3
Lie Detection

A fluttering Heart, and unequal Pulse, a sudden Palpitation shall evidently confess he is the Man, in spite of a bold Countenance or a false Tongue.
Daniel Defoe,
An Effectual Scheme for the Immediate Preventing of Street Robberies
and Suppressing all other Disorders of the Night

Our exploration of contemporary authenticity testing begins with lie detection. It may appear to be a curious beginning, for lie detection occupies the less-reputable neighborhoods in the city of testing. German courts have held that lie detection's aim to bypass the conscious mind (which may wish to conceal certain information) so as to dredge facts directly from the unconscious compromises the right to avoid self-incrimination and erodes the basic human quality of free will.^[1] No less august a personage than Pope Pius XII was moved to condemn lie detection for its capacity to "intrude into man's interior domain."^[2] Subjects I have interviewed often reported feeling demeaned by the suspicion of wrongdoing implicit in the request or insistence that they submit to a lie detector test and violated by the prospect of machine-enhanced snooping into their private affairs. Attitudes such as these have given lie detection an unsavory enough reputation in official circles that a federal law greatly curtailing its use in the private sector took effect at the end of 1988.

These foibles notwithstanding, lie detection is of great interest for our analysis because it brings certain common characteristics of testing into peculiarly interesting and instructive focus. Most important, because the whole purpose of lie detection is to reveal information that subjects wish to conceal, the power that test givers typically exercise over test takers is particularly prominent here. After analyzing certain exaggerated—occasionally even caricatured—features of testing as they appear in lie detection, it will be easier for us to identify the subtler versions of those same features that lurk in more respectable forms of testing.

The Ghost in the Machine

In tracing the context of assumptions within which lie detection operates, it is helpful to begin over three hundred years ago with French philosopher René Descartes. Standing at the source of modern philosophy, Descartes distinguished sharply between the body and the mind. His theory, now known as Cartesian dualism, holds that while activities of the body take place openly in the physical world, where they can be observed by anyone, the thoughts and emotions that constitute mental activities occur in some metaphorical "place" deeply buried within a person. The doings of the mental "ghost in the machine" (as Ryle termed it)^[3] are hidden from public view. Observation of them is the exclusive prerogative of the self, who (until Freud and others muddied the waters with the concept of an unconscious mind) has immediate and total access to them by introspection.

People have long cherished the prospect of intruding on the private precincts of other minds, and no society has lacked shamans or psychics who lay claim to the pertinent preternatural powers. In our own society and century, the strongest bid for direct communication with ghosts in other machines comes from a persuasion that claims to have solved the problem by the judicious application of modern science and technology in a family of procedures known generally as lie detection or (a somewhat more genteel designation often used by proponents) the detection of deception. In essence, lie detection operates on the premise

that a mind-body connection exists such that it is possible to know something about what a person is thinking on the basis of certain measurable physiological responses. Lying—or, more precisely, anxiety about being discovered in a lie—is thought to produce bodily perturbations. These may be no more than slight fluctuations in blood pressure or subtle changes in the voice, however, so liars are often able to conceal their prevarication-provoked physiological responses from the ordinary scrutiny of human observers.

One scheme for unmasking callous deceivers was advanced by Daniel Defoe. Although he proposed it in 1731, the tactic of scrutinizing subtle bodily signs for evidence of a guilty ghost lurking within is totally in accordance with the assumptions that underwrite contemporary lie detection:

Guilt carries Fear always about with it; there is a Tremor in the Blood of a Thief, that, if attended to, would effectually discover him; and if charged as suspicious Fellow, on that Suspicion only I would always feel his Pulse, and I would recommend it to Practice. . . . It is true some are so hardened in Crime that they will boldly hold their Faces to it, carry it off with an Air of Contempt, and outface even a Pursuer; but take hold of his Wrist and feel his Pulse, and there you will find his Guilt; a fluttering Heart, an unequal Pulse, a sudden Palpitation shall evidently confess he is the Man, in spite of a bold Countenance or a false Tongue.^[4]

Today, thanks to the ultrasensitive instruments of modern technology, the imperceptible quickenings of the most accomplished, cold-blooded liar are now susceptible to precise measurement, recording, and scientific analysis. This points, incidentally, to another way in which the analysis of lie detection reveals a general characteristic of testing in a special light. Testing has a tendency to fragment or decenter the subject, to characterize the person in terms of one or a very few dimensions rather than as a fully rounded, integrated human being. This appears in the case of lie detection in a peculiar, extreme, and (from the suspect's point of view) threatening form. Not only is the subject divided but the parts are actually set against each other, so that the mental

ghost is betrayed when the blabbermouth bodily machine spills the beans. But bad news for the subject may be good news for the community, especially for that branch of it devoted to law enforcement. As one interrogator happily described the distinctive achievement of lie detector tests, "the criminal can no longer hide in the deepest recesses of his mind."^[5]

Many of the medieval tests described in chapter 2 sought to plumb the deepest recesses of criminals' minds by putting questions to God, who would answer through the outcome of ordeal or combat. As part of an overall secularization of society, one of the most salient transformations in assumptions about tests in general as we pass from these antecedents to their modern forms is the renouncing of all reliance on God's omniscience in favor of purely human knowledge, as enhanced by the methods and mechanisms of science. This is especially visible in lie detection, where issues that would previously have been put to trial by ordeal or battle are addressed from a putatively scientific perspective and with heavy reliance on technological devices. Lie detector tests deal with human qualities in terms of things. The polygraph itself is a thing, a machine. Its job is to measure the activities of other concrete things: the heart, the lungs, and the skin. The ultimate purpose is to detect truth and deception, but these too are understood as correlates of physical things and events rather than purely as subjective qualities of mind or morality. The aura of science is further fortified by a tendency to classify subjects according to their physical responses in polygraph tests. To examiners, people become objectified as "spot responders," "reactors," "nonreactors," "respiratory reactors," or "diminishing reactors."^[6]

As we have seen, modern testing has its roots in positivism. Lie detection's redefinition of human qualities in terms of things is an outstanding example of the agenda charted for positivism by Saint-Simon. He argued that human misery, strife, and war stem from decisions and actions in the realm of human affairs that are taken on the basis of opinions. If they could instead be based on things, as scientifically determined and understood, then human affairs would be regulated according to positive facts rather than ephemeral, uncertain opinion.

It is no longer men controlling men. It is truth alone which speaks; it is impersonal, and nothing is less capricious. In short it is things themselves—through the mediation of those who understand them—that indicate the manner in which they should be handled.^[7]

Advocates of lie detection have anticipated similar happy results from the extended application of polygraph technology to, say, the legal system. Unlike the capricious, opinion-laden decisions of judges and jurors, the scientific lie detector test is not subject to bias.^[8] Robert Ferguson, Jr., provides example after example where polygraph tests have rectified the errors of juries, proving the innocence and securing the release of falsely convicted defendants.^[9] In one case, a judge had his doubts about a jury's conviction of a man tried for robbery and assault with a deadly weapon. He delayed sentencing and ordered a polygraph test. The results indicated that the convicted man was in fact innocent. Refusing to allow a man to be convicted under these circumstances, the judge vacated the jury's verdict.^[10]

The results of polygraph tests are grounded in concrete things as known systematically by scientific methods of measurement and are therefore objective and certain. The polygraph speaks truth as a matter of principle. Therefore, if the results of a particular polygraph test are manifestly mistaken, the problem lies not with the nature of the test but with how it was conducted or, perhaps, with the person who was being tested. On the analogy that one does not lose confidence in fingerprinting because particular prints are occasionally smudged, "there is no reason, therefore, why any derogatory implication should be attached to the lie-detector technique because of the occasional unfitness of the person upon whom a deception diagnosis has been attempted."^[11]

The situation is fascinating: a testing technique that has been touted as being so thoroughly saturated with science as to be nearly infallible has been so widely perceived as fundamentally flawed that a law has been passed against it. Arguments on both sides of the issue will be reviewed here and in chapter 4, as we examine more closely what lie detector tests are, how they work, and what social consequences they have.

