Attorney Kirk A. Migdal of Akron, Ohio, who last month convinced a judge to allow polygraph results into evidence over objection in the rape trial of New York law student Sahil Sharma, has evidently convinced Cleveland Scene writer Rebecca Meiser of Sharma’s innocence, based on nothing more than polygraph results:
Duke Comes to Akron
Is an overzealous prosecutor hunting an innocent man?
By Rebecca Meiser
Published: June 20, 2007Kirk Migdal is used to clients insisting on their innocence. After 20 years of defending low-level bad guys, he fancies himself immune to their sob stories. It’s how he gets through the days.
“I can count on one hand the number of cases I’ve had where I really believe the guy’s been innocent,” he says.
But despite his best efforts, once in a while a case lodges in him like a kidney stone. “It’s easy to defend the guilty people,” Migdal says. “It’s the innocent ones that keep you up at night.”
Consider the sexual battery case against Sahil Sharma. Last summer, the then 25-year-old law student at New York’s Touro College traveled to Cuyahoga Falls for a cousin’s wedding. On the night before the nuptials, Sharma and his young relatives snuck out to a bar for reprieve. His cousin brought a friend, former Kent State student Michelle Sacia.
It was a muggy night, the kind that inspires consumption. At a bar at the Cuyahoga Falls Sheraton, Sharma and Sacia became acquainted, chatting about fashion and city life. But six drinks later, Sacia began to feel woozy. She’d asked earlier that day whether there’d be a place to rest before driving home. Sharma’s cousin said his room would be fine. Around 1 a.m., she asked whether she could lie down. She was let into a suite the cousin was sharing with Sharma.
Two hours later, Sharma opened his hotel room and found Sacia lying on his bed. She was not in the best shape. She had just vomited. He brought her a glass of water.
Afterward, the two continued their conversation. One thing led to another, Sharma claims, and soon their libidos took charge. They were drinking, he admits. “It probably wasn’t the most responsible thing to do.”
Then he fell asleep — only to awake the next morning to the sound of police pounding on his door. Sacia claimed she’d been raped.
Migdal wasn’t anxious to take Sharma’s case. Rape, by nature, is hard to defend. Most are he said/she said battles, soiled further by alcohol. But Sharma kept insisting on his innocence. So Migdal, unimpressed, did what he often does with clients: He took him to see Bill Evans, a polygraph examiner and lawyer.
Evans asked about specific facts. “Was Michelle awake while you were having sex?” “Did you insert your penis in Michelle’s vagina without her knowledge?” The American Polygraph Association asserts that asking questions based more on fact than opinion leads to a 90-percent accuracy rate. Sharma passed.
While the American Polygraph Association — an organization with a documented history of turning a blind eye to fraud by its members (see here and here) — may assert that polygraph “testing” has an accuracy rate on the order of 90%, such notions have been roundly rejected by the scientific community. Indeed, there is broad consensus amongst scientists that polygraphy has no scientific basis at all. Because polygraphy lacks both standardization and control (within the scientific meaning of the word), no meaningful accuracy rate is knowable. The only persons claiming 90th percentile accuracy rates for polygraphic lie detection are those with vested interests in promoting this pseudoscience.
Migdal, now convinced of his client’s story, went to the Summit County prosecutor, hoping to get the charges dropped. But he wasn’t approaching a receptive foe.
Sherri Bevan Walsh has spent much of her career prosecuting sexual assault cases and is known as a women’s rights darling in Summit County, where she’s gotten life sentences for three men who’d sexually abused young children. But some believe her success has bred zealotry.
A year and a half ago, recently recovered DNA evidence cleared inmate Clarence Elkins of raping his niece in 1998. Even Attorney General Jim Petro, a man loath to act on behalf of any convict, believed Elkins’ innocence was irrefutable and wrote a letter to Walsh demanding his release. But Walsh held her ground, claiming that DNA tests weren’t conclusive.
The Plain Dealer called her lack of action “unconscionable.” Defense attorney Jana DeLoach claimed that she and other prosecutors were “more interested in protecting their conviction than protecting the public.”
Only after pressure mounted did Walsh consent to Elkins’ release.
Is Clevelend Scene writer Rebecca Meiser suggesting that polygraph “evidence” should be accorded a probative value similar to that of DNA evidence? While lawyer Migdal claims to be persuaded by polygraph results, a prosecutor would be remiss to allow such unreliable evidence to guide her decisions.
Walsh refused to drop the charges. But Assistant Prosecutor Connie Lewandowski agreed to discuss allowing the polygraph test into evidence. (In Ohio, polygraphs can only be admitted if both sides concede.) Yet there were conditions: The prosecutor wanted to use her own examiner and her own questions.
