Experts say voice stress analysis no more accurate than coin flip
For at least three decades, the discord about computerized voice stress analysis has grown louder.
Universities have conducted studies debunking the theories of voice stress as a measure of a person’s deception. Medical journals have published stories stating the same.
The Pentagon ceased using it on terrorism suspects.
Ironically, its widespread use among law enforcement agencies has continued.
The Hernando County Sheriff’s Office has relied on it for more than 10 years, said Deputy Wendy McGinnis, an agency spokeswoman.
Col. Mike Maurer, the sheriff’s office’s second-in-command, emphasized the CVSA is merely one tool the department uses to detect whether a person is being truthful.
“Generally speaking, when questioning reveals inconsistencies in a person’s interview, their body language, or when that person has case specific information that no one else would know unless they were there, whether during a truth verification examination or otherwise, the detective is going to explore the interviewees’ inconsistencies,” Maurer said. “The idea that we depend solely, or even primarily, on the truth verification results during a criminal investigation is jumping to an incorrect conclusion.”
The sheriff’s office has two computers with the CVSA software — each was purchased for about $11,000. There are six certified CVSA operators employed with the agency.
Sgt. Phil Lakin, while a detective investigating the 2006 slaying of Pat and Evelyn DePalma, was assisted in the interrogation of the first suspect by now-retired Detective John Cameron Sr., who was a certified CVSA operator.
Cameron told the first suspect, 18-year-old David Alexander Bostick, a distant nephew of the slain couple, the machine was “90 to 98 percent accurate.”
The test took 45 minutes — an extended amount of time because Cameron’s first laptop was malfunctioning and he had to start over, according to the video footage of the interrogation.
After the test, Bostick yawned, stretched and told the detective it “wasn’t bad at all.” He was eager to know how well he did.
Cameron and Lakin, who returned to the interrogation room, showed him he lied on two questions — including whether he knew who killed his great aunt and uncle.
The interrogation continued through the night and into the following morning. It ended only after Bostick gave a forced confession. He would later be exonerated and another suspect, connected to the crime through DNA evidence, was eventually convicted and sentenced to life in prison.
The sheriff’s office answered no questions about Bostick or whether his CVSA results added to the detectives’ suspicions or whether it was used as an interrogation ploy. The tone of the interview changed dramatically after Bostick was told of his failed test.
“This particular case has an order to expunge records in effect for one of the individuals listed in the report,” said McGinnis. “For us to comment on any specific facts would be inappropriate and against the court order to expunge.”
CVSA critics galore
“Validity of voice stress measures was poor,” was the conclusion of one study 27 years ago from Psychological Medical, a prominent medical journal.
“Accuracy was not significantly greater than chance for the CVSA,” wrote one researcher with the Department of Defense Polygraph Institute.
Even the slightest proof of accuracy was met with skepticism from the Western Journal of Speech Communication.
“Only one subgroup showed a detection rate significantly better than chance,” according to the study, “and it did so by the thinnest of margins. Use of questionable statistical methods in this study suggests the modest positive findings would not be replicated in other research.”
ABC News, the University of Florida and Georgetown University all have done stories that chipped away at the validity of the CVSA.
“The use of truth verification as an investigative tool, when used properly and in conjunction with other tools and evidence, can be an important element of any investigation,” McGinnis said.
One local prosecutor and a former prosecutor are among the throngs of outspoken critics.
“There is less science and less research behind it and that makes it suspect,” said former Assistant State Attorney Don Barbee.
Pete Magrino, who prosecuted the DePalma case and criticized the sheriff’s office for arresting Bostick, called the longer-established polygraph a “great interviewing tool” and said the CVSA “is nothing close to it.”
Al Joseph, a former detective with the Rochester, N.Y., Police Department and author of “We Get Confessions,” said the CVSA was used on Susan Smith, who was convicted in 1995 of murdering her two sons after initially telling police they were abducted during a carjacking.
Some have pointed to that case as proof that CVSA was a useful tool toward her eventual confession.
“I just watched her on television and I could tell she was lying,” said Joseph, who said he is “dead set” against both CVSA and the polygraph. He added neither is admissible in court.
He conceded, however, that a polygraph feels more ominous, which could add a “little more pressure to a suspect.”
Either way, he said, they are “mostly props.”
Little value or no value?
The sheriff’s office uses the CVSA during the hiring process. Prospective deputies are administered the test.
McGinnis said job candidates are not turned away solely on CVSA results.
“The decision to hire an individual is based on the totality of the circumstances and not exclusively on the results of a CVSA exam,” she said. “The CVSAs are used as a tool in our hiring process just as it is used during an investigation.”
McGinnis cited the website for the National Institute for Truth Verification, the manufacturer and “sole source for the patented” CVSA.
She said the site includes stories about “numerous cases that have been solved” using the device.
Many self-described old-school police investigators remain steadfastly opposed to the use of the CVSA.
“There has never been a study produced by CVSA people that has proven its accuracy,” said Leonard Bierman, a South Florida polygrapher. “There’s no value to it at all.”
CVSA is designed to detect “microtremors in the voice,” which supposedly vibrate more rapidly when someone is lying.
Bierman said a recent study by the University of Florida debunks even the existence of such microtremors.
“The substance of it is, you’re better off flipping a coin,” he said.
