While the physiological targets of polygraph testing have not changed much since the 1930s, numerous testing techniques, question formats, scoring systems, and specialised applications have emerged since then.
The plethora of approaches and the associated lack of standardisation have made it difficult to provide clear estimates of polygraph accuracy. A number of initiatives have meant the situation has improved (Kraphol & Shaw, 2015). Chart scoring, as opposed to decisions based on a global overview of the polygraph chart, was introduced in the 1960s, a hardening of testing protocols took place between the 1960s and 1990s, increased acceptance of blind scoring of charts as a means of Quality Control to overcome the risk of examiner bias became more commonplace in the 1990s, research in the early 2000s better clarified response patterns that are indicative of deception (and just as importantly, response patterns that aren’t) and the amount of variance explained by the different physiological channels, and in the late 2000s the American Polygraph Association undertook an exercise to validate testing techniques (American Polygraph Association, 2011). All of this has provided a better scientific basis on which to evaluate the efficacy of polygraph testing.
To this end Norman Ansley found in 1983 – When research is conducted using real polygraph cases in which independent means are employed to check truth or deception, the average validity is 96 percent, with a range of 86.3 to 100 percent (Twelve criminal studies, involving 1,964 polygraph examinations were considered) These statistical results do not include those examinations in which the results were reported as inconclusive.
When research is conducted in a laboratory setting when truth and deception is known (except to the examiner) the validity of polygraph techniques averaged 93.6 percent, with a range of 69 to 100 percent. (Twenty-Eight laboratory studies, involving 1,113 polygraph examinations were considered). But not all of these laboratory studies cited were conducted to determine validity, some were projects to evaluate variations in techniques, methods of analysis, specific and often single physiological recordings, and specific types of subject population. For example, the third study by Heckel was of delusional psychotics which showed low validity, 69 percent, while the studies of psychopaths resulted in a surprise, with an average detection rate in excess of 90 percent.
In 1990 he Norman Ansley followed up with a report on validity from all studies of real cases, conducted since 1980. Examiners decisions in these studies were compared to other results such as confessions, evidence, and judicial disposition. The ten studies reviewed considered the outcome of 2,042 cases, and the results, assuming that every disagreement was a polygraph error, indicated validity of 98 percent, and for non-deceptive cases 97 percent. These studies were from police and private cases using a variety of polygraph techniques, conducted in the United States, Canada, Israel, Japan and Poland.
That polygraph techniques are cross cultured is evident from the similarities of the results of studies made in Poland, Israel, Iceland, Japan, Canada and the United States.
But the most definitive review of polygraphy accuracy to date has been carried out by the National Academies of Science in the United States. It concluded that “polygraph tests can discriminate lying from truth telling at rates well above chance, though well below perfection” (National Research Council, 2003, p. 4). Accuracy for the most commonly used test format, the comparison question test (a version of which is employed in PCSOT), was estimated to be between 81 to 91 percent, which is highly supportive of a meaningful association between what the polygraph records, truth telling and deception. The National Academies review was carried out on behalf of the US Department of Energy.
Also see the Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques, a report prepared for the American Polygraph Association Board of Directors – Nate Gordon, President (2010-2011).