Every year at least one million Americans undergo a lie detector test, and the number is steadily rising. Polygraph results are cited increasingly, and not just in the courts. Officials from government agencies as well as banks, department stores and fast-food chains are using lie detectors to screen job applicants or uncover theft by employees. The only catch, says David T. Lykken, 52, author of the recently published book A Tremor in the Blood, is that polygraph tests don't work. The innocent will fail them 50 percent of the time. Thousands of people, he says, are being refused employment, fired from their jobs and, in some cases, sent to prison—without having committed any crime. A psychiatry and psychology professor at the University of Minnesota who has studied lie detectors for over two decades, Lykken talked with PEOPLE's Linda Witt in his Minneapolis office about the evils of the polygraph.

Why are you against lie detector tests?

Because there is no such thing as a lie detector. A machine—or test—known as a polygraph picks up your emotional reactions to questions, measuring breathing, sweating responses and blood pressure. The examiner uses this information and other subjective evaluations for a diagnosis of what he thinks is truthful or deceptive.

Is this physical evidence conclusive?

The most any examiner can infer is whether or not one question is more disturbing than another—but not why. About 90 percent of the damaging reports made to employers are based not on physiological reactions but on the examiner's assumptions, or on incriminating confessions made during an interview. This subjectivity is part of the reason why the detectors are accepted as evidence in criminal cases in only about 20 states, and then only when both sides agree in advance.

How does the machine work?

Two soft rubber belts are strapped around you—one around the stomach, the other around the chest. Wires are fastened to the ends of two of your fingers. And a blood pressure cuff is wrapped around your arm.

Is there a specific physiological "symptom" of lying?

Absolutely not.

Can the experience of the test itself induce stress signals on the charts?

Yes. It's easy to make people frightened and angry. But the machine cannot tell if one person is angry, another frightened, or whether one or both are being deceptive. Statistics show tests are heavily biased against the innocent. If you've ever had the experience of denying a false accusation and still feeling guilty, you can understand. Wouldn't your palms sweat if you were suspected of murder? Ironically, the true criminal may be so accustomed to the psychodynamics of lying and denial that he can fool the examiner more easily.

How reliable are the tests?

Half of innocent people fail them. You'd do as well flipping a coin. In particular, people with strong consciences and religious beliefs can be easily made to feel guilt and anxiety.

Who is officially qualified to give lie detector tests?

In most states, anyone who passes the typical six-week polygraph course. Yet these inexperienced, untutored people are asked to make difficult judgments that may be literally matters of life or death. Polygraph expertise is touted as science, yet only about 10 of the thousands of practicing examiners are Ph.D.s in psychology, and few could meet the requirements for any of my basic courses.

Can a person refuse to take a polygraph?

Yes. Legally neither an employer nor the police can force you to take a lie detector test. The problem is that many people may then associate refusal to take the test with actual guilt.

Why do employers use polygraphs?

To solve thefts, mainly. They are also used in evaluating job applicants.

Would a "lie box" have helped the Washington Post deal with its reporter who won her Pulitzer for a made-up story and had been hired on a phony résumé?

Giving her a lie detector test might have led her to confess her misdeeds earlier, but if it didn't produce a confession, the test results would be ambiguous at best.

Have innocent employees been fired after failing polygraphs?

Yes, and in increasing numbers. In one case, a Detroit woman was awarded $100,000 from the Kresge stores. But there are tragedies—I will testify soon for the widow and young son of a highly decorated ex-Marine who killed himself after innocently failing a test.

Has anyone been wrongly convicted after failing a polygraph?

It's too common. Peter Reilly, then an 18-year-old from Canaan, Conn., was convicted of murdering his mother largely because he failed a lie detector test. Peter was persuaded that the polygraph showed he had killed her, even though he had no memory of it. Peter had strong physiological reactions to questions like "Peter, did you hurt your mother?" and "Can you remember stomping on her legs?" His eventual confession was meaningless. His conviction was later overturned because vital evidence had been withheld from the defense.

Is design of the questions a problem?

The Floyd Fay case in Ohio can be used as an example of how dumb they can be. Fay failed a lie detector test and was convicted of murder. But he had volunteered to take the test because he knew he was innocent. Typically, he was asked relevant questions like "Did you do it?" along with control questions like "Is today Tuesday?" Because Fay responded more strongly to the "Did you do it?" questions than to "controls," he failed. He served two years in prison before the real killers were found.

Can you outwit the lie detector machine?

Yes. While in prison Fay read an article of mine that said you could make the polygraph needles jump during the control questions by biting your tongue or rubbing your foot against a nail hidden in your shoes. Fay claims he taught the techniques to 27 prisoners who were in trouble over rules. All had told Fay they were guilty, yet 23 beat the test. Anything that produces tension during a question—even tightening your fanny muscles—will make the needles dance.

Are there other polygraph abuses?

Yes. In many parts of the country rape complainants undergo polygraph tests before they can file charges. I find this particularly distressing. How could such a victim, even while telling the truth, not react violently to the relevant questions?

Why do we believe in polygraphs?

I don't know exactly. The lie detector is almost exclusively an American artifact. Many Europeans have never heard of it. Americans are hooked on the mystique of science and technology—an aura exploited by advocates of the devices. There is nothing scientific about them. We began romanticizing the "lie box" in the '20s and '30s as we became aware of the horrors of the third degree and police brutality. The lie detector seemed clean in comparison to hitting someone with a rubber hose. And as a matter of fact, if that is the alternative, I prefer the polygraph.

Does it have any valid uses?

The Los Angeles police are able to get a 30 to 40 percent confession rate by using the polygraph. If lie detectors help close the books on some of the cases in cities with big crime problems like L.A., I'm all for them. I've seen cops grab a gullible guy fleeing the scene of a crime and wrap a cord from the squad car radio around his wrist and tell him it's a lie detector. I put myself in the cop's shoes. So he's got this guy by ruse—of course, the confession must be verified by other means, but at least he's got him.

Isn't this contradictory—they're good if they can scare the guilty, but evil when they scare the innocent?

I'm sensitive to civil liberties, but a person can make a fetish out of civil liberties and forget the police have a serious, difficult job to do—as long as they don't violate rights.

Should lie detectors be banned?

I'd like to think one could impeach the lie detector simply by unmasking its mystery. The lie detector has no more place in the courts or in business than a psychic or tarot cards.