In fact it is possible to test the efficacy of polygraphs, but the trouble is such experiments are rare – at least according to Richard Wiseman, professor of the public understanding of psychology at the University of Hertfordshire. Because you can’t simply ask a criminal whether he was lying or not, the only method would be to use polygraphs in cases where the true answer had already been established using other evidence. Then it would be possible to check the polygraph’s responses against reality in a high-stakes situation. “If the examiner was unaware of the true answers, then that would be a proper double-blind scientific study. But there have been very few trials like that,” says Prof Wiseman.
From what evidence there is, he adds that “the polygraph seems to be very good at catching liars. But it seems very likely to give false positives as well, to 'catch’ people who are telling the truth”.
Horder agrees, pointing to research which found that polygraphs were “70-97 per cent accurate for people who turned out to be guilty, but between 12-94 per cent accurate for people who turned out to be innocent
so between 12 and 94 % accurate when the suspect is innocent
