Accuracy of the Polygraph

The term accuracy takes two conditions into account – Validity and Reliability.

Validity refers to whether a test does indeed measure what it claims to.

Reliability refers to the extent to which a test (or instrument) is consistent in its measures. It also addresses the issue of consistent results if another examiner retests a person.

Scientific evidence supports the high validity of polygraph examinations. However, a valid examination requires a combination of a properly trained examiner, state of the art polygraph instrumentation, and the proper administration of an accepted testing and scoring system.

Norman Ansley reported that from all studies of real cases conducted from 1980 to 1990, which involved 10 studies considering 2 042 police and private examinations using a variety of techniques administered in the US, Canada, Israel, Japan and Poland, validity was 98%.

Four studies of police and private examinations, where independent evaluators rescored polygraph charts concluded that reliability was 95%.

In 1983, Ansley showed average validity for 1 964 people in criminal cases was 96.3% and the average validity for 1 113 people in laboratory cases was 93.6%.

The American Polygraph Association has a compendium of research studies available on the validity and reliability of polygraph testing. The 80 research projects listed, published since 1980, involved 6 380 polygraph examinations or sets of charts from examinations. Researchers conducted 12 studies of the validity of field examinations, following 2 174 field examinations, providing an average accuracy of 98%.

Researchers conducted 11 studies involving the reliability of independent analyses of 1,609 sets of charts from field examinations confirmed by independent evidence, providing an average accuracy of 92%.

Researchers conducted 41 studies involving the accuracy of 1,787 laboratory simulations of polygraph examinations, producing an average accuracy of 80%. Researchers conducted 16 studies involving the reliability of independent analyses of 810 sets of charts from laboratory simulations producing an average accuracy of 81%. Tables list the authors and years of the research projects, which are identified fully in the References Cited. Surveys and novel methods of testing are also mentioned.