This week at the AAAS : What chance we've got the wrong man?
点击量： 时间：2017-04-02 01:01:08
By Peter Aldhous IF EVER you find yourself on a jury asked to consider a case that hinges on DNA evidence, pay close attention to the way the statistics are presented. Whether the numbers are expressed as a percentage or as a frequency could make all the difference to your verdict. The fact that a DNA sample from a suspect matches DNA found at the scene of the crime does not necessarily prove that the sample came from the suspect. The key to interpreting such evidence is the probability that the samples will match even if the DNA found at the scene came from someone else. In the past, lawyers, forensic scientists and statisticians have argued bitterly over how these numbers should be calculated. But Jonathan Koehle, a psychologist at the University of Texas at Austin, says getting the right number is just the start. Making sure the jury understands what the probability means is an even bigger problem. “I think there’s no more pressing issue in DNA forensics than how jurors think about and use the statistics that are presented in court,” he told the AAAS. Koehle has run experiments with mock juries made up of students from his university. In one experiment, carried out just days before the AAAS meeting, Koehle asked his jurors to consider a case in which a sample of semen found on a dress belonging to a former White House intern was found to match a DNA sample from a US President. “This is hypothetical, obviously,” he says. The students were divided into two groups. One group was told that there was a 0.1 per cent probability that the President’s DNA would produce a match even if he were not responsible for the stain. The other was told that 1 in 1000 men in Washington DC would produce a match. To a statistician, these statements mean exactly the same thing. But Koehle has shown that most people interpret them very differently. “They’re identical, but they elicit different ways of thinking,” he says. In the first group of jurors, 28 per cent thought the President was virtually certain to have left the stain, compared to only 8 per cent of jurors in the second group. So which way of presenting the statistics is likely to serve the cause of justice best? Koehle thinks that frequencies are better than percentages, pointing out that 74 per cent of the mock jurors in the second group responded correctly when asked how many people in a city of 500 000 would have DNA samples matching the President’s. Only 26 per cent of the group given the number as a percentage got the answer right. “I think people have a very hard time distinguishing between numbers like 0.1 per cent and 0.01 per cent,