THEY call it the CSI effect. The seemingly endless succession of glossy TV shows about forensic science has created a popular perception that the dazzling array of techniques now available can produce fast and firm evidence for convictions.
It is well known that traditional methods such as bite-mark and fingerprint analysis rely to some extent on analysts鈥� opinions, and as such they are viewed with a healthy measure of scepticism. However, DNA evidence has so far escaped this treatment.
Profiling is generally seen as routine, infallible and always able to get its man. One recent study found that jurors rated DNA evidence as 95 per cent accurate, while mock trials showed that they were more likely to convict when presented with DNA evidence than other types of evidence.
Advertisement
鈥淛urors see DNA evidence as highly accurate and are more likely to convict when it is presented鈥�
But, as we demonstrate in this issue, DNA profiling is far from perfect, and in many cases can run into the same problems of subjectivity and bias seen with the more old-fashioned techniques.
This is an evolving problem, one rarely encountered in the early days, when analysts only tended to tackle large samples of DNA from one or two people. Forensic scientists are now increasingly swabbing crime scenes for DNA evidence as a matter of routine, which means samples are often very small or contain DNA from more than one person. These factors make analysis far more challenging.
Even though doubts were raised about DNA evidence as long ago as the early 1990s, albeit for quite different reasons, it is still seen as rock solid. There is, of course, plenty of very convincing DNA evidence, but little distinction is made, either in labs or in court, between samples of different quality: DNA is just DNA.
There鈥檚 a deeper problem, too: a pervasive idea that the role of forensic science is to convict criminals, rather than to evaluate the available evidence and present it in the right context.
One way to begin to tackle these issues would be to use statistical methods to ensure that the correct weight is placed on each piece of evidence and allow juries to better assess the likelihood of a suspect鈥檚 innocence or guilt. In next week鈥檚 issue, we will examine the options on offer and show that it is possible to use them to remove some of the subjectivity around DNA analysis and highlight those worrying cases where the evidence is feeble.
However, few labs currently use the recommended statistical analysis, despite the fact that there are several methods available. Why is this? Many scientists we spoke to complained about a lack of funding. Peter Gill, former principal research scientist for the UK鈥檚 Forensic Science Service, points out that even though improved statistical methods were devised over a decade ago, the lack of investment means that they have not yet been translated into practical techniques. If and when they are, the resulting software packages should be open-source, so that all labs can do the same analysis. For there to be real justice, labs must have equal access to all the latest technology.
If forensic science is ever to live up to the hype of the CSI effect, statistics need to become as cool and important as the techniques deployed at the crime scene.