The Validity of Tests: Caveant Omnes

01 Jan 1987

A great debate swirls about the use of polygraph tests in criminal cases. Similar concerns about individual privacy and freedom arise with proposals and projects involving widespread testing of government employees for drugs and deception. Required diagnostic testing for certain diseases - most notoriously, for AIDS - raises similar concerns. Incorrect conclusions about who has taken illicit drugs, who has AIDS, and who is lying can be devastating. Yet, perfect knowledge is unattainable. Errors are inevitable. Questions of what the tendency is for these tests to err, which measures are appropriate for deciding whether to use a screening test, and what measures of error have to do with the admissibility of evidence in court surround this discussion. Dr. Lykken, a well-known critic of the polygraph test, argued that polygraphs could be invalid when applied to a group in which the base rate for lying is low, and he also made a similar observation about drug tests and other screening tests. This article elaborates on Dr. Lykken's description of the terminology commonly applied to medical screening tests, and shows how the biostatistical and psychometric quantities relate to issues of public policy and forensic proof.