Lie Detector Machines

Modern lie detection boils down to a machine-enhanced capacity to detect minute bodily perturbations, together with the professional skill to interpret their mental correlations. The machine most commonly used in lie detection is the polygraph. Quickly reviewing some of the highlights of its development, the polygraph was foreshadowed in the mid-nineteenth century when Italian criminologist Cesare Lombroso utilized measures of pulse and blood pressure in the interrogation of suspects. By 1908, the well-known psychologist, Hugo Munsterberg, was advocating the use of blood pressure for detecting deception in the law courts. William M. Marston, one of Munsterberg's Harvard students, claimed to have discovered a specific physiological response that accompanied lying. An avid publicity hound, he made several unsuccessful efforts to apply his technique in the investigation of the kidnapping of the Lindbergh baby. Ultimately, Marston went on to other pursuits and, under the pseudonym of Charles Moulton, created the comic strip character "Wonder Woman."^[12]

Marston's claim that a specific physiological response exists for deception did not survive, but exploration into more general links continued. John A. Larson developed a large and unwieldy machine capable of providing continuous measures of blood pressure, pulse, and respiration rates. Working closely with the criminologist and police chief, August Vollmer, he used the instrument for investigations in the Berkeley Police Department. His most famous success was a shoplifting case in the early 1920s. The culprit was known only to be one of thirty-eight women living in a certain dormitory. Larson interrogated all of them with the assistance of his machine and found that one reacted much more strongly to questions about the thefts than the other thirty-seven. Confronted with the findings, she made a full confession.^[13]

A high school student named Leonarde Keeler, fascinated by police work, used to spend his spare time around the Berkeley Police Department in those days. He took a keen interest in Larson's and Vollmer's experiments, and in 1926, Keeler developed the first compact, portable polygraph machine. It remains the

prototype for current models. Keeler went on to become one of the founders of the lie detection profession. Based in Chicago, he utilized his polygraph extensively in criminal investigations during the 1930s and 1940s. Keeler developed some of the formats for lie detector tests that are still in use and founded the first polygraph examiner training institute as well as a company that manufactures polygraph machines.^[14]

The polygraph is a device that simultaneously monitors several different physiological processes. The machine is attached to the subject in several ways. An expandable band is placed around the chest (and perhaps a second one around the upper abdomen) to measure breathing patterns. A blood pressure cuff is applied to the upper arm and inflated to about halfway between the diastolic and systolic pressure, to measure fluctuations in blood pressure and pulse rate. A small clip placed on a fingertip measures galvanic skin response (GSR): an electric current, so small that the subject cannot feel it, is passed through the clip, and changes in conductivity reveal variations in the subject's perspiration. Another, similar clip may also be applied to measure pulse and blood volume in the fingertip. Readings from these sensors are recorded by fluctuating pens on a long sheet of graph paper that moves steadily beneath them. During the course of the test, the examiner writes the number of each question on the graph paper as it is asked. This makes it possible, when the chart is analyzed, to connect physiological responses as recorded on the chart with the various questions.

No one (at least, no one since Marston) claims that deception can be measured directly. As with all tests, lie detection deals in representations. In the case of polygraph tests, these representations take the form of a chain of causality leading from prevarication to anxiety or psychological stress and then to physiology. As already noted, the general theory is that emotional stress produces variations in one or more of the physiological functions monitored by the polygraph. One form of stress that may produce such physiological variations is apprehension about being caught in telling a lie. Assuming that other reasons for stress can be ruled out, significant changes in the polygraph tracings that accompany the response to a given question (when a subject

"hits" on a question, in the parlance of the trade) may be interpreted as signifying deception.

The type of signification that operates in polygraph tests is metonymy. The presence of an observable effect metonymically signifies the co-presence of its cause, even if the cause is not observed directly. Precisely in the same way as smoke signifies fire, in the logic of the polygraph, certain physiological perturbations, properly analyzed, signify deception.

Stanley Abrams sets out to provide a more detailed account of the rationale behind polygraph examinations.^[15] In an apparent effort to drench his discussion in as much science as possible, he even includes information on the human cell and the operation of neurons, although, alas, no relation between these particular matters and the operation of the polygraph is identified. In the more pertinent part of his account, Abrams keys polygraph to the "fight or flight" reaction that has been indelibly imprinted in human beings and other animals over eons of evolution as an adaptive response to situations of danger. The threat of being discovered in a lie provokes this reaction, which consists of a set of physiological changes that prepare the individual to fight or to run away. "Vasoconstriction takes place in the peripheral blood vessels and causes an increased flow of blood to the skin, thereby allowing for a dissipation of the heat engendered by muscular effort and fostering a reduction of blood loss should injury occur."^[16] Moreover, sweat appears on the palms and hands "to aid in locomotion or grasping. . . . Stronger contractions of the heart send more oxygenated blood through the body and an additional blood supply is directed to the skeletal muscles allowing for a more effective utilization of the arms and legs."^[17] These developments are monitored by the various measuring devices of the polygraph. Therefore, lie detection by the polygraph, as Abrams cogently points out, produces an interesting twist on evolutionary patterns of self-preservation: "ironically, those same responses that typically serve to get the individual out of trouble get him into difficulty if he is deceptive during a polygraph test situation."^[18]

Although this polygraph machine is the one in general use today, technological elaborations have been tried from time to time. The Darrow Behavior Research Photopolygraph (fig. 3), mar-

[Full Size]

Figure 3.
Darrow Behavior Research Photopolygraph. From Paul V. Trovillo.
"A History of Lie Detection." Journal of the American Institute of Criminal
Law and Criminology 29 (1939):874.

keted half a century ago, seems to have been attached to the subject's body in every way imaginable. Apparently, the apparatus used by the CIA, which makes frequent use of instrumental lie detection, is similar. An informant said that it expands the number of physiological characteristics monitored from three or four to twenty-eight and that an individual wired up to it "looks like an astronaut." At the other extreme, Israel's Weizmann Institute has developed a machine that does not touch the subject's body at all. It measures palpitations of the stomach remotely, by microwaves, on the theory that such movements increase in frequency when an individual is lying. The device is in use at border checkpoints, where individuals may be subjected to lie detection without their being aware of it.^[19]

Lie detection by remote sensing, of course, obviates the necessity to hook the subject up to clips, tubes, and wires. These attachments frequently provoke a good deal of consternation.

Several polygraph examiners told me that it is frequently necessary to assure subjects that the machine will not give them electric shocks; occasionally, people observe with apprehensive humor that they feel like they are going to the electric chair.

The psychological stress evaluator (PSE) is a type of lie detector machine that attaches nothing to the subject's body. It is designed to analyze stress in the voice. As its history was explained to me by a PSE examiner, the device was invented by two army officers, Charles McQuiston and Allan Bell. They wanted to give lie detector tests to suspected double agents and others in the jungles of Vietnam and found that the polygraph was not suitable for such use in the field. With the PSE, one asks the same sort of questions as would be used in a polygraph test, but the subject answers (yes or no) into a tape recorder microphone. The tape is then run, at one-quarter speed, through a special analyzer that produces a chart composed of closely packed lines of varying heights and positions on a narrow strip of paper similar to that used in cash registers. The expert studies the chart for points of vocal stress or tension, not audible to the ear, that may be found in the answers to certain questions on the test.

