The problems with Efficacy of Repeated Psychophysiological Detection of Deception Testing, by Andrew B. Dollins, Victor L. Cestaro, and Donald J. Pettit (J. Forensic Sci. Vol. 43, pp. 1018-1023) begin with the title. Tests like the lie detector can be evaluated for their reliability (e.g., do repeated tests give the same diagnoses?) and also for their validity (is the diagnosis correct?) "Efficacy" ambiguously suggests that this was a validity study when in fact neither validity nor reliability was assessed.
Each of 22 subjects attempted to conceal from the examiner the fact that the number written on a card in their pocket was 64. Each test involved asking if their number was 60-66 while recording breathing rate, skin activity, and heart rate responses to each question. These tests were repeated 6 times on each of two days. This is the "peak of tension" text (POT) format, virtually never used by professional polygraphers.
The POT is like one multiple-choice item in a Guilty Knowledge Test (GKT), the one method of polygraph testing proven to have high validity. The GKT attempts to identify not lying, but whether the suspect recognizes crime scene facts or photographs that a guilty suspect would recognize. Because polygraphers are committed to the myth of lie detection, the GKT is almost never used in practice.
As expected, the average subject gave larger autonomic responses to the number 64 than to the other numbers. But we are not told what proportion of the subjects gave their largest response to 64 on any of the 12 tests (the true-positive validity).
Another group of 22 subjects did not carry the 64 card in their pockets but we are not told how many of them happened to give their largest responses to 64 (the false-positive invalidity). And we are not told how many of either group received the same diagnosis on both days of testing (the test's reliability). Thus, this study is not concerned with any of the methods of lie detection in wide use in the U.S. nor does it provide any useful information about either the reliability or the validity of the seldom-used POT. If this is an example of the quality of 'research" being conducted, then polygraphic lie detection will continue to be based more on myth than on science.
