On 16 May 2024 from 1-2.15pm the Statistics and the Law Section held a meeting with Professor Sir David Spiegelhalter on “Communicating Statistical Evidence in Public Inquiries and in Court”. David discussed his involvement with a number of high profile inquiries including the infected blood inquiry, an inquiry on children’s heart surgery at the Bristol Royal Infirmary, and on the murders by Dr Harold Shipman. He showed how combining insight with data was key in much of the analysis, for example in noticing that the times that Harold Shipman’s patients tended to die differed from the norm.
David also discussed with us the use of likelihood ratios in investigation, inquiries and the court room. One example of this was how they helped in the identification of the remains of Richard III. He pointed out that Alan Turing had first used log-likelihood ratios to compute the weight of evidence at Bletchley Park long before they were being used by forensic statisticians for assessing evidence.
The event finished with time for questions and discussion. One interesting point discussed was how statisticians can convince jurors and the judge in court rooms of their expertise. Part of this boils down to effective communication. It may not be necessary to understand all of the technical detail of the calculation, but the statistician must be able to convince the decision-makers of the results of the analysis. We also discussed the need for young people interested in moving into this field to gain experience by building up from the traditional statistical techniques and clean datasets into working with messy datasets and unclear questions.
David also discussed the difference between noticing deviations from patterns and assessing causality. While it was possible to flag that Harold Shipman had an unusually high death rate, it was not possible to show from the data analysis that he was criminally responsible for the deaths.