The general theory behind the PSE is identical to that informing the polygraph: anxiety about being caught out in a lie produces certain physiological responses (this time in the voice), and these can be detected and recorded by the machine. In an interview, a PSE examiner provided a more detailed rationale that (in contrast to Abrams's reliance on the fight or flight reaction) trades on the relation between the subconscious and conscious sectors of mind. It is an outstanding example of the tendency of lie detection to disarticulate the person into fragments that are then set against each other. The subconscious mind, he explained, is like a computer data bank with no discriminating controls. In response to any question, the subconscious simply dumps everything it has on that topic. The conscious mind, which is a highly discriminating gate-keeping device, reviews what the subconscious has produced and decides what to reveal. The verbal answer to any question is what the conscious mind lets pass. Even if a question has potentially damaging implications for the self, such as whether one stole something (when one in fact did

steal it), the subconscious foolishly churns out the correct answer: yes. The conscious mind, however, in effect says (and this is a quotation from the PSE examiner), "You dummy, I'm not going to say that. It will get me in trouble." Hence the subject gives a false answer to the question: no. Whenever one lies, then, there is a difference between the responses of the unconscious and conscious components of the mind. That difference produces psychological tension, or stress, and that in turn affects the voice in a way that is measurable by the psychological stress evaluator. The logic of signification is the same as for the polygraph, operating on the assumption that perceptible effects signify, by metonymy, concealed causes. The word "stress," incidentally, was frequently used both as a noun and as a verb by the two PSE examiners I interviewed. Whereas polygraph examiners tend to say the subject "hit" on a certain question, the PSE examiners would say the subject "stressed" on it.

As with the Israeli remote sensor, it is possible to give PSE tests without the subject being aware of it. A polygraph examiner told me that some insurance companies record telephone conversations with policyholders concerning loss claims so as to subject them to analysis by the PSE. Another pointed to a PSE tape prominently displayed on the wall of his office. It was, he said, the voice of Patty Hearst as recorded from a telephone conversation while she was with the Symbionese Liberation Army, and he indicated several points on the chart where she stressed.

The lie detection profession itself has stressed on the question of the PSE. PSE examiners argue that their newer and simpler technique is as accurate as the polygraph. Polygraph examiners, who are much more numerous and whose technique is more common, tend to dismiss the claim, often with expressions of contempt not unlike those that critics of the whole idea of lie detection level against the polygraph. Thus an unfriendly schism has emerged between PSE and polygraph examiners.

Uses of Lie Detection

The U.S. Congress Office of Technology Assessment has estimated that as of 1987, some two million polygraph tests were

given annually in the United States.^[20] Lie detector tests are regularly employed by the CIA and the National Security Agency to detect spies or double agents in their midst. Police departments test criminal suspects and their own personnel in various internal investigations. In 1972, Imperial Wizard Robert Shelton ordered Ku Klux Klansmen to take lie detector tests in an effort to root out informers and FBI plants in their midst.^[21] George Steinbrenner had lie detector tests administered on several occasions to New York Yankees executives when he was angered that news of trades leaked out before the official announcements.^[22] In 1985, the Reagan administration was planning to plug leaks by demanding lie detector tests of high government officials, including Cabinet members, until Secretary of State George Shultz shot the plan down by saying he would resign rather than submit to such a demeaning and ineffective procedure.^[23] One examiner told me that he was engaged by the church to test a Catholic priest who was accused of having an affair with a female parishioner. The church fathers decided they would take the opportunity to also look into the state of the priest's faith. In addition to questions about his relationship with the woman, while the priest was hooked up to the machine, he was asked if he assented to the various propositions of the Nicene Creed. It seems that he passed that part of the test with flying colors, although he did not do so well on the morals charge.

Prior to the 1988 legislation banning it in the private sector, by far the most common users of lie detector tests, accounting for 90 percent of the two million annual tests estimated by the Office of Technology Assessment, were private businesses whose employees regularly handle cash or items of value, such as retail firms, pawn shops, and banks. Businesses using the polygraph included over 30 percent of the Fortune 500 companies and at least half of the retail trade firms.^[24]

Business turned to lie detection in an effort to control employee theft. Not many years ago, retail firms chalked up most of their unaccounted losses to external factors such as shoplifting. Now they attribute the lion's share to internal theft. Not that the reality has changed. Instead, managers have reluctantly come to the conclusion that their employees simply

do not (and never did) deserve the trust that was once placed in them. Although internal disciplinary measures or termination are much more common responses to employee theft than legal action, in 1982, some 335,000 American workers were arrested for stealing from their employers. When employees who are not prosecuted and those who are never caught are added, the totals become impossible to verify. Estimates vary widely; among the more pessimistic are that 70 percent of all workers steal something during their employment, that employee theft is responsible for some one-third of all business bankruptcies, and that annual losses suffered by American business due to employee theft total $40 billion or more.^[25] Staggering as this total is, some perspective is provided by a National Institute of Justice estimate that corporate management steals three times that amount through securities fraud, corporate kickbacks, embezzlement, and insurance fraud. This moves critics of the business use of lie detection to ask why hourly employees have been forced to submit to lie detector tests while executives are rarely tested.^[26]

However they may answer that question, employers have sought to combat employee pilfering and other problems by applying lie detector tests in four main circumstances: preemployment screening represents an effort to avoid hiring untrustworthy employees; current employees may be kept under surveillance and deterred from unacceptable behavior by periodic tests (given at regular intervals) or random tests (given unannounced at any time); and specific tests are given in the course of investigations into thefts or other particular acts of wrongdoing that have occurred. (Tests of the last sort are also used by law enforcement agencies in criminal investigations.)

Until very recently the use of polygraph tests by American business was dramatically on the rise, having tripled in the decade prior to 1987.^[27] Lie detection expanded in this period not only because of the growing perception that employee theft was more common than had been previously suspected but also because legislation protecting employee interests made it extremely difficult to learn about applicants from previous employers. To protect themselves from possible legal action, employers will

often divulge no more about former employees than the period of their employment and the position they held. Potential employers are interested in knowing more about their work history: their reliability, congeniality, honesty, and habits that may have adverse consequences for the company, whether they are likely to agitate for a union or make other trouble, why they left earlier jobs, and so on. Two issues of special interest are drugs and homosexuality. Drug use may have an impact on job performance and place the employer in serious legal jeopardy if the employee is in a job that affects the safety of others. Moreover, drug addiction is extremely expensive, and addicts frequently steal to support their habit, often from their employers. Intravenous drug users and homosexuals represent high-risk categories for AIDS, and this disease is so expensive to treat that one or two cases have serious repercussions for a company's health insurance program. Unable to ascertain information of these sorts from previous employers, many companies resorted to lie detection as a means to discover it for themselves.

If the experience of the Vermont State Police is any guide, what they discover may be startling. A preemployment lie detector test was used in the screening of the 203 individuals who applied for positions in 1983. Nineteen applicants did not even show up for their scheduled tests, and only 75 were retained for further consideration after the test. The other 109 admitted, under the scrutiny of the polygraph, to a total of 238 kinds of disqualifying offenses. These included drug use, larceny, lying on their application forms, abnormal sexual practices, and "immature acts." Sixty-five of the 75 who "passed" the lie detector examination made admissions of misdeeds that were not considered serious enough to drop them from further consideration.^[28] One wonders, incidentally, if any of those who did not show up for the test or were disqualified by it had a record as distinguished as a former police officer who (unsuccessfully) applied for a job in the Washoe County (Nevada) Sheriff's Department. In the course of the preemployment lie detector test, that worthy "admitted accepting bribes on two occasions, 'rolling' a homosexual and beating him to a point that required hospitalization, stealing property while answering an alarm at a burglary scene, and ransacking the

apartment at a death scene and stealing the property of the deceased."^[29]

Some employers use lie detection to weed out not only applicants who have engaged in criminal behavior on previous jobs but also persons who are thought to be potential troublemakers or union agitators. The following assessment of the polygraph was made in 1984 by a uniform dry cleaning chain in Buffalo, New York.

