Two weeks ago, besieged by speculation that he had killed former Washington intern Chandra Levy, Rep. Gary Condit, D-Calif., launched a strategic retreat. He granted the police a long interview, reportedly admitting to an affair with Levy. The cops emerged to say he had answered all their questions. Condit gave them a DNA sample and let them search his apartment. Then his lawyer, Abbe Lowell, delivered the coup de grâce: He announced that Condit had passed a private polygraph exam. The cops, the Levy family, and many commentators cried foul, but the speculative bubble around the congressman had burst. The old headlines about his role in Levy's disappearance faded, replaced by talk of serial killers, Web sites, and fresh searches for her body.
Step back for a minute and think about this: A man has attached himself to a machine, uttered the word "no" three times, and derailed an investigation that had become a national frenzy. No information or explanation was necessary. The search for truth used to be understood as a verbal process of questions, answers, and probing contradictions between statements and evidence. But the ascent of technology has nurtured confidence in a more mathematical approach. We think the polygraph can reveal the truth. And it can—but only as part of an enterprise older, larger, and subtler than itself. The verbal and social elements of investigation are just as influential inside the polygraph room as they are outside it. By focusing our attention and faith on the machine, we allow its users—police and suspects alike—to manipulate those elements to their advantage.
In the old days, cops questioned suspects while scrutinizing them for signs of nervousness or evasiveness. A polygraph basically systematizes that scrutiny. It replaces rough observations—sweaty foreheads, blinking eyes, fidgeting fingers—with quantitative data such as pulse and respiratory response. To the extent that these data can be compared among thousands of suspects, and specific elevations of blood pressure or perspiration can be correlated with answers that turn out to be false, the polygraph measures responses more reliably than the unaided human eye does.
But measuring responses is only part of interrogation. The answers won't reveal much unless the questions are well-chosen. In 1991, a CIA polygrapher perfunctorily asked intelligence officer Aldrich Ames, who was secretly on the Soviet payroll, whether Ames was withholding any financial information from the U.S. government. Ames registered a blip on the polygraph but explained that the withheld information was his contemplation of retiring from the CIA and going into business. When Ames retook the test, he passed it easily. CIA polygraphers later indicated they would have constructed the test differently had they been given the latest background file on Ames, which flagged two recent cash purchases as suspicious. The machine didn't fail. Its human masters did.
In 1998, Barry Colvert, the polygrapher who would later examine Condit, testified that he had grilled former Teamsters President Ron Carey on the "primary questions" in a money-laundering scandal and that Carey had passed with flying colors. But according to the Associated Press, "Colvert's questions were limited to charges that about $735,000 was donated to the Teamsters to generate contributions to Carey's re-election campaign. There was no indication he asked Carey about allegations that other labor leaders funneled prohibited donations to his campaign." This year, Carey was indicted on charges that, under oath, he had falsely denied knowledge of those prohibited donations. Polygraphy, like polling, is an art of framing questions concealed in a science of measuring answers.
It's also a social process disguised as a mechanical one. Critics dismiss the lie detector as an empty, pseudo-omniscient box used by cops to bully suspects into confessions. ("The machine's having a problem with your answer about where you got the TV set, Jerry.") But you don't have to think the box is empty to recognize that it benefits from a sobering environment. "I'm breaking [the suspect] down psychologically," polygrapher Bill Majeski explained this week to the New York Daily News. "If I'm dealing with someone who is desensitized to lying, I need time to establish a fear in them that they'll be detected." That fear depends in large part on where, how, and by whom the test is administered. The box works in a room. Take it out of the room, and it doesn't work so well.
To measure the importance of this factor, look at the data posted by the American Polygraph Association, which represents the nation's top polygraph examiners. The APA says the machine's average accuracy in real-life "field examinations," where livelihoods and jail time are at stake, is 98 percent. But in "laboratory simulations"—in which, for example, college students take the test after being instructed to "steal" envelopes stuffed with a few dollars—the figure falls to 80 percent. "The accuracy is assumed to be lower [in simulations] because the motivation to deceive is assumed to be less," explains Dr. Frank Horvath, who chairs the APA's research center. "It's commonly believed that the fear of detection is greater in the real world than it is in the laboratory."
