A few years ago in southern England a drunken teenager threw a brick at a truck. The brick hit the driver’s window; the driver had a heart attack and died. The police forensic services obtained a DNA sample from the surface of the brick, but could not find a match. The case went cold…
Then last year the Forensic Science Service told the police that the newly developed process of “familial searching” might help. This is based on the reasoning that close relatives of the criminal will have partially similar DNA, and there may be a close relative somewhere in the police DNA records, especially as criminality tends to run in families.
On this basis the FSS searched their databases (containing about 2.35 million DNA profiles in England), and found one profile with an 80 percent match to the sample from the brick. The police listed close relatives of the partial match, identified a likely suspect, and obtained a DNA sample from him which proved to be a perfect match. After initially denying all knowledge of the crime, when confronted with the DNA evidence the suspect pleaded guilty to manslaughter. He has been sentenced to six years in jail.
So-called “civil liberties” organisations are said to be worried at the implications. Personally, I set a higher value on the civil liberty of not having drunken louts throwing bricks at me.
However, the case does raise some delicate questions of methodology. The use of DNA matching depends on a calculation of probabilities. In this case it was said (according to the report in the London Times, April 20) that the probability of a perfect match from someone who was not the true villain was one in a billion. But if the ‘sampling frame’ is constructed by identifying close relatives of people with a good partial match, then the probability of getting a perfect match by chance (i.e. from an innocent person) among those relatives must be greatly increased as compared with the probabilities in the general population. What is needed is the probability that someone identified by this method is the true villain.
I presume that the experts were satisfied that the probability of a perfect match by chance was still sufficiently remote, but as the suspect ‘copped a plea’ the matter was never tested in Court. It all sounds like an episode of CSI – I wonder if the same issue has come up in any American cases?
Posted by David B at 03:28 AM