A few days before Sharma was supposed to fly out for the test, he got a call. The prosecutor had changed her mind. The examiner wanted Sacia also to take the test. And due to the trauma of rape, the tests were considered “unreliable” for victims, says Chief Assistant Prosecutor Brad Gessner. The talks were off.
Polygraph results are unreliable for victims and suspects alike. Polygraph procedure is inherently biased against the truthful: perversely, the more honestly and fully one answers the so-called “control” questions (answers to which are secretly assumed to be less than honest), the more likely one is to wrongly fail. On the other hand, deceptive persons may pass the polygraph using simple countermeasures that polygraphers have no demonstrated ability to detect.
Migdal feared he was now playing witness to a “Duke case in Akron,” referring to the botched prosecution of the Duke University lacrosse team. So he had his client take the polygraph anyway, using the prosecutor’s questions and examiner.
Then he contacted Dr. Louis Rovner, considered among the nation’s foremost polygraph experts, and had him conduct a test as well. Sharma passed both times.
The odds of three experts all screwing up is “about one in a thousand,” says Donald Krapohl, president of the American Polygraph Association.
Donald Krapohl is dead wrong. He is inappropriately applying a simple statistical method to what is, in essence, an unscientific, unstandardized procedure. His simplistic assumption is that given a 90% accuracy rate (again, a rate unsupported by the scientific evidence), the chance of an error on any single “test” is 10%, on any two tests one tenth of that (1%), and on any three tests, one tenth of that (0.1%). Such a conclusion with regard to polygraph outcomes is unsupported by any peer-reviewed research. In fact, there are no peer-reviewed studies on polygraph test-retest reliability.
To Migdal, the evidence was undeniable. But when presented with the results, Walsh again refused to admit them in court. “We never agreed for a polygraph to be a determinant in the case,” Gessner says.
That’s when Judge Judith Hunter took control. In late May, she ruled the tests admissible. “All three polygraphists used the most advanced computerized polygraph machines,” she wrote. “All three individuals independently found that Mr. Sharma was not being deceitful during the examination.”
Judge Hunter was duped to the extent that she believes using “the most advanced computerized polygraph machines” means that polygraph results are reliable. As with astrology, computerization cannot make up for the underlying procedure’s lack of scientific underpinnings. Judge Hunter’s 15-page order may be downloaded here (252 kb PDF).
Migdal was vindicated. Prosecutors were appalled. “Judge Hunter disregarded Ohio law,” Gessner says. His office is appealing the decision.
Meanwhile, Sharma’s trial is set for August. If he loses, he will never be able to practice law, so he tries not to think about such things. “In law school, we were taught that justice prevails,” he says.
But as Walsh learned in the Elkins case, justice in Ohio is far from perfect.
Indeed, to the extent that justice in Ohio is influenced by the results of pseudoscientific polygraph results, it is far from perfect.
So even if it’s 50-50, then the odds of three being right ar 78%?
With 50% accuracy (and hence a 50% error rate, as in a coin toss), the odds of three successive outcomes being erroneous is equal to .50 x .50 x .50 = .125, or one-in-eight.
Taking the statistical accuracy to the power of 3 to determine the chance that the outcome of the polygraphs is correct is simply wrong because reliability and validity are two different concepts altogether.
Reliability is the tendency at which one person will get the same result each time they take the test. So when, the accused gets the same result three times, it sugests high reliability of the polygraph.
Validity is the frequency at which results of the polygraph are correct, or detect the condition that they’re supposed to detect. This case indicates nothing about the polygraphs validity because ground truth is not established.
Take for example, personality tests, which many sales or marketing firms use to screen job applicants because it is believed that individuals who posess extrovertive and creative qualities will perform better than those who are introvertive and linear thinkers due to the nature of sales work. Personality tests are considered very reliable, but not very valid for this purpose. When someone takes the test and scores extrovertive, it is likely that this person will continue to be rated extrovertive in future personality tests because they are very reliable. However, human research studies have shown that people who are rated as extrovertive do not display superior performance at sales type work than the people who are scored to be introvertive. Personality tests have low validity for the purpose of screening applicants for favourable characteristics.
It is wrong to say that passing the polygraph over and over again indicates any more accuracy than simply taking the test once. Getting the same result three times merely suggests that future polygraphs are likely to come to the same conclusion as the first three. The orginal validity does not increase with succesive tests.
An overly strict definition of rape could explain why the accused has continued to pass his polygraph. He openly admits that woman in question had both passed out and vomited in his room before they had sex, if not after or during. Although consent laws vary widely from state to state, many states have accepted that women do not have the capacity to freely consent to sex while they intoxicated or extremly sick, and that any non-consentual sex is rape by definition. Asking questions about the wakeful state of the victim during sex ignores the argument that vomiting is another perfectly clear indicator that consent was not, and could not have been freely given. Unconciousness is not the only indicator that rape has occured.