False identification of GKT with POT

[George W. Maschke]

Professor John J. Furedy of the University of Toronto has sent us the following constructive criticism regarding the 3rd edition of The Lie Behind the Lie Detector. While the problems he points out will be corrected in the 4th edition, with his permission, we're making his comments available here now:

I just managed to print off your excellent 3rd edition, but have come to a problem on p. 114 where, under the major heading of *Other polygraph formats*, your first sub head is *Peak of Tension (POT) or Guilty Knowledge Test*, and in a sentence spanning pages 114-5 you cite Lykken and give the impression that the POT is based on sounder principles than the CQT.

In fact the rationales underlying the CQT and POT are different, but they are both nonsensical for detecting even guilt, let alone deception. The POT may have been what led Lykken to propose the GKT in 1959, but that's really the only connection. See my discussion of the difference between the CQT and GKT in my 1996 article on the North American polygraph (see the table 1 especially).

Among the differences between the GKT (as it is now common used in the field for criminal investigations only in Japan) and the POT are the following:

1. The term "control" is has scientific sense only in the GKT and not the POT.

2. A statistically adequate GKT (that permits one to detect guilt reliably, say, 19 times out of 20 (the usual group inferential statistical level of p = .05 level of significance) requires about 5 topics repeated about 3 times, where each topic has one critical question (for the guilty suspect) and four or five control questions. The POT usually just has one topic with 5 items, so no statiscal-inference based conclusion is possible just on grounds of reliability alone.

3. The GKT requires that only the suspect and the police know details of the crime, so one necessary condition is that such details be kept from the public. The POT like the CQT, is applicable, like snake oil, under all conditions.

4. North American polygraphers (especially APA-approved practitioners) don't use the GKT (except for the numbers test, which is not employed for the detection of guilt), whereas they do use the POT.

These assertions on my part are not mere pedantry. Polygraphers have recently made use of this confusion between the GKT and POT, and have referred to Lykken as if he supported the POT which, of course, he has not....