Maths in the dock

By Rebecca Davis 7 October 2011

It is common for forensic experts giving evidence in court to use mathematical formulae to show the probability of someone having committed a crime. But mathematicians are up in arms about the fact that a UK judge has now forbidden the application of statistical analysis to crime trials. By REBECCA DAVIS.

Bayes’ theorem is a mathematical formula which measures how likely something is to be true. In its application to crime, it provides a scientific way to calculate the likelihood of guilt or innocence, or the reliability of evidence. Forensic experts use Bayes’ theorem as a way of quantifying their certainty. In other words, instead of saying “this DNA evidence is probably accurate”, they can say “there is a one-in-three-million chance that this DNA match is wrong”.

But Bayes’ theorem has now been ruled inadmissible by a UK judge. A convicted killer was in the dock last summer appealing against his conviction and among the evidence which had seen him found guilty was a shoeprint from a pair of Nike takkies, found at the crime scene, which seemed to match a pair of Nikes at his house. An expert used Bayes’ theorem to make calculations about the probability of the match being random, which was unlikely: there are 42 million pairs of takkies sold every year, and Nike alone has about 1,200 different sole patterns.

But in this case the judge objected to the fact that the numbers weren’t firm enough. The expert couldn’t say, for instance, exactly how many pairs of these particular Nikes were in the country. As a result, the judge threw out the case and also ruled against the use of statistical analyses in future. Mathematicians say that losing the right to use the theorem will have “shattering” consequences. They are calling on the court of appeal to reconsider the ruling. DM

Read more: