The Deception of Detection
March 1986
In the piney hills of northeastern Alabama sits an old cinderblock building about the size of a small-town body and fender shop. One exterior wall has recently been given a fresh coat of beige paint, which the other three could use. Rural Army installations all across America look like this, but Building 3165 at Fort McClellan houses something special: a school that teaches people to believe that they can tell when someone is lying.
The present class of 26 students—drawn from the FBI, Secret Service, National Security Agency (NSA), and military investigative branches—is part of a burgeoning program to supply the Reagan administration with enough polygraph operators to grill tens of thousands of federal employees with access to information the government wants to keep under wraps. Secretary of State George Shultz has said he would resign rather than take such an exam, but others, lower on the bureaucratic ladder, won’t find the choice so easy to make.
That polygraph tests are given so much weight in Washington—and are widely administered in American businesses—is one of the darksome curiosities of life in the United States. While lie detector exams are employed in few other countries (Israel and Japan are the main exceptions), government use in the U.S. has more than tripled since the early 1970s, from about 7,000 exams in 1973 to 23,000 in 1982. This is startling in light of the fact that the tests are little more than sophisticated carnival tricks. Although many polygraphists claim their procedure has an accuracy rate of 95 percent, evidence indicates that, at best, their results are comparable to standardized psychological exams like the Minnesota Multiphasic Personality Inventory, which claim to be 80 percent accurate. At worst, you’d be better off tossing a coin.
“It’s a nice idea that, like Pinocchio, we all might give off some physiological indicator when we’re lying,” says Leonard Saxe, a professor of psychology at Boston University and principal author of a 1983 study of polygraph validity for the congressional Office of Technology Assessment (OTA). “Unfortunately, we don’t really know.” The polygraph does not detect lies, he emphasizes; it merely records such telltale signs of anxiety as increased pulse, breathing rate, and perspiration. You might be lying, or thinking about family problems, or faking—or just about anything.
“On a more pragmatic level, the lie detector does work as long as the subject believes it works,” Saxe says. “A good examiner scares the crap out of you. It’s theater.”
This theater, however absurd from a scientific point of view, has been a social issue for much of this century, and continues to be a source of debate among law enforcement experts, congressmen, and a handful of scientists. Under the Reagan administration, which on November 1, 1985, authorized expanded use of polygraph tests by federal agencies—which could mean that an additional 10,000 senior officials will be subjected to the machine’s scrutiny—the debate has heated up. (Secretary Shultz’s objections did cause the administration to back off from the plan, at least for the moment.) It has also sparked discussion of commercial polygraphing, to which hundreds of thousands of workers are subjected every year, mostly in pre-employment screening.
Advocates of lie detector tests found themselves challenged when the OTA report concluded that the scientific validity of polygraphing could not be established from available research. They now argue that the tests have utility if not validity—the exams are useful despite being bogus, in other words—and they want more research.
“The people who have experience with polygraph services, the people who use them, the people who pay for them, find them extremely valuable,” says Frank Horvath, a professor at Michigan State’s school of criminal justice and a director of the American Polygraph Association, a trade organization. “The marketplace, not the scientific community, has decided the need.”
Horvath, who runs a research center at Michigan State financed by his association, published in 1977 a doctoral dissertation that included a re-analysis of polygraph charts. He found polygraphs to be 77 percent accurate in detecting the guilty and 51 percent in exonerating the innocent—or one percent more reliable than flipping a coin.
Field studies reviewed by the OTA showed a broad range of accuracies. Lie detection rambled between 50.6 and 98.6 percent. Truth detection swung even more wildly, from 12.5 to 94.1 percent. Inconclusive tests, in which the polygraph failed to point one way or the other, ranged from zero to 25 percent of total cases per study. “Although some differences can be explained methodologically, other differences cannot,” the OTA stated.
“The polygraph itself doesn’t do anything at all, except make recordings,” Horvath says. “When we talk about whether polygraph testing works or not, we mean whether the instrument can be used in a method of collecting data that are useful in inferring something about truthfulness or deception.” In other words, the little machine in the suitcase with lots of wires and knobs can help make edgy stupes spill the beans, so why knock it?
