Issue 186 of SOCIALIST REVIEW Published May 1995 Copyright © Socialist Review

This month Britain opened the world's first ever national DNA database in Birmingham. Home secretary Michael Howard claimed that the database was 'the most significant scientific advance in fighting crime since fingerprints'. At the same time, new police powers came into force as part of the Criminal Justice Act. As well as ending the right to silence and reintroducing stop and search laws, the police now have the right to take DNA samples from suspects. This can include taking swabs from the mouth, blood samples or more intimate intrusions. The DNA profiles can then be kept on the national database and compared against samples obtained from the scenes of crimes.

Recently, following a murder case in Cardiff, all young males on a particular housing estate were asked to come forward to offer blood samples. They were told that they would be treated suspiciously if they refused. The civil liberties implications of the new powers are enormous. As well as police having the right to stick instruments into any part of your body, they also have the right to keep your DNA profile on computer. Samples can be taken from anyone arrested on suspicion of any crime, from shoplifting to murder. According to the police, the files will be deleted if you are acquitted, although last year Howard was reported as saying that deleting profiles may be 'technically difficult'.

Clearly socialists should oppose these moves on this basis alone. But what about the science? Much has been made of the ability of DNA techniques to identify an individual by their unique profile. Some advocates have stressed that this can benefit defendants by eliminating them from an inquiry. There are two problems with this. The first is that DNA profiling is very expensive and beyond the reach of most ordinary defendants, so it is almost always the prosecution that calls for it.

Secondly, the scientific method itself is fundamentally flawed and challenged by many respectable scientists in the field.

DNA molecules are the building blocks of genetic material. They determine to a large extent what each individual will look like. It is therefore true in principle that each person has a unique DNA profile. However, it is only practical to use a bit of DNA at a time for sampling. A sample taken from your body will only give a glimpse of your total DNA sequence--it will only show part of your profile, not the whole picture. This will then be matched against only a part of someone else's DNA sequence taken from the scene of a crime--opening up the possibility of finding a match even though you are perfectly innocent.

To account for this chance similarity the prosecutors (or the profiling companies working for them) calculate what the probability would be of an individual selected at random matching the sample taken from the crime scene. The calculation of this so called 'match probability' is the subject of much dispute amongst scientists, and this lack of agreement in itself has been enough to throw DNA evidence out of court in several cases in America.

There is a further sleight of hand used to impress juries into securing a conviction, known as the 'prosecutor's fallacy'. Basically the prosecution confuses the match probability, that is the probability that the similarity of DNA is just a random match--which can be very small, say one in a million--with the odds that the suspect is innocent despite the apparent match. The second calculation will depend on the size of the pool of possible suspects from which the defendant was selected. In an article in Nature last year two prominent scientists showed that even with a match probability of one in a million, the chance of innocence could be one in three or even greater.

In the first DNA based conviction to come to appeal in Britain, Andrew Deen was given a retrial when the 'prosecutor's fallacy' was exposed. During the appeal it was also heard that discrepancies in two out of the ten bands of DNA used to match Deen's sample were not taken into account sufficiently by the judge. A professor of forensic medicine from Germany said at the appeal that in his opinion the DNA evidence may even be sufficient to show Deen's innocence, not his guilt! Clearly there is a lot of ambiguity involved in deciding what is a good match in the first place. On the basis of DNA evidence at the original trial, Deen was sentenced to 16 years in prison for rape.

Another problem with the calculation of the match probability was first outlined in 1991 by Richard Lewontin and Daniel Hartl, a population biologist at Harvard University. The probability of achieving a match at random will be calculated by comparing the frequency of certain alleles (specific bits of DNA used for matching) in a representative sample of the population. But what is a representative sample? Here the question of ethnicity becomes very important. The FBI uses three reference populations--Caucasians, Blacks and Hispanics. But there are many ethnic sub-groups, and they may have more genetic variation than the three main groups. This means that the chances of, say, an Amish suspect having a certain combination of alleles could be higher than for other sections of the population, increasing their chances of being randomly matched. In Britain this could mean a bias against people from Bangladesh or Somalia, for instance.

Following Lewontin and Hartl's arguments, the FBI collected information on the genetic makeup of ethnic subgroups, but in an article in Nature last year Hartl claimed they had not changed their practice despite the new evidence. He said that the FBI 'argued that the differences between ethnic groups didn't matter. Their data show that they do, and it's very inconvenient.'

Similar arguments apply to the case of close relatives. In a recent Scottish case the forensic scientist gave a match probability of one in 49,000. He then accepted that the chance of a match from a particular brother was about one in 16. As it happened, the defendant had five brothers.

Anyone with any illusions in the neutrality of the science involved should look to the example of the United States. When the National Academy of Sciences released its report on DNA fingerprinting in 1992, its conclusions were far from enthusiastic. In particular, on the question of 'one in a million' claims that the likes of Howard want to impress us with, the report stated that 'there is not a sufficient body of empirical data on which to base a claim that such frequency calculations are reliable or valid' (my emphasis).

Not satisfied with this, the FBI pressurised the NAS into a second report. They urged them to look at an alternative statistical method, the so called 'ceiling principle'. This method gives more conservative estimates of the match probability and it was hoped that it would be more politically acceptable. But all it consists of is a series of guesstimates. It would not be very comforting if you were up on a murder charge to think that your likelihood of conviction depended on a scientist's best guess--especially when that scientist was in the pay of the prosecution. Newton Morton of the Cancer Research Campaign attacked the illogicality of this method, saying that if the committee simply wanted a more conservative estimate it could have urged expert witnesses to 'move the decimal point a couple of places'.

In pushing for a fresh report, the FBI deliberately focused on alternative statistical methods, and chose to ignore other parts of the old report, such as the requirement for forensic laboratories to publish error rates after independent inspection.

The point for socialists is not to argue about which is the best statistical method, but to demonstrate that the whole idea of DNA fingerprinting is scientifically unsound and an attack on our rights. The profiling companies in America have a direct interest in securing as many positive matches and convictions as possible. Cellmark, the company that will handle the DNA analysis in the O J Simpson trial, has been accused of sloppy analysis following false matches made during a proficiency study in the late 1980s. Cellmark's errors result from misinterpretation of data, ambiguous data and mixing of samples, according to William Thompson, associate professor at the University of California at Irvine.

The same process will be at work in Britain. Big money will be thrown into the race to secure convictions. The cost of maintaining the database for the first year alone is estimated at around £28 million. It will be in the interests of the police and the scientific companies conducting the analysis to find positive matches. The result will be more miscarriages of justice, more abuse of police powers, and more attacks on black people and the working class. The DNA database should be closed down and the Criminal Justice Act opposed anywhere and everywhere that it is used.