The polygraph very definitely has improved industrial relations at our facility and every other non-union facility where it is used properly. The simple reason is that with polygraph an employer tends to eliminate bad employees, those who are deep into drugs, dishonest, troublemakers, etc. . . . In a union facility, usually the bad employees are protected by the union bosses because they are strong union supporters and when there is polygraph involved these bad employees will do everything in their power to fight it. . . . Trade unions and civil rights organisations continually complain about polygraph, apparently because it promotes good relations between employer and employee and when you have these good relations you likely will not have employees voting a union in. . . . A refusal to take a polygraph test is taken into consideration along with other information at hand in making a decision to promote or to hire. Polygraph is the most valid aid in selecting staff known to man. . . . The polygraph, properly utilized, prevents labor disputes.^[30]

It has even been possible to exert the force of lie detection in the absence of actual tests. The application forms used by many companies included a question about whether the applicant would be willing to take lie detector tests prior to being hired or as an employee with the firm. Often companies did not actually give tests to job applicants because they are too expensive. But a fast-food chain proprietor told me that if a person answered "no" to that question, "unless the application is incredible , there is no way that we would consider that applicant further." The reason she gave is that "philosophical" problems with taking a lie detector test can signal an overly independent turn of mind that might lead to noncompliance with other company policies down the line.

The Test

Because the polygraph is a portable machine, a lie detector test can take place in almost any surroundings. Sometimes subjects are directed to go to the examiner's office for the test. Other examiners may visit the company or other locale where people are to be tested and set up shop in any available room or office. Some examiners prefer to bring their own office with them and work out of a trailer or camper in the parking lot. In general, the room in which the test is given should be reasonably plain so as not to distract the subject. Some polygraph examiners, in an effort to heighten subjects' perception of the test as a professional procedure, go so far as to wear a white coat, stethoscope dangling from the neck, and to spray the air of the examining room with ethyl alcohol.^[31] About the only piece of furniture specifically designed for lie detection is a straight-backed chair with concave armrests long enough to provide support for the subject's hands, where the fingertip clips are attached. This is not essential, however, for the test can be effectively conducted with the subject's arm resting on a table or on chair arms of normal length.

The typical lie detector test begins with a pretest interview. The examiner explains how the polygraph machine works and reviews the questions to be asked. While most polygraph examiners agree that the pretest interview is an extremely important part of the overall process because it sets the general ambience for the test, they are not of one mind as to precisely what that ambience should be. Some attempt to set the subject at ease, while others strive to increase nervous tension. In the latter case, the examiner's goal is often not so much to conduct a reliable polygraph test as it is to extract a confession. A police officer told me that occasionally the polygraph is used in this manner as a last resort in cases where the evidence against a suspect is not conclusive and the likelihood of a conviction is small. The suspect is told that things look very bad but that it might be possible to clear the record by means of a polygraph test. If the person can be induced to confess during the course of the test, charges will be filed (on the basis of the confession, not the polygraph test). If not, the suspect will

be released regardless of what the polygraph chart shows about innocence or guilt (the evidence of polygraph tests rarely being accepted in court).

The polygraph is an effective tool in the hands of a skilled interrogator, who can use it to help convince suspects that their lies are not fooling anyone. When the aim is confession, other interrogation techniques may be used in the pretest interview, such as the suggestion of one polygraph instructor that the questioner sit close to the subject and that there be no table or other obstacle between them. Any kind of obstacle gives the person being questioned a certain degree of relief and confidence. The questioner may start with his chair two or three feet away and move closer as the questioning proceeds, so that ultimately the knees are in close proximity. This physical invasion of the subject's territory by the questioner, the crowding in as he is questioned, has been found in practice to be extremely useful in breaking down a subject's resistance. When a person's territorial defenses are weakened or intruded on, his self-assurance tends to grow weaker.^[32]

Whether or not the examiner wishes to intimidate the subject, a ubiquitous aim of the pretest interview is to convince the subject that the polygraph really works. "The polygraphist . . . ," writes Abrams, "must engender enough of a feeling of confidence in the subject to relieve the anxiety of the innocent and at the same time increase the fear of the guilty."^[33] Fred Inbau and John Reid suggest that the examiner tell the subject quite emphatically, "if you're telling the truth this machine will show it; if you're not, the machine will show that, too."^[34] An examiner I interviewed habitually tells subjects just prior to the test, "Every story has three sides—your side, his side, and the truth. This machine is going to get the truth."

Lie detector tests are used not only to ascertain the subject's own improprieties but also to learn what one may know, or suspect, about the wrongdoings of others. The stage is often set for this during the pretest interview by convincing the subject that "divulgence of the suspicions is necessary for the subject's own test purposes."^[35] The polygraph, that is to say, is presented as a machine with rather mysterious properties that necessitate that a

person betray any suspicions about other employees or suspects to be found innocent oneself.

One common means of instilling respect for the machine is to run a "stim" (stimulation) test prior to the actual test.^[36] The examiner, saying that he is going to demonstrate the accuracy of the polygraph, hooks the subject up to the machine and directs her, in one type of stim test, to draw a card from a deck. The subject is told to sit still, look straight ahead, and, for the purposes of the stim test, answer "no" to all questions. The examiner proceeds to ask if it is any of a number of possible cards and then identifies the correct card on the basis of the polygraph chart readings. (Some polygraph examiners, it seems, cover their bets on the stim test by using a trick deck of cards.)^[37]

Actual examinations vary according to whether they are concerned with the subject's general honesty and history of wrongdoing (as in applicant screenings and periodic and random tests of current employees) or are part of the investigation of a specific crime. In general, however, the test consists of some ten to fifteen questions of three basic types. Irrelevant questions have to do with nonthreatening matters regarding which the subject can fully be expected to give honest answers: questions such as "Is today Thursday?" or "Do you live in Chicago?" or "Were you born in 1946?" The chart readings for these questions depict the subject's physiological profile when responding truthfully. Control questions are designed to be threatening and to evoke an untruthful response from anyone. These may vary a good deal with the particular sort of wrongdoing one is interested in uncovering and the particular history of the subject. For a person with a criminal record, a control question might be, "Did you ever commit a crime that was not found out by the police?" A person being investigated for assault might be asked, "Did you ever desire to hurt anyone?" An applicant in a preemployment screening might be asked, "Did you ever steal anything?" The chart readings for control questions show the physiological signs of anxiety connected with deception. Finally, the relevant questions (or "hot" questions, as they are often called in the trade) pertain to the particular issue under investigation. If this is a specific crime, the question focuses directly on it, such as, "Did you rape Judy Barnes

on July 17?" or "Do you know who stole $500 from the cash register two weeks ago?" Relevant questions in periodic tests of current employees or applicant screenings are perforce somewhat more general, such as, "Have you stolen any money from the company during the last six months?" or "Did you steal any merchandise from your previous employer?" or "Have you ever been fired for reasons of dishonesty?"

Analysis of the chart is basically a matter of comparing responses to the relevant questions with those to the irrelevant and control questions. If the subject is lying in responding to the relevant questions, the readings for them will resemble those for the control questions; if the subject is telling the truth, they will resemble the readings for the irrelevant questions. Or, phrased somewhat differently, a deceptive subject will show greater perturbations for the relevant questions than for the control questions, while a nondeceptive subject will react more strongly to the control questions. The reason is that the guilty individual is more threatened by the relevant questions than the control questions, while the innocent individual, having nothing to fear from the relevant questions, is more threatened by the control questions.^[38] Test results are usually reported as "DI" (deception indicated), "NDI" (no deception indicated), or "Inconclusive." The last is used when no clear pattern is discernible on the charts.

Polygraph examiners are interested in creating conditions in which the charts will be as "good" as possible; charts, that is to say, that lend themselves most unequivocally to an interpretation of either DI or NDI. Examiners may differ, however, on how to bring about these optimal conditions. According to Phillip Davis and Pamela McKenzie-Rundle (herself a former polygraph examiner), the subject who has been "pumped up" during the pretest interview is most likely to produce the sharply different responses to relevant, irrelevant, and control questions that are essential for "good" charts.^[39] However, a polygraph instructor told me that his main goal is to put subjects at ease during the pretest interview, partly to facilitate the production of "good" charts. An overly nervous subject, he explained, is likely to give erratic responses even to the irrelevant questions. This makes the chart as a whole much more difficult to interpret and produces a result of Incon-

clusive. Moreover, the subject who has been "pumped up" by the test situation may even be driven by the pressure to make a false confession.

