Let’s Not Rely on Lie Detectors--They Lie

Secretary of State George P. Shultz is raising proper questions about the President’s directive ordering polygraph testing of persons in and out of government who have access to classified information. But the most disturbing thing about the directive is the President’s reliance on what his advisers say about the polygraph, rather than on the scientific community’s virtually unanimous finding that--as Shultz pointed out--such tests are not reliable and are easily evaded by guilty parties.

When the scientific community outside government criticizes President Reagan’s Strategic Defense Initiative as impossible to achieve, its defenders inside government resort to the old trick of alluding to secret data not available to academic scientists. But the defects of the polygraph are all in the public domain; the President and his advisers just refuse to listen. Perhaps they think that this “20th-Century witchcraft” (Sam Ervin’s phrase) will at least work as a deterrent.

Operatives of the Central Intelligence Agency are tested periodically, but that did not deter Edward Wilson when he was planning arms deals with Libya. Nor did the prospect deter CIA employee Sharon Scranage, who conducted an affair with a Ghanaian intelligence officer and gave him secrets. Holiday Inns runs Harrah’s casinos in Nevada, where employees at the gaming tables are screened on the polygraph, and in New Jersey, where state law forbids polygraph screening. Their losses to employee theft are no different in Las Vegas than in Atlantic City, suggesting that the deterrent effect of the polygraph is as weak among Nevadans as it is in the CIA.

Defense Secretary Caspar M. Weinberger takes his advice from his own security people--people like Norman Ansley, head of polygraph testing for the National Security Administration. Ansley told Congress last year that the polygraph test is 98% accurate. His claim was based on a study in which questionnaires were sent to polygraphers in Virginia, asking how many of the tests that they gave in the last year had turned out to be correct. With studies like that, one could show that astrology, tossing the bones and other forms of divination also are 98% correct.

Another of Ansley’s sales pitches consists of a portfolio of success stories--cases in which some spy was unmasked by a polygraph test. One wishes that decision-makers in the top levels of our government could see through such hype on their own. Suppose that the next 10,000 tests given at the Pentagon were scored just by flipping a coin--heads, you’re lying; tails, you’re truthful. Since respondents must either be lying or not, about half those tests will be correct. Suppose that 100 of those “correct” tests are subsequently verified--there you have Ansley’s portfolio.

It is not necessary to assume that Ansley has a test miss locked up in his safe for every test success that he parades around Washington. Polygraphers seldom find out for sure, after scoring a test, whether their conclusion was right or wrong; that is why they can continue to believe in their own mythology. Moreover, they decide themselves which tests are “verified.” In 1983 Congress asked its Office of Technology Assessment to review the scientific status of the polygraph test. The study concluded that the polygraph detects stress or fear, not lies, and that it goes wrong as often as it does because truthful people often are upset by the ordeal of polygraphy, while some of the best liars are not. The OTA pointed out that the kind of test used in screening is basically different from the “control question” test used in criminal investigation, and that there is no scientific evidence whatever about the accuracy of screening tests.

Congress apparently has chosen to go with scientific opinion, because bills that would ban polygraph screening of employees and job applicants in the private sector have been introduced in both the House and the Senate with wide bipartisan support. Seventeen states already have such legislation.

A federal district judge in Georgia recently held that requiring public employees to submit to polygraph tests (and to sign the waiver of rights that invariably precedes such testing) is in violation of the Fifth Amendment. Polygraph evidence cannot be introduced at trial in most jurisdictions. In the last few years the Supreme Courts of Wisconsin, Illinois, North Carolina and Colorado have decided that the polygraph is too inaccurate to be considered in evidence, even if both sides agree to it.

Like the scientists, the courts know that the polygraph should not be relied on. Tens of thousands of ordinary Americans, victims of invalid polygraph tests, learn the same lesson each year at great cost. And the KGB even knows how to teach people to beat the lie detector.

It is gratifying to learn that Shultz knows the truth, too, and has the guts to put his job on the line. Let us hope that the President listens to him.