Miscarriages of justice are likely to happen because courts are dumbing down complex statistics, one of Britain’s leading forensic scientists has warned.
Peter Gill, a pioneer of DNA profiling, has alerted the Home Office to a series of trials in which judges allowed expert witnesses to give a subjective summary of complex DNA evidence in place of a robust statistical analysis.
The departure from accepted scientific standards risks misleading the jury over the strength of evidence at a time when more cases are hingeing on traces of DNA found at crime scenes, he said.
“I’m convinced that miscarriages of justice will occur, because of the unscientific methods being used to convict people,” said Professor Gill, who is based at the University of Oslo. “There is a risk of cases being constructed around weak matches, which could be seen as fitting people up.”
Advertisement
Sue Pope, the former head of DNA interpretation at the Forensic Science Service, has independently raised concerns with the Home Office about the precedent set by three recent cases in which convictions were secured partly based on subjective interpretations of DNA evidence.
In one trial, in which two teenage boys were convicted of murdering a security guard at a London brothel, the judge rejected a request for a statistical assessment of DNA found on a knife to be put before the jury. David Balding, a professor of statistical genetics at University College London, was told that software that he had developed to analyse complex mixtures of DNA did not have “sufficient scientific basis” in the hearing in 2012, even though the software had been used in previous and subsequent hearings.
Instead, the judge allowed another forensic scientist to give a qualitative view on whether it was likely that one of the defendants had contributed to the sample.
A pattern of “hostility to scientific evidence” among judges was especially worrying given that such evidence was playing an increasingly decisive role in courts, Professor Balding said.
Courts have only recently had to grapple with complex DNA evidence, because technical advances have made it possible to obtain readouts from tiny quantities and mixtures of genetic material. In the US, mathematical software is routinely used in trials but English courts have been more reluctant to admit the latest methods.
Advertisement
“Judges want the scientist to stand up and say ‘I’ve been doing this for 20 years and it’s him’,” Professor Balding said. “I think that’s horrifying.”
Norman Baker, the crime prevention minister, said that an investigation into what may safely be put to juries had been commissioned by the Forensic Science Regulator, and that the results would be published in due course.
• Case study: Kuba Dlugosz, 36, was convicted in 2011 of the manslaughter of Eveline Kelmenson in a case that hinged on a match with a mixed DNA sample. In a landmark appeal court ruling last year, Dlugosz and defendants in two unrelated cases — a murder and a child sexual assault — unsuccessfully challenged their convictions based on the subjective interpretation of DNA evidence.
In the Dlugosz case, mixed DNA samples, with DNA from at least two people, were taken from two chisels found at the home of the 83-year-old victim.
In DNA profiling, 20 components are read-out from a suspect’s DNA and compared against the sample. For a simple DNA sample, a score of 20 out of 20 would give a probability of close to one in a billion that you had the wrong person. However, with mixed DNA samples the likelihood of a chance match is far higher.
Advertisement
The Polish burglar was identified as a suspect after he appeared as one of the top hits when the mixed DNA sample was run through the DNA database, which contains genetic data from five million people.
The court was told that it would be “rare” to see all twenty components match up, as the defendant’s did, if he had not contributed to the sample. However, new calculations by Professor Gill show there is a 5 per cent chance that you would find such a match in the database by coincidence.
For a second DNA sample used as evidence in the trial, the chance of a random person from the database matching was estimated at 95 per cent – almost guaranteed rather than rare.
The jury was not told that the “match” was discovered by a database trawl, meaning that this information was effectively hidden from the court proceedings, according to Professor Gill. “The jury cannot place the DNA evidence into context, without this crucial information,” he said.