A careful polygraph test involves running several charts, that is, going through the same set of questions three or more times, with periods of interview/interrogation in between. This provides additional opportunities for the subject to make a confession, or, conversely, for areas of apparent deception or guilt that showed up on the first chart to be resolved in the subject's favor. In either event, the charts and the overall result of the test are clarified with repetition.

One technique for such clarification, which seldom works to the benefit of the subject, is the following:

Once a test has been administered to a guilty individual it is extremely effective to display the records to him and point out the deception criteria—at the same time reminding the subject that the recordings represent his own heart beats, his own blood pressure changes, etc., and not those picked up by the machine out of thin air or placed there by the examiner.^[40]

A particularly powerful technique is to go through the test two or three more times, pointing out to the suspect how, with each successive chart, the indications of deception become more pronounced. Eventually the evidence from the polygraph becomes so self-evident that only the most intransigent subject can avoid confessing.

But, of course, the alternative explanation is not difficult to imagine. Subjects "hit" more and more decisively on the relevant question with repeated testing not necessarily because of guilt but because they are distressed about how the response to it appears on the previous chart and are increasingly apprehensive about it when the test is run again.^[41] This, incidentally, sheds additional light on the practice of reviewing the questions with subjects prior to administering the test. This is done, they are told, as an assurance that no surprise questions will be sprung while the chart is running. That is true enough, but the practice also has certain advantages for the test itself which are not so

immediately apparent. For one, a subject might be surprised by an unexpected question, and even if the person has nothing to hide on that question, the surprise itself could produce physiological responses that might be difficult to distinguish from those associated with deception.^[42] For another, to review all questions in advance alerts deceptive subjects to just when in the test a relevant question is coming. Their attention focuses on it, and they become increasingly apprehensive as it approaches and then relax after it has passed. This produces a characteristic response pattern known as "peak of tension," which is taken as particularly damning.

Is is possible, however, that the peak of tension might also characterize the responses of innocent individuals who go into lie detector tests knowing full well that they are suspected of certain misdeeds and who therefore are apprehensive about the relevant questions. This is obviated in another testing format, known as the Guilty Knowledge Test.^[43] Suitable only for investigations into specific acts of wrongdoing and not for general preemployment or periodical checkups, the Guilty Knowledge Test trades on information that only the guilty individual would know. The precise location of a rape, for example, or some characteristic of the clothes the victim was wearing might not have been made public. One could then construct a polygraph test with a series of alternatives, for example, she was wearing a red blouse, a blue blouse, a sweatshirt, a woolen sweater, and so on. Regardless of how much anxiety an innocent suspect might have about the test, he would not be likely to "hit" on the correct alternative because he simply does not know it. The guilty suspect, however, is much more likely to react. And if the possibilities and the order in which questions would be asked were reviewed in advance, it is reasonable to expect that the responses would manifest the typical peak of tension pattern.

Polygraph examiners I have interviewed enjoy telling stories of their most interesting cases, including their greatest triumphs. One of these was a classic use of the Guilty Knowledge Test and is interesting in addition because it demonstrates that the verbal answers given by the subject are really not an essential part of the test. A murder suspect agreed to be hooked up to a polygraph but

refused to answer any questions. "That's all right," said the examiner, "all you have to do is to sit there." He informed the subject that they were going to find out just where the body was hidden. He produced a map of the city, divided it into quadrants, and, pointing to each in turn, asked, "Did you hide the body in this section? In this section?" and so on. Although the subject sat mute, the chart showed a larger response for one of the quadrants than for the other three. The examiner divided that section into quadrants and repeated the questions. The narrowing process continued with, it can be imagined, the subject becoming increasingly apprehensive. Finally, a bit of territory the size of a house lot was specified. The police dug there and discovered the body.

Polygraph examiners may use ingenious—if on occasion remarkably simple—techniques to ascertain if someone is lying. Keeler, who also tried the technique of identifying the location of a body by means of pointing to quadrants on a map,^[44] was once asked to use the polygraph to determine if an individual who claimed to be blind was in fact so. Keeler did not find it necessary to ask any questions of the individual at all. He simply hooked him up to the polygraph and then held a picture of a nude pinup girl in front of him. The needles went wild, and they had their answer.^[45]

Polygraph subjects too have developed a set of techniques—some simple and some more ingenious—to "beat" lie detector tests. The technique that usually first springs to mind, especially among persons with little experience of the polygraph, is to maintain such rigid control over one's physiology that no telltale perturbations will disturb the polygraph when one lies. This is extremely difficult to accomplish, although in one case an enterprising individual may have made it work. A highly experienced examiner told me about a subject whom he had tested before and whose previous tests had indicated a great deal of deception. This time, however, the charts indicated no deception at all; they looked, indeed, too perfect. The examiner said to the subject, "This is too good to be true. What are you doing?" The subject, apparently prouder of his ruse than apprehensive about being discovered in an effort to beat the test, unbuttoned his shirt and

displayed his torso, completely wrapped in aluminum foil. In principle, this should have no effect whatsoever on the polygraph readings, but the examiner surmised that the trick gave the subject such confidence that he succeeded in muting the physiological responses that normally accompany deception.

Another technique is to use some substance prior to the test in an effort to mask one's responses. Typewriter correction fluid such as White-Out is believed by some to be effective for this purpose. (Does the rationale have to do with some supposed generic capacity of the liquid to cover things up?) Subjects have been known to paint their fingertips with it to thwart the galvanic skin response measure—surely a ruse that would not be difficult to detect. Alternatively, the subject can drink it. One individual who was told by a friend that he could beat the test with correction fluid "drank five bottles of White-Out, threw up during the pretest interview, and confessed."^[46]

The most effective way to thwart a lie detector test, however, is not to attempt to minimize one's reactions to the relevant questions but to maximize them on the irrelevant and control questions. Chart analysis rests on identifying differences among these three types of question. The assumptions are that all subjects will show little reaction to irrelevant questions, that an innocent individual will react more strongly to control questions than relevant questions, and that a guilty subject will show the reverse profile. Thus, a conclusive finding of deceptiveness is not possible if the responses to control questions are stronger than to relevant questions,^[47] while the entire test is thwarted if responses to irrelevant questions are stronger than those to relevant or control questions.

Subjects may resort to a variety of techniques if they wish to intensify their responses to irrelevant or control questions. Pain is effective for this, and one practice is to come into the examination room with a tack in one's shoe and to press the foot down on it when one wants to provide a heightened response. Somewhat simpler is to bite one's tongue; easier still is to tighten the sphincter muscle, which produces a minor perturbation in the blood pressure. Finally, it is easy for any subject to confound a lie detector test simply by refusing to sit still. Wriggling, coughing,

and intentionally varying the rate of one's breathing all have the effect of defeating the polygraph's physiological measures. A common method to distort the readings produced by the psychological stress evaluator, which measures voice patterns, is to wear a necktie and to press one's throat against its large Windsor knot while answering irrelevant or control questions.

Many countermeasures are not difficult for the examiner to detect. People with tacks in their shoes often limp painfully into the examination room. Some examiners have subjects sit on an air-inflated pillow that is connected to the polygraph machine in order to detect sphincter tensing. And, of course, subjects who deliberately cough and squirm are often not even trying to hide their refusal to cooperate.

A deceptive subject who uses subtle techniques that escape the notice of the examiner may achieve the false negative test result of NDI (no deception indicated). The examiner who detects efforts to defeat the test is likely to attempt by cajoling or by threats to convince the subject to desist from them and to cooperate fully with the test. If the subject will not do so, it will be impossible for the examiner to reach a conclusion of DI (deception indicated) on the basis of the charts. The only option is to report an inconclusive test result, although very possibly with a notation on the report that the subject's refusal to cooperate might in itself indicate deception.

