Debate flares over effectiveness of lie detectors
Some researchers aren't confident that voice-based lie detectors work
By Elizabeth Gibson
The Columbus Dispatch
GROVEPORT, Ohio — When people sit in front of a lie detector, they expect it to detect lies.
But some researchers aren't confident that the voice-based lie detectors common in many Ohio police agencies are accurate.
Dektor Corp., the Lansdale, Pa., manufacturer of one voice-stress analyzer, says Groveport police insulted its services and then bought what it called an inferior system from another company, the National Institute for Truth Verification, based in West Palm Beach, Fla.
Some researchers say they don't have much faith in either device. Studies, most of which focus on NITV, have found that the machines are little better than a coin toss.
Some law-enforcement agencies have backed away from voice-stress detectors, fearing false confessions or lawsuits. But supporters say academics have unfairly tarnished the technology's reputation.
Most people are familiar with polygraphs, which measure physiological responses such as breathing, pulse, blood pressure, body temperature or skin conductivity.
A voice-stress analyzer uses a microphone hooked to a computer to pick up on stress in a suspect's voice.
"They do not work," said Mitchell Sommers, a psychology professor at Washington University in St. Louis, who studied voice-stress devices for the U.S. Department of Defense.
"I'd love for them to work and I think they have the potential to work someday, but for the first year (of research) I just went around with my mouth open, thinking, 'Why on Earth are you using this?' "
Dektor doesn't keep tabs on devices sold, but NITV estimates that about 1,800 agencies in the U.S. use its machines. A few state and federal agencies, meanwhile, have banned the use of anything but a polygraph.
Similar questions have been asked about polygraphs, which is why the tests generally aren't admissible as court evidence in 49 states.
Arthur Herring III, president of Dektor, blamed polygraphists for lobbying lawmakers to stick with the polygraph.
Critics and supporters each find fault with the other side's research.
Voice-stress proponents point to legitimate rape and murder confessions. Critics cite a 1998 case in California in which a boy confessed to a fatal stabbing he didn't commit after being confronted with voice-stress analyzer results.
The basic argument is that voice-stress detectors might detect stress but that the technology has yet to separate the stress of a liar from any other sort of stress, said Kelly R. Damphousse, associate dean of the College of Arts and Sciences at the University of Oklahoma in Norman.
In Damphousse's study, interrogators used NITV devices to ask people who had been arrested whether they were on drugs and then verified the results with urine tests. The accuracy was no better than chance, he said.
But the mere presence of a lie detector makes people more honest, according to Damphousse's study. If police say they have scientific proof that a man is lying, he's more likely to confess.
But Jim Chapman, a regional director for the National Association of Computer Voice Stress Analysts, said researchers are testing fake lies using the wrong procedures in laboratory settings where there's no real stress. He said it has discredited a technology he's used successfully for 30 years.
Groveport, meanwhile, is happy with its choice.
"When officers are telling me they are happy with the equipment I have now, I support that," Police Chief Ralph Portier said.
Before Portier became chief, the village sent three officers to Philadelphia in 2008 for seven days of training on a new Dektor device.
Herring failed them and offered retraining for more money, but the village traded in the device for an NITV machine with free training.
Portier said it works fine.
Whitehall Police Chief Richard Zitzke said he considers voice-stress analysis a helpful investigative tool but not the final word.
The Columbus Police Division uses polygraphists. The Franklin County sheriff's office stopped using voice-stress analysis in the 1980s, Chief Deputy Steve Martin said.
Sommers, the psychology professor, said police and manufacturers may really think the products work, but it's a case of seeing what they want to see.
Copyright 2010 The Columbus Dispatch