By taking a polygraph under the auspices of his lawyer rather than the police, Condit essentially moved his test from the field to the lab. It was a simulation, a rehearsal. He knew that if he flunked, no one but himself, Lowell, and Colvert would hear of it. At his press conference, Lowell bragged about Colvert's FBI career. But now that Colvert is in the private sector, he reports only to his client. The polygraph he gave Ron Carey was arranged by Carey's legal team. When Colvert testified that Carey had passed it, he was forced to concede that if Carey had failed, the results would never have been disclosed. Horvath says people who take polygraphs arranged by their lawyers are twice as likely to fail as are people who take polygraphs arranged by cops. But surely that's because those who think they might fail the test screen themselves out of the latter sample. The probability that Condit would flunk his private polygraph and that the cops or the public would find out about it was zero. He had no fear of detection.
He also controlled the questions. When Levy family attorney Billy Martin first proposed a polygraph, he said it should include questions about the origin and exact nature of Condit's relationship with Levy. The police agreed. They wanted additional questions about chronological inconsistencies in Condit's recollections and about whether he had introduced Levy to anyone else who might know something about her disappearance. They pressed Lowell for broad latitude. By circumventing them and paying for the polygraph himself, Lowell got to narrow its scope. At his press conference, he asserted that according to Colvert, Condit's answers showed "a probability of deception of less than [one-]100th of one percent to the only questions that mattered. These were, first: Did the congressman have anything at all to do with the disappearance of Ms. Levy? Second, did he harm her or cause anyone else to harm her in any way? And third, does he know where she can be located? … Any other questions might be of interest to tabloid journalists but would not be pertinent to the issues involving Ms. Levy now."
A week later, the cops are complaining that Colvert's data and analysis haven't been presented to them in a scientifically reviewable format. They miss the point. The polygraph wasn't aimed at them, and its sleight of hand wasn't in the numbers. The target was the media, and the game was in the questions. The cops had no evidence that Condit was involved in Levy's disappearance. If he was guilty, the only way to unravel the case was to tug at the thread they knew about—the affair—and extract one disclosure after another. If he was innocent, the media were going to pull this thread anyway, along with every other tip about his sex life, until his career and family were destroyed. Lowell's best way out was to catapult the story forward to its ostensible bottom line—did Condit kill Levy?—thereby rendering the intermediate questions moot. Stripped of their excuse for investigating Condit's affairs, reporters would have to shift attention to other aspects of the case. They wanted to keep the mystery of the relationship and the mystery of the disappearance intertwined. Condit wanted to separate the two issues. The polygraph did the job.
The police, the Levy family, and the piranhas of the press corps have only themselves to blame. Frustrated with Condit's lies and evasions, the Levys and their lawyer, Martin, first demanded a polygraph early last week, claiming that it would "end this darkness." Washington, D.C., Police Chief Charles Ramsey said Condit should submit to a lie detector because the Levys "have a right to know the truth." The Daily News fed the frenzy with articles hyping the machine's deadly accuracy. A Gallup poll showed that 83 percent of respondents thought Condit should take the test. So he did. The metaphysical certainty that his accusers had attributed to the lie detector became his best weapon against them.
Now Martin, the cops, and the Daily News are trying to dig themselves out of the myth they dug themselves into. They're trying to explain that the power of the polygraph isn't just in the box; it's in the crafting of the questions and the context of the conversation. The polygraph "was our idea," Martin complained last weekend. Condit "picked up the idea" and "snuck off and took a polygraph on his terms," the lawyer charged. But the public has no idea why the "terms" of the test matter. With Martin's encouragement, the legend of the polygraph's scientific precision has overwhelmed the verbal and social nuances that go into it. That's why Lowell was able to brush aside inquiries about the questions in Condit's test by pointing to Colvert's "unassailable credentials."
Colvert's critics note that his résumé says he polygraphed Ames. Both Ames and Carey, they observe, turned out to be liars. But Colvert wasn't the examiner who cleared Ames. He didn't test Ames until the spy had been caught and was spilling his secrets to the FBI. What happened between the Ames case and the Carey case? Colvert left the FBI and began hiring himself out. Same examiner, different clients, different questions, different settings, different results. So much for "credentials." It would be easy to dismiss Colvert as corrupt or incompetent. It's harder to face the truth: He isn't. Honesty and expertise aren't enough. It matters who's paying for the exam, who's drawing the subject boundaries, and who's writing the questions. Polygraphs don't reveal truth. People with polygraphs reveal truth. Or, if we're not careful, they bury it.