The Examiner

Examiners vary greatly in their interest in and skill at detecting efforts to beat the test and in all other aspects of polygraph tests. Therefore, many experts in lie detection strongly advance the opinion that the examiner is the single most important factor in any lie detector test. Professional quality of examiners is a sore point in lie detection, for licensing regulations vary widely among states (and are completely absent in some), so it has been possible in many places for a high school graduate with less than six weeks of specialized training to hang out a shingle as a professional lie detector examiner.^[48] There is, moreover, considerable

incentive to do just that, for lie detection can be a fairly lucrative profession. The fee for preemployment and periodic tests varies from about $25 to $75 or more, while tests in specific criminal investigations, done in more elaborate circumstances with attorneys present, may cost more than $200. Businesses are understandably concerned to keep their costs down, so many of them opt for lie detector tests at the cheaper end of the scale. Some examiners accommodate this by offering low-cost tests, but then they attempt to maximize their income by increasing their volume. In 1972, for example, one examiner in Alexandria, Virginia, earned a handsome living by giving more than 2,000 lie detector tests annually (an average of 8 per day for 50 five-day weeks), although he was scarcely at the lower end of the cost scale because even at that time he charged $75 per test.^[49]

Examiners I interviewed frequently stated that the greatest threat to their profession is the presence of incompetent and/or unscrupulous individuals in their midst. They call them "chart rollers," and they accuse them of producing highly unreliable results through sloppy testing. It is essential for valid results, examiners told me, to ascertain whether the anxiety that accompanies the response to certain questions is produced by deception or by some other circumstance. This often involves running several charts and long interviews, in a process that may require two or more hours. The process is so mentally demanding on the conscientious examiner, one told me, that he is utterly exhausted if he does two tests in a day. Chart rollers, whose primary concern is to maximize the number of tests they give, are unwilling to devote the necessary time and care to this imperative. They do a perfunctory pretest interview, run a single chart, simply report "deception indicated" for the questions where stress is recorded, and move on without exploring the issue. Thus their rate of "false positives" (erroneous conclusions that the subject is lying) is high. A grocery store manager recounted an experience with a chart roller who polygraphed two cashiers for "sweetheart checking"—a procedure whereby the cashier charges the customer (a relative or friend) far less than the actual price, such as ringing up a ham as a pack of chewing gum. The polygraph results indicated deception by one of the checkers, and the examiner

reported that she had been giving merchandise to her brother, John. She was fired but then sued for wrongful termination, pointing out that she did not have a brother named John. The store settled out of court and decided it was time to get someone else to do their polygraph testing. One examiner said that the most unscrupulous chart rollers will run people through in about ten minutes, sometimes without even hooking them up to the machine, and then provide a fabricated report and charge the fee for a completed test.

The PSE has perhaps been especially attractive to would-be chart rollers because the training program is shorter still—only two weeks, as one PSE examiner told me. The PSE is also appealing to business because it often costs less than the polygraph. The cheapest rate I have come across is from a policeman trained as a PSE examiner, who told me that he used to moonlight by doing PSE tests for $8 each. Under his system, the employer would ask the questions, tape record the subject's responses, and send the tape to the examiner. He would analyze the tape on the PSE machine, identify the questions on which deception was indicated, and mail his report to the employer. The report would recommend that the employer discuss those questions with the applicant to determine if the stress that showed up on the chart stemmed from deception or some other cause, but he does not know if employers actually followed up on the recommendation. Certainly the PSE examiner did not do it because he never set eyes on the applicants. This seems to be a highly evolved form of chart rolling. When I asked him why he did it, he replied, "If some fool pays some other fool to do that, why not?"

Partly in an effort to achieve a more respectable image and to work toward establishing self-policing ethical and professional standards, polygraph examiners formed a professional society, the American Polygraph Association (APA). The APA publishes the journal, Polygraph , and attempts to present polygraph tests and polygraph examiners in the most favorable light possible to the outside world. The seal of the association is the blindfolded figure of Justice, holding a balance scale in her left hand. But in her right hand, instead of a sword, she holds a long polygraph

chart, one end of which trails on the ground near her feet. The motto "Dedicated to Truth" is printed beside her.

Another effort to police the profession was through state licensing of polygraph examiners. While this did not exist in all states, the concept came to be supported by many polygraph examiners as a means of rooting out chart rollers, giving them the aura of true professionals and fending off moves to curtail lie detection by legislation. The importance placed on state licensing is clear from the following passage by two polygraph practitioners.

State polygraph licensing boards began to crop up [in the early and middle 1960s] and make purposeful headway in the investigation of public complaints that alleged personal, embarrassing, or sexually oriented questions associated with polygraphy. The word spread quickly. A good number of examiners had their license to practice suspended or revoked. It was not long before that kind of complaint became a rarity. By the end of 1982, polygraph licensing boards were tougher than boot leather. As a result, strict licensing administration and regulation became a blessing to ethical examiners.^[50]

Legal Control of Lie Detection

The salutory promise of a professional association and state licensing turned out, however, to be wishful thinking. Public suspicion of lie detection persisted, and by the end of the 1970s, at least sixteen states had enacted laws designed to control or prohibit lie detector tests in the workplace.^[51] According to polygraph proponent Ferguson, who welcomes tough state licensing of polygraph examiners, such legislation was really a grotesque aberration of a dark time in the nation's history. It was because of the liberal-dominated 1960s, he explained, that several states "let witch hunts initiated by Mob controlled labor unions, and other rotten-stinking sources, cause 'elected' state representatives to enact ridiculous antipolygraph statutes of the flimsiest sort."^[52] But legislative threats to the polygraph were also emerg-

ing on the federal level. Polygraph control bills were proposed by Senator Sam Ervin (North Carolina) in 1971 and 1973 and by Senator Birch Bayh (Indiana) in 1977.^[53] These bills were not adopted, but in 1987, Representative Pat Williams (Montana) introduced another bill that successfully passed through the entire legislative process and became law. The Employee Polygraph Protection Act, which took effect on December 28, 1988, drastically curtailed the use of lie detection in the United States. In essence, this law prohibits most private sector employers from requiring prospective or current employees to submit to preemployment or periodic lie detector tests. An employer may still request an employee to submit to a lie detector test by polygraph (but not by PSE or other devices) if there is reason to suspect the employee of some specific wrongdoing. All local, state, and federal governmental agencies are exempt from the law and may therefore continue to use any form of lie detection for employment or any other purpose. A number of private firms are also exempt from the provisions of the law, notably those under contract for sensitive work for the Department of Defense, CIA, or FBI; those involved in the manufacture or distribution of controlled substances; and those in the security guard, armored car, or security alarm fields. These exemptions mean that lie detection will certainly not disappear from the American scene. Nevertheless, it will be dramatically reduced: some 80 percent of the two million lie detector tests given annually in the United States fall under the law's ban.^[54]

Those who supported the bill argued that lie detection violates constitutional and legal rights to privacy, to due process of law, and to be presumed innocent until proven guilty.^[55] But perhaps the most important concern that led to the legislation had to do with the accuracy of lie detector tests. Especially disquieting is the problem of false positives: that lie detector tests label as dishonest some people who in fact are honest. The particular circumstances of lie detection in the employment setting produce a magnification effect with reference to false positives, such that even with an accuracy rate of 85 percent, the tests generate some alarming results. Representatives of the American Psychological Association explained it in testimony before the

House Subcommittee on Employment Opportunities in the following way:

Assume that polygraph tests are 85 percent accurate, a fair assumption based on the 1983 OTA report [on the validity of lie detector tests]. Consider, under such circumstances, what would happen in the case of screening 1,000 employees, 100 of whom (10 percent) were dishonest. In that situation, one would identity 85 of the dishonest employees, but at the cost of misidentifying 135 (15 percent) of the honest employees. As you can see, in this situation the polygraph tester identifies 220 "suspects," of whom 61 percent are completely innocent. It can be shown mathematically that if the validity of the test drops below 85 percent, then the misidentification rate increases. Similarly, if the base rate of dishonesty is less than 10 percent, and it most likely is, the misidentification rate increases. It is obvious that in the employment screening situation it is a mathematical given that the majority of identified "suspects" are in fact innocent!^[56]

Congress found it to be intolerable that preemployment screening by lie detector tests would reject more honest than dishonest applicants, especially when the stakes involve people's reputations, self-esteem, and opportunity to earn a living.