The staff at the Fort McClellan school takes polygraphing very seriously. They point to the role of polygraphists in investigating the 1969 My Lai massacre of civilians by American troops in Vietnam and the 1980 murder of three American nuns and a lay missionary in El Salvador by a Salvadoran military death squad. Agents from the governments of Israel, Venezuela, Taiwan, South Korea, and Canada have been trained there. For the first time since the school opened in 1951 (then at Fort Gordon, Georgia, from where it moved to Fort McClellan in 1975), the Defense Department recently let a reporter spend a day there attending classes and talking to instructors.
On a blackboard in the main classroom of Building 3156 was a list of terms labeled no-no’s, including Ouija board, hot question, squiggly lines, innocent, guilty, thingamabob, and lie detector. Every time a student utters one of these no-no’s, he must plunk some change into a basket in the center of the room. The kitty will go toward a graduation picnic.
“We’re at the mercy of everybody who wants to hammer away at the polygraph to make it sound cheap,” says Colonel Jon McFarland, assistant commandant of the Military Police School at Fort McClellan. In early January, 1986, McFarland found his school being hammered away at by the Washington Post, which reported that a polygraph training manual used at Fort McClellan suggested that operators pose such impertinent questions as “Have you ever received sexual stimulation in a crowded area?” and ask black subjects about the degree of their participation in the NAACP—as if it were a subversive organization.
The 14-week course at Fort McClellan consists of four weeks of lectures on law, semantics, ethics, physiology, pharmacology, psychology, operation and maintenance of polygraph machines, and testing procedures, plus ten weeks of practice training. A professor of psychology from nearby Jacksonville State recently gave a talk at the school on the “dynamics of normal behavior.” His lecture seemed to serve little purpose except to expose the students to a lot of jargon, as if he had been invited only to lend a scientific aura to the proceedings. When questioned about his reasons for teaching the class, he distanced himself from polygraphing and claimed a research interest in “psycho-physiological measurements.”
Fort McClellan is acknowledged for turning out proficient polygraphists. Although the course is no more or less valid than the technique it propounds, students at least come away knowing they are administering a kind of psychological test, not a truth serum.
The academic professional organizations that might be expected to cast a stern eye on polygraphing have been oddly silent about lie detectors until recently. In December, 1985, the directors of the American Psychological Association approved a statement of “great reservations about the use of polygraph tests to detect deception,” which warns of “great damage to the innocent persons who must inevitably be labeled as deceptors.”
Saxe says scholarly groups have taken so long to take a stand because none of the professions is threatened by polygraph tests: “It’s not as if the tests were replacing medical procedures, or that people were going into polygraphy and not psychotherapy.” Perhaps the whole subject is perceived as rather low-brow, too.
The paucity of good research on polygraph techniques is largely the result, Saxe thinks, of “the lack of an interesting underlying theory. We already know that anxiety has multiple causes, so there’s no hypothesis to test here.” This is a scientist’s diplomatic way of saying that something is just plain stupid.
What is interesting, at least from a pop culture standpoint, is the complexity that has grown up over the years around a machine whose roots lie in nineteenth-century pseudo science. Cesare Lombroso, the Italian psychiatrist and criminologist who in 1895 claimed he could detect a lie by measuring blood pressure changes, was also convinced he could recognize a criminal by the shape of his skull. The first polygraph was built in 1906, but it was used solely as a medical instrument to measure pulse rate and blood pressure. It was not until the next decade that William Moulton Marston, a Harvard-educated psychologist, popularized the device as a lie detector. In 1932, Marston offered to place his expertise at the disposal of Charles Lindbergh (who did not reply) after his baby had been kidnapped. Three years later, Marston caused a public stir by suggesting a lie detector test for the condemned kidnapper, Bruno Hauptmann, who continued to maintain his innocence (he was never tested).
It took the Chicago gangland murders of the same era to make the polygraph a fixture of American law enforcement. But Marston eventually became an embarrassment to serious practitioners by suggesting that polygraphs could tell if your wife was enjoying your kisses less than someone else’s. He even left the fold for awhile to create the comic book character Wonder Woman.
