The Magic of Science

Lifer Jeremy Bamber writes in your December issue of polygraph testing that the innocent have nothing to fear. Doubtless this sentiment owes something to Jeremy’s understandable euphoria following his own apparent vindication by polygraphy. I congratulate Jeremy on this, but a word of caution seems indicated. Jeremy is right to point up the significance of polygraph testing beyond the accuracy of the tests themselves. He is right to suggest that this significance has something to do with the fears of the guilty. But what is behind this is a cultural belief in the magic of science and technology equivalent to the so-called superstitions of primitives who believe that Witch Doctors can identify guilt in suspects who cannot swallow a piece of magic dry bread.

Many studies in the USA have shown that when the accuracy of polygraphy is hyped, it can indeed act as a truth facilitator. Those who believe in it start to confess rather than be found out. Interestingly in these conditions, the proportion of offenders claiming to have been victims of sexual abuse can fall dramatically (from 65% to 32%) and many more offences and risky behaviours are admitted (Hindman and Peters, 2001).

It can’t be said too often that lie detection is a misnomer. What polygraphy can measure is not truth or falsehood but some of the physiological and neurological correlates of stress that can sometimes be detected in certain people under some circumstances. Given that these correlates of stress result from belief in the technology, it is not surprising that the Home Office, the media and private practitioners with a financial interest such as Polygraph Security Services Ltd wish to talk-up its accuracy. Its inaccuracy is rarely their problem.

It is in relation to this inaccuracy that I return to Jeremy Bamber’s assertion that the innocent have nothing to fear from polygraph testing. When Grubin (2003) looked at this issue, he assumed a 90% accuracy of the technology and that 90% of 1,000 subjects were telling the truth. Given those assumptions, there would be 90 false positives people identified as deceptive who were actually truthful. The same theoretical model would identify 10 false negatives people incorrectly classified as truthful. Of the 180 identified as deceptive, 50% (90) would actually be telling the truth. They would be the innocent whom Jeremy believes have nothing to fear even from highly accurate testing. It could be worse than that. Sticking with the (optimistic) 90% truthfulness assumption and plugging in a more realistic 70% accuracy for the polygraph (Jaccono and Lykken 1997), 340 of the same 1,000 subjects are now reported as deceptive. Of these, 270 (79%), would actually be telling the truth. The claim that the innocent have nothing to fear is beginning to sound even more hollow.

False positives are not a problem for administrators or the public, which might be why they are played down. In the above model, there are only 30 false negatives those incorrectly classed as truthful. No wonder the Home Office is so keen to get the system running and it won’t stop with sex offenders.

One final point. For those who are actually innocent of some or all of their convictions (up to 30% of us according to the Inside Time survey) it is unrealistic to expect us to start believing the rhetoric of justice over our own bitter experience. If we took the huge risk of submitting ourselves to polygraphy, a risk our accusers are not generally willing to take, and if we were fortunate enough not to fall into the trap of false positives, the same people who enthuse about the polygraph are likely to dismiss the result as an artefact and/or to further condemn us as skilful liars. Evidence rarely impacts much on prejudice.