Stealth,


Quote:

...However, in my discussions with people that have undergone polygraphs, it is my unscientific opinion that polygraphs are generally accurate (more than 50%).

Note that a correct decision rate of greater than 50% is not necessarily indicative of greater than chance accuracy. For example, in the Department of Defense's counterintelligence-scope screening program, for the past two fiscal years, everyone who did not make "substantive admissions" passed (which is nearly everyone). Certainly, very few of those polygraphed are likely to have been spies, and it could be argued that the accuracy of this program is in the high 90th percentile. But similar "accuracy" could be achieved by the chance-accuracy method of arbitrarily deciding that everyone passes.


Quote:

While your book is thorough with regards to discussion of the control/relevant test.  It is sparse with information for this type of test (Irrelevant/Relevant, peak-of-tension, or whatever), particularly in discussing possible counter-measures.

You're right. Our discussion of the Relevant/Irrelevant technique is brief not just because it's less-commonly used, but also because it's not as well documented in the polygraph literature. For example, polygrapher James Allan Matte barely mentions the technique in his 800-page volume, Forensic Psychophysiology Using the Polygraph. Perhaps this is because the R/I technique is largely discredited, even within the polygraph community.

As for countermeasures to the R/I technique, I don't presently have anything to add to what we've included in The Lie Behind the Lie Detector, except to note, as I did above, that if the polygrapher is using the "clinical" method of scoring, then I agree that producing any reaction to any relevant question would likely be "shooting yourself in the foot."