There is nothing special about the polygraph machine—it is just a portable cousin of common medical instruments. “There really hasn’t been a great deal of change” over the years in the basic set-up, according to Alfred Cross, product manager at the Stoelting Company of Chicago, a principal manufacturer. As a subject is questioned, three physiological responses related to arousal are indicated by pen needles sliding across a moving chart: depth of respiration, measured by pneumatic tubes strapped around the chest and abdomen; cardiovascular activity, by a blood pressure cuff on the bicep; and skin conductance (a gauge of how much the subject sweats), by electrodes on the fingertips. Modern machines, which have names like The Fact Finder and The Diplomat and cost upward of $5,000, are very sensitive. Being hooked up to one is disconcerting—perhaps because of the wiggling styluses silently tracing out body signals that cannot be felt.
One of Marston’s professional descendants, Cleve Backster, made a name for himself in the 1970s by attaching his polygraph to plants. He became convinced that the plants were responding to events in the lab—clipping seemed especially traumatic. Backster also made a serious contribution to polygraphy, a system of numerical scoring on charts that took some of the subjectivity out of examinations.
Polygraphists use three main interrogation techniques. The relevant/irrelevant (R/I) method assumes that somebody who is lying will have a greater physiological reaction to questions linked to a crime he has committed, say, or an undesirable character trait than to innocuous questions. It also assumes that a truth-teller, because he has no specific fears, will respond with equal calm to all questions. Refinements of this method are widely used in commercial employment screening and by the NSA for nonspecific “fishing expedition” sweeps.
During an R/I test, the examiner asks relevant questions (“Do you use drugs?”), interspersed among irrelevant ones (“Is today Monday?”) that provide a baseline. In the periodic exams given to some federal employees for security purposes, almost all the questions are relevant, and signs of arousal to any of them lead to further queries into the subject’s activities.
The control question technique (CQT) was developed to get around some of the R/I’s faults. It incorporates questions designed to be arousing for nondeceptive subjects and less arousing for deceptive subjects than the relevant questions. That is, it assumes that deceivers will be more nervous about those relevant questions, but it adds to the mixture questions that are supposed to provoke stronger responses among innocent subjects. For example: “Before the age of 20, did you ever take anything that wasn’t yours?” The subject is not told there is a distinction between relevant and control questions, which are carefully constructed after pre-test interviews so that they are almost impossible to answer truthfully with an unqualified no. The CQT is the most common method used in investigating crimes.
The OTA report pointed out that since the subject’s mood is so crucial in response to control questions, the validity of the CQT depends on how the subject feels about the examiner. “In this sense,” the report said, “CQT examinations . . . require examiners to make important judgments about and interventions in their interactions with subjects.” This is an amorphous skill, to say the least.
A training film shown at Fort McClellan demonstrates the use of CQT to exculpate a suspect in a theft case. In the movie, which has the cinematic feel of Reefer Madness, a suave examiner manipulates a G.I. into ratting on his larcenous girlfriend by pretending he has enough polygraphic evidence to nail the soldier. The gullible G.I. breaks down and incriminates his sweetie to save himself. Here the machine is really just a prop for intimidating the subject—which might be what it is all about, anyway.
The third technique, known as the concealed information test, is based on a different premise. Its goal is to find out whether a subject knows something about a crime that only a guilty party could: Was the stolen car red? Was it green? Was it yellow? Et cetera. One question will evoke the most nervous response, or so the theory goes.
Polygraph proponents insist that the machine can unmask security risks. “In the experience of the Central Intelligence Agency and the National Security Agency, the examination has repeatedly produced information relevant to an individual’s trustworthiness that failed to surface in background investigations or by other means,” wrote retired Army general Richard G. Stilwell (1917-1991), Deputy Undersecretary of Defense for Policy from 1981 to 1985, in reply to views expressed by Jeane Kirkpatrick (1926-2006)–the American ambassador to the United Nations from 1981 to 1985 and a hardliner on security issues—on the Washington Post’s op-ed page. She called the polygraph “a gross instrument that is probably least effective against the most effective professional spies.”
It is the polygraph’s reputed ability to elicit confessions that apparently caught President Reagan’s attention. At his press conference on January 7, 1986, he maintained that “the very nature of the test has led to a multitude of confessions of various crimes and so forth.” Since such claims are based entirely on anecdotal evidence, they pass blithely over the issues of polygraph validity and reliability.
They also ignore one point of agreement among those who study lie detector tests: they are least reliable doing the kind of broad security screening Reagan wants them for. The OTA found “meaningful scientific evidence of polygraph validity” only in investigations of specific criminal incidents, warning that even here the research was far from conclusive. “No field studies exist to directly test the situations for which the Defense Department and the President propose to expand polygraph use,” the report said.