If lie detection is so drastically flawed, one wonders why the Employee Polygraph Protection Act has so many exceptions. Is it not hypocritical, as some members of the House Committee on Education and Labor registered in their dissenting views on the bill,^[57] to marshal such powerful arguments for outlawing lie detection in the private sector but then to place no impediment whatsoever on its continued use for any purpose in the public sector? Indeed, one journalist reports that lie detection has actually increased in the federal government since the bill went into effect late in the Reagan administration: "The Reagan and Bush administrations have vastly expanded the use of polygraphs in recent years, routinely administering them in the course of 'leak' and national security investigations."^[58] In the Conference Report that adjusted differences between the Senate and House versions of the bill, the exemption of public sector employees was explained as a matter of legislative committee jurisdiction:

By exempting public sector employers and private contractors engaged in intelligence and counterintelligence functions, the conferees recognize the functions performed by these employers are not within the jurisdiction of the committees which reported the legislation, and the policy decisions as to the proper or improper use of such tests are left to the committees of jurisdiction and expertise.^[59]

The Conference Report does not explain, however, why exemptions from the bill are also granted to private companies that operate armored cars, other security service firms, and companies that manufacture or dispense controlled drugs.^[60] Contradicting the arguments about polygraph inaccuracy on which the bill is supposedly based, continued use by the government and these private sector exemptions seem to betray a belief that lie detection really is an effective tool for identifying dishonest employees and applicants, even if certain excesses and imperfections compromise their civil rights. But, the thinking seems to go, that is justified for people charged with keeping political secrets or with special responsibilities in the war on drugs or the custody of money.

A similar ambivalence surfaced during the Senate Judiciary Committee hearings on the sexual harassment charge raised against Supreme Court nominee Clarence Thomas in October 1991. Polygraph tests seem to have a special appeal in cases of sexual harassment, which by their nature tend to occur in private, with no disinterested witnesses available. If in past centuries disputes that came down to the word of one litigant against another recommended soliciting the opinion of God via trial by ordeal or battle, in our own time, a lie detector test promises to cut through efforts at dissimulation and get the answer directly from the principals themselves. Judge Thomas's accuser, Anita Hill, did in fact take, and pass, a polygraph test. Although all parties acknowledged difficulties with polygraph tests and their inadmissibility as legal evidence, this was perceived by some as speaking in her favor. Senator Paul Simon (Illinois) noted, "I don't find generally that people who are not telling the truth volunteer to take lie-detector tests,"^[61] while Senator Patrick Leahy (Vermont) said that the test result supported her credibility.^[62]

Curiously, although the test was mentioned by senators and members of the media many times during those tension-filled days of October (and roundly scorned by avid supporters of Judge Thomas such as Utah Senator Orrin Hatch), hardly anyone thought to point out that Congress itself had passed a law against lie detector tests just three years before.^[63]

Written Integrity Tests

Also not covered by the Employee Polygraph Protection Act are paper-and-pencil tests that purport to measure honesty, such as the Reid Report, the Personnel Selection Inventory, the Stanton Survey, and the Wilkerson Pre-Employment Audit. R. Michael O'Bannon, Linda Goldinger, and Gavis Appleby identify forty-six of these so-called honesty or integrity tests currently on the market.^[64] Most of them are designed for preemployment screening, although a few are for use with current employees. Many employers seem to be turning to them as a replacement for the now-banned polygraph. Test publishers report dramatic increases in sales since the antipolygraph act was passed. Approximately 2.5 million such tests are administered annually by an estimated 5,000 to 6,000 companies.^[65] This is already a considerable increase above the two million polygraph tests administered in the United States in 1987, well before the antipolygraph law went into effect.^[66] Budget-conscious employers might have reason to prefer them to polygraph tests because they are considerably cheaper. They cost from $4 to $35, with most tests clustering in the $6 to $15 range.^[67]

Questions are of two basic types. Some solicit outright admission of previous wrongdoings, while others explore attitudes and personality characteristics that may be correlated with honesty of dishonesty. The former ask whether, during the last few years, the subject has shoplifted, broken into a home or business, stolen from an employer, used an alias, driven while intoxicated, written a bad check, had a friend who steals, and so on. Attitude-probing questions might ask the subject to respond with "agree," "disagree," or "undecided" to a broad series of propositions, such

as that it is acceptable to break unjust laws, that people are usually not what they seem to be, that a store employee should not be fired for taking pens and pencils, that it is necessary to have laws governing morals, that smoking a marijuana cigarette is no worse than drinking a Coca-Cola, that employers should not be concerned about drug or alcohol use so long as the employee performs well on the job, that a thief deserves a second chance, that most people cheat on their income tax, that the subject has occasionally felt like breaking something, would never lie, sometimes feels sorry for oneself, has wild daydreams, likes to create excitement, has not had many lucky breaks, would prefer to talk to an interesting person at a social gathering, sometimes goes to places where it would not be good to be seen by an acquaintance.

Integrity tests and scoring techniques come in a variety of formats. In some cases, an answer key and instructions for scoring are included with the packet of tests. In others, written answer sheets may be mailed to the test publishing company for scoring, or the employer may have the result immediately by calling the company and reading the answers over the telephone. Still other tests are taken on a computer and scored automatically.^[68] Telescreen, Inc., markets a test that is taken by means of a touch-tone telephone. The subject listens to questions on the telephone and has three seconds to punch keys signifying "yes," "no," or "not applicable."^[69]

Integrity tests have been developed outside the mainstream of psychological testing, often by polygraphers and always by companies interested in marketing them. Nearly all research on their validity and reliability has been conducted by persons associated with those companies rather than by independent scholars and is unpublished or appears in reports issued by the test publishers rather than in peer-reviewed journals.^[70] Although most of that research reflects favorably on integrity tests, one reviewer has been quoted as saying, "As an analogy, would we consider it good science to publish a review of research on the effects of smoking on health when almost every study was supported by the Tobacco Institute?"^[71]

Even the research on integrity tests conducted by interested parties raises grave suspicion as to their value. The Office of

Technology Assessment identified five studies, all conducted by the publishers of the tests under study, where applicants in retail enterprises were tested and hired regardless of the test result. Their work records were scrutinized to determine how many of those identified as dishonest by the test were found to engage in theft or other dishonest behavior on the job. Study results are summarized in table 1. The briefest examination of these figures indicates that something is seriously wrong with integrity testing. It is, of course, unlikely that all dishonest employees will be apprehended, but it is alarming that (except for Study 4, with its small sample size) those labeled dishonest by integrity tests outnumber those who are actually caught in dishonest behavior by between ten and twenty to one. In four of the five studies (again Study 4 is the odd one out), over 40 percent of those who took integrity tests failed them, and in two of the studies (including Study 2, with by far the largest sample), the dishonest ones represent large majorities. If we are to believe the testimony from representatives of the American Psychological Association that

10 percent or less of workers are dishonest, it is obvious that staggering numbers of honest individuals are branded as dishonest by integrity tests. Reversing the principle of the American justice system that it is better to acquit several guilty individuals than to convict a single innocent one, integrity tests seem to be more akin to the philosophy that "if you hang 'em all, you'll be sure to get the guilty ones."

