In May 1922, the then–highly competitive Bay Area newspaper industry was mesmerized by a case, and a device, that would influence crime lore for the next century. San Franciscan Henry Wilkens had been accused of hiring a pair of hoodlums to stage the murder of his wife, Anna, during a family trip to the Santa Cruz Mountains.
San Francisco cops proposed hooking up Wilkens to a machine being popularized at the time by Berkeley Police Chief August Vollmer, the father of modern police work, who introduced cops to modern filing, police school, merit-based promotion, crime-mapping and, in 1921, the polygraph.
"So long as the machine was hooked up to the living, breathing subject, it became kind of a Frankenstein's monster, an artificial human given life by the subject to whom it was intimately attached," Ken Alder wrote in his book The Lie Detectors: The History of an American Obsession. "No wonder Wilkens was reluctant to undergo the ordeal. But the police gave him the veiled threat they always give in such cases; they told him that if his 'hands were clean,' he had nothing to fear."
Wilkens emerged from the session deemed to have told the truth in professing his innocence. Soon after, Anna's killer confessed he had been hired by Wilkens. Evidence at trial showed Wilkens had lied about matters other than the murder during his polygraph test, but a jury deadlocked and he went free. Nonetheless, San Francisco police were furious that Wilkens — whom they believed was guilty — tricked the polygraph machine. And the SFPD captain of detectives declared at the next meeting of the International Association of Chiefs of Police that lie detectors "would not be countenanced."
But the department eventually came around, and polygraph examinations have constituted a routine part of checks conducted on potential recruits for as long as the SFPD's current background examination supervisor can remember. And as in the case of Henry Wilkens, it seems as if people are still fooling the machine.
Seven years ago, a San Francisco police recruit named Michelle Alvis found herself at a consultant's office, hooked up to an updated version of the device championed by Vollmer, with monitors measuring her pulse, sweat, and heartbeat.
"I was with her in the waiting room, and she was very nervous going into the test," her ex-husband, Michael, wrote in a 2002 declaration filed in court as part of child custody proceedings. "When she passed, she was very relieved and bragged to me, my mother, and my aunt (and possibly other persons) that it was easy to lie during the polygraph examination and pass the test. She admitted to me that she was compelled to lie and did so successfully on four out of five questions."
Attorney Lidia Stiglich, who represented Michelle Alvis in a recent, unrelated theft case, said it was Michael Alvis who wasn't truthful. "Ms. Alvis never lied on the polygraph nor boasted of doing same," Stiglich said in an e-mail. "The declaration to which you refer is from a hotly contested child custody matter."
Michelle Alvis went on to become notorious in San Francisco, following allegations about her involvement in two incidents. In 2006, she shot a man to death after claiming to have seen him draw a weapon as he was hiding in an apartment attic. She was later charged with stealing nearly $2,000 from a police evidence locker. After a jury deadlocked seven to five in favor of her guilt, prosecutors dropped the charges this past May.
It's impossible to conclude merely from statements in a custody proceeding that Alvis' rocky police career can be traced to a faulty polygraph exam. As her lawyer suggests, declarations in such a case warrant skepticism because angry ex-spouses can be motivated to make strong — and even false — accusations against each other.
But the mere question of whether Alvis managed to trick a lie detector is interesting because it throws light on the fact that the SFPD depends on the device even though scientific experts say polygraphs are all but worthless for job screening, with a significant likelihood of identifying truthful applicants as liars and giving dissemblers a passing grade. "The vast majority of studies show that the polygraph is better than chance, but far from perfect, and far from reliable," said Carnegie Mellon University statistics professor Stephen Feinberg, who led a National Academy of Sciences panel that concluded in a 2003 report that lie detectors were "intrinsically susceptible to producing erroneous results."
Debunking the validity of the polygraph has been a routine exercise among scientists, journalists, and other experts since the Wilkens trial. Polygraph exams are prohibited from being used as evidence at trial in most cases. In 1988, Congress passed a law prohibiting private companies from using polygraphs in employee screening.
In screening, "it has no validity whatsoever," said Drew Richardson, a retired FBI special agent with a Ph.D. in physiology who has testified before Congress about the polygraph. In this context, he said, the interesting question becomes not so much whether Alvis lied, but why she was sent to a polygrapher at city expense in the first place: "I don't think people have a right to a job, but they do have a right to due process, and this is not due process. It's very bad." The SFPD background check polygraph exams are administered by the Burlingame-based company Patrick Coffey and Associates. Coffey dismissed scientific reports questioning the accuracy of lie detectors as overblown and outdated.
Coffey said the Feinberg-led panel probably didn't include scientists with relevant expertise, particularly in the field of psychophysiology, and that polygraph science has advanced since then. Johns Hopkins University researchers have given a thumbs up to a technique in which an individual subject's sweat, pulse, and breathing results are compared to a computer database of known liars and truth-tellers who underwent the same test. "All those studies they looked at are dinosaurs," he said. "Once you came into the computer age, when you have objective scoring, now you're looking at a database that is purely mathematical."
"The claim that we missed something and there's a lot better science now is simply incorrect," Feinberg retorted. "Polygraphers are not scientists. It's not that everything they do is wrong. It's just that most of it is not fully based on science. It's just that they make lots of errors. It's just that they don't know the difference between errors and accurate results."
But Sergeant Martha Juarez of the SFPD's background division says that the polygraph has been helpful in vetting potential cops. "Has it ever been challenged by anybody? I don't think so," she said. "If they've told us the truth about everything, it will come out in the polygraph, so it's been very useful for us. It's been a fact-finder, basically."
She said only a few of the hundreds of polygraph exams the department conducts come back with suspicions that a candidate lied. And "we still do the old-fashioned police work," Juarez said. "It's all about footwork, telephone calls, computer database work. If this person says they are not a drug dealer but the polygraph says they are, we say, 'Let's fact-find.' We go to their neighborhood. We talk to their friends. We check everything."
Regarding Michael Alvis' 2002 declaration about his ex-wife's supposed lack of candor during her exam, Juarez said, "I hope they investigate that person to see if what the husband said is true. If it's bad on us, we want to know."
Ironically, Coffey said that test subjects like Michelle Alvis might be mistaken in believing they have fooled the test. Most polygraph examiners depend on the idea that everyone is a liar, but that some lies are worse than others. They begin by asking "control questions" which are designed to create an irresistible temptation to deceive, causing the subject to sweat, pant, or palpitate in a way that provides a specific basis of comparison for more serious lies down the road.
Alvis "may think that she fooled those control questions," Coffey said. "Whether or not she was responding to control questions, or the examiner wasn't on his best that day, it's hard to say."
This raises the question: Given that the city is already performing thorough, shoe-leather background checks, why not stop spending tax dollars on the discredited product known as the lie detector?