstract
The broad introduction of DNA evidence in the 1980s was early considered to be the firm path towards a new gold standard when it comes to proving contact between a suspect and a victim (or something else related to a crime). Even if the estimated random match probabilities were much higher than today, it was generally anticipated that the probability was low that another person than the suspect could have left the DNA under investigation. Such an interpretation is actually erroneous, since evidence that a person has left their DNA is not solely determined by the rarity of it.
Unfortunately, among the actors of law enforcement, an obtained DNA match between person and a trace is still imagined to be almost sufficient to have that person convicted for having taken part in the criminal activity under investigation. But it is no longer fruitful to confront a suspect with statements like: ‘“Listen, we’ve got your DNA, so why not confess?”. In the childhood of DNA evidence, criminals simply had to surrender to such evidence, but nowadays they have learnt a lot about its weaknesses.
The development of using DNA as evidence has been very comprehensive since its introduction, and many of the end-users have not paid enough notice to this development. Today, it is possible to extract genetic information from very tiny amounts of DNA (picograms), but just because it is possible to find small amounts of DNA for instance on the victim’s clothes that matches a suspect, this does not constitute sufficient evidence that the suspect was the perpetrator. Any kind of innocent contact could have left such traces. It is today rare to question who the source of the recovered DNA is, but instead focus is on why and how, and even when it was deposited.
In this talk I will give a short crash course on probabilistic evaluation of forensic evidence and then focus on the challenges met today for DNA evidence, how probabilistic modelling can be used to discover which the crucial parts are.