The Office of Technology Assessment report is a carefully researched document that finds inadequate evidence to confirm or deny the claim that integrity tests predict dishonest behavior^[72] but nevertheless identifies sufficient problems with them to reach the final conclusion that "the potentially harmful effects of systematic misclassification, possible impacts on protected groups, and privacy implications of integrity tests combine to warrant further governmental attention."^[73] Another report on integrity testing, prepared by a special task force of the American Psychological Association, is also critical of integrity tests at numerous points. Many tests have cutoff scores that mark the point between passing and failing or that define zones such as high, medium, and low risk. But information about how test publishers have determined the proper location of cutoff points is insufficient or absent. (The data in table 1 concerning how many pass and fail integrity tests dramatize the cogency of this criticism.) The report also chides some publishers for promoting their tests with fraudulent claims and complains that research and evaluation of test reliability and validity are seriously hampered by the fact that test publishers often conceal the necessary information under a cloak of proprietary interests. Test results could well be affected by insufficient attention to linguistic and cultural variables when tests are translated into other languages such as Spanish.^[74] Curiously, however, and especially in light of the strongly negative testimony from representatives of the American Psychological Association in the congressional hearings on polygraph tests, this report is equivocal. In tenor, it tilts toward employers' needs to curtail theft and promote productivity as against the interests of employees in privacy and personal dignity. Among the assumptions that guided the investigation, the one they identify as "perhaps of the greatest importance" is the

highly pragmatic consideration that if integrity tests were not available, employees would subjected to something worse.^[75] Hence the bottom line:

Despite all our reservations about honesty tests, however, we do not believe that there is any sound basis for prohibiting their development and use; indeed, to do so would only invite alternative forms of preemployment screening that would be less open, scientific, and controllable.^[76]

The OTA report was commissioned by the House Committee on Education and Labor. It is likely that the request was made as a step toward considering legislation to control integrity testing. A similar OTA report on the polygraph was requested and used as an important basis for the development of the Employee Polygraph Protection Act. In earlier forms, the bill that eventually became that law included written and oral as well as mechanical tests, although references to the written and oral forms were deleted in committee pending further study.^[77] Those inclined toward legal control may find integrity testing to be more elusive than polygraph testing. For one thing, the OTA report representing the further study on integrity testing desired by Congress is not nearly so decisively negative as the earlier report concerning the polygraph. Thus Rep. Pat Williams, the Democrat of Montana who requested the OTA report, indicated that he would not sponsor legislation until still more research has been completed. Moreover, efforts to establish legal control of integrity tests might encounter problems in stating exactly what they are. This is relatively straightforward with the polygraph because such a test may be unambiguously defined in terms of the use of a machine that measures physiological responses. To define a written or oral integrity test is much more difficult.^[78] It could scarcely be by label, because publishers could (and often already do) call them "inventories" or "surveys" rather than "tests." Nor would it seem prudent or practical to forbid all employer inquiries into applicants' theft, vandalism, absenteeism, or other problematic behavior in previous jobs. Therefore, potential legislation controlling integrity tests might prove to be considerably more

complicated to write and to enforce than was the case with polygraph and other mechanical tests.^[79]

These issues aside, it is nevertheless possible to criticize the sheer logic that grounds the concept, design, and scoring of honesty tests. In the first place, these tests are intended to measure "honesty," but a single character trait corresponding to that designation may not exist. In a classic study in social psychology, Hugh Hartshorne and Mark May conducted experimental tests of lying, cheating, and stealing among 11,000 children aged 8 to 16 and found little evidence that these behaviors clustered in the same individuals. Someone who might lie is not particularly disposed to steal, a child who might steal could not therefore be expected to cheat, and those who might cheat on one kind of test might not cheat on another. Their conclusion:

Neither deceit nor its opposite, "honesty," are unified character traits, but rather specific functions of life situations. . . . There is no evidence for supposing that children who are more likely to resort to deceptive methods than others would not use honorable methods with equal satisfaction if the situation in which dishonesty is practiced were sufficiently controlled by those who are responsible for their behavior.^[80]

Hartshorne and May's conclusions are not unequivocally accepted. J. Philippe Rushton argues forcefully that their own data, when examined from a different statistical perspective, suggest that lying, cheating, and stealing do tend to cluster (together with selfishness and a few other qualities) in certain people. He concludes that a general personality trait of morality does exist and is more characteristic of some individuals than others.^[81] The issue is not finally resolved, but no one seems to question that one important determinant of whether people will behave honestly or not is the specific situation in which they are placed. This would imply that employers should pay at least as much attention to creating conditions in the workplace that discourage stealing and other undesirable activities and to rewarding honest behavior as they do to tests that claim to categorize some people as honest and others as dishonest in any and all situations.

Again, questions on integrity tests that pertain to past activities are based on the assumption that people who have stolen from employers or engaged in other dishonest behavior in the past are likely to do so again in the future. This is probably correct (within the limits of the situational considerations mentioned above), and about 25 percent of those who take integrity tests do admit to delinquencies of various kinds.^[82] But, of course, it does not follow that only those who admit to past misdeeds have performed them. Surely the group of those who deny previous dishonest behavior contains some wrongdoers who perpetuate their dishonesty by concealing it on the test. These people would be at least as likely to be dishonest employees as those who made admissions on the test. Following the convention of considering a positive result to be an indication of dishonesty and a negative result to be no such indication, these questions are susceptible to false negative results. That is, they may well identify certain persons as not being dishonest who in fact are.

Certainly it is a primary purpose of the attitude-probing questions to smoke out those false negatives, but again there are logical problems. It is not always possible to know precisely how the tests are scored because testing companies often protect this information as proprietary.^[83] This is another factor that complicates research on the tests. Nevertheless, David Lykken reports that highest marks on attitudinal questions go to those who favor harsh punishment for infractions (e.g., that an employee who steals a few dollars each week should be fired) and who assume a high degree of honesty in everyone (e.g., a belief that very few people cheat on their income tax). "The rationale for such scoring," he writes "is that a thief will be unlikely to recommend harsh punishment for acts he might himself commit and he will probably contend that most people are as dishonest as he is."^[84] True enough, but this net catches many more than moral degenerates because, as Lykken points out, "the logic does not work in both directions."^[85] Even if all thieves oppose harsh punishment, it is likely that many who are not thieves also oppose it. Even if all scoundrels believe that people in general are not much good, many who are not themselves scoundrels may share that dark assessment of human nature. Therefore, these questions

have the reverse flaw of those that inquire into past wrongdoing. Now the problem is false positive results: attitudinal questions are likely to label as dishonest a number of people who in fact are honest.

Given these difficulties, the most effective strategy when confronting an honesty test may be to throw honesty to the winds—to learn the kinds of considerations that go into the scoring and to answer the questions accordingly, without regard for truth. One experiment with college students found that a group told to fake the most honest results they could on an integrity test scored a full standard deviation higher on attitudinal questions and half a standard deviation higher on admissions of past misdeeds than a group that was instructed to answer all questions completely honestly.^[86] As for the alternative: "If the questions are to be answered honestly, then what is required [to score well on the test] is a punitive, authoritarian personality combined with a worldview like that of the three monkeys who hear-no-evil, see-no-evil, speak-no-evil."^[87] Or, as I was told in earthier language by the proprietor of a fast-food chain in Texas, "We used written honesty tests for a while, but quit when we found that the people we were hiring were just too weird ."

A strategy of beating the test by giving the preferred answers may not always work, because on occasion the rationale behind preferring one answer to another may not be possible to perceive. It may, in fact, not exist. Consider the following two questions, from a Stanton honesty test:

1. Do you think people who steal do it because they always have?

2. Do you agree with this: once a thief . . . always a thief?

According to scoring expectations, honest people will answer "yes" to the first of these questions and "no" to the second.^[88] It is possible, however, that many people—some of them honest and true—may be able to discern little, if any, difference between the questions. While this scoring strategy might foil efforts to beat the test, it is scarcely conducive to confidence in its validity.

To end on an ironic note, honesty tests are not always entirely honest themselves.^[89] In some tests, the questions about previous

wrongdoings are labeled optional, and the directions state that they may be left blank if the subject does not wish to answer. But the instructions to the test administrator advise that anyone who does not answer them should be viewed as probably having something to hide. Again, interviewers may be encouraged to lure test subjects into making damning revelations. In one test, a section to be completed by interview includes a question about the applicant's attendance in school. The instructions to the interviewer, stressing that school attendance record is a highly significant indicator of the likelihood of success at work, recommend minimizing its importance by posing the issue with a wink and a smile, such as, "John, when I was in school I often found that going to class was a waste of time. How about you?"