From Mike Keating
Angela Saini’s article about misunderstanding probability in courts makes me wonder how it’s possible to verify assertions such as the probability of having DNA matching that found at the crime scene being “1 in 200 million” (24 October, p 42). I doubt that 2 billion cases have been examined for this purpose. With DNA, I gather that the samples are compared in 20 or so places on the genome. A figure is then given for the chance of a match at any one place, and raised to the power of 20 to give the chance of 20 matches. This assumes that matches in different places are independent of one another.
As we derive our DNA from a small pool of ancestors, not all possibilities will be realised in the population, so 20 matches will occur more frequently than the independence hypothesis suggests.
Cornhill-on-Tweed, Northumberland, UK