Beyond the sphere of the federal government, with its training of polygraphists and regulations that limit how test information can be used, stretch the wilds of commercial testing. The Fort McClellan staff tends to look down its collective nose at these outsiders, noting that many private examiners get only six or seven weeks schooling. Moreover, government tests last several hours, while private ones often take less than twenty minutes. Commercial firms are also known to use devices that proponents of polygraphing consider bogus, such as voice stress analyzers. Among seven companies listed under “lie detection service” in the Washington D.C. yellow pages is one that advertises “lie detection by telephone.”
Frank Argenbright, a private polygraph examiner and chairman of the APA’s board of directors, estimates that there are about 4,000 commercial examiners in the U.S. (the trade group has some 3,000 members). No firm number exists for how many commercial tests are given each year, though an association official told a House committee in 1985 that “hundreds of thousands” of American companies use them. The exams are especially popular among retailers with large cash flow and high employee turnover: fast-food chains, discount stores, supermarkets. The APA reports that 60 percent of retailers, 50 percent of banks, and 20 percent of large businesses perform at least some polygraph screening. An examiner can gross between $50,000 and $60,000 a year, Argenbright says, at around $35 per test.
Statutes in about half the states limit polygraphing in the workplace. They generally keep employers from requiring exams, but many allow them to solicit or request the tests. Even in places that prohibit such exams, like the District of Columbia, employers can skirt the law by doing their hiring in a state, like Virginia, that permits testing.
When faced with bipartisan congressional support of legislation that would virtually outlaw polygraphing in the private sector, the APA sent a letter to congressmen in October, 1985, warning that “the handwriting is clearly on the wall for all who choose to see it.” It announced the formation of a pro-lie-detector PAC, named the Committee for Integrity in the Workplace in marvellous Orwellian fashion, and called on members to urge Congress to pass a regulatory bill rather than a ban. The legislation outlawing polygraphy is still pending, but it is expected to pass in the House this spring.
A separate bill that would ban federal usage has not fared as well on Capitol Hill, mostly because of the CIA’s nearly invincible constituency there. Representative Jack Brooks (D-Texas), chairman of the Government Operations Committee, for years has campaigned in vain against federal polygraphing. House aides concede that the legislation calling for a private sector ban is carefully written so as not to step on the toes of the national security establishment.
This means that Fort McClellan will probably be turning out its graduates at least for the remainder of the Reagan administration, adding to its 1,200 alumni of the past 35 years. The school aims to expand its output from 48 students a year to 108, and has high hopes for new facilities. According to one student from NSA, there are new, though unannounced, pay incentives to go into polygraphing. And the Secret Service has joined the CIA and NSA in requiring pre-employment exams.
These agencies claim that polygraph exams are only one element in an extensive background checking and screening process. John Schwartz, an instructor at the McClellan school, says that nobody in the Defense Department has ever been denied a security clearance solely on the outcome of a polygraph test. This may not be the case at the CIA, where “a considerable number” of job candidates are turned away every year because of their performance on polygraph tests, according to spokesman Patty Volz. In any case, a bit of simple arithmetic demonstrates why even the Schultzes and Kirkpatricks are alarmed about any reliance on the devices. According to the OTA study, a polygraph security screening operation might detect one “guilty” person out of 1,000 employees. Even assuming a very high rate of polygraph validity, say 95 percent, the one leaker or spy might be identified as deceitful, but so might 50 innocent employees. Even assuming 99 percent validity, there would be ten mistakes for every correct detection.
Perhaps more important, as Kirkpatrick pointed out, is that those the government would most want to catch may be the best trained to fool the polygraphist. “Anybody can sit in that chair and distort and move and make the test inconclusive,” says Ronald Decker, the chief of Fort McClellan’s polygraph division. “The whole procedure requires that the subject cooperate.”
He also notes that the NSA has taken one precaution against an examinee who tries to fool the polygraph by tightening his sphincter muscle, and thereby momentarily raising his blood pressure. Instead of sitting in the traditional straight-backed chair, the subject must recline on a Barcalounger. Apparently that makes it harder to pucker.
Postscript: The polygraph school at Fort McClellan moved to Fort Jackson, South Carolina, when the entire McClellan base closed in 1999. It is now known as the National Center for Credibility Assessment.