Ohio Court Admits Lie Detector Tests As Evidence

An anonymous reader writes "Last month, an Ohio court set a new precedent by allowing polygraph test results to be entered as evidence in a criminal trial. Do lie detectors really belong in the court room? AntiPolygraph.org critiques the polygraph evidence from the this precedential case (Ohio v. Sharma)."

No, I don't believe they have a place in court

[HouseArrest420]

Lie detectors don't detect lies, contrary to the device's name. They detect physiologal responses to stress, such as elevated blood pressure, pulse, respiratory changes, and sometimes body temperature. All of these things can be "faked". During the initial questioning process that they use to gauge your bodies responses, simply answer a question you know as being true while you clench your ass checks (change in respirations and pulse), or while you try to mimick pain (thumb tac under big toe concealed in shoe). Most lie detector tests ask the same questions in 2 (or more) cases-1 being the control (graphing your responses), and the 2nd time (trying to anticipate your lie). That way when you really do have to lie, your response seems more natural, thus providing false negatives for the test.

Any test that can possibly provide false results should never be used (IMHO) when the resulting information could possibly deny a man his freedom.

Lie detectors are very unreliable

[slashqwerty]

The most common lie detector, the control question test, takes a set of baseline readings where the suspect is expected to tell the truth on some questions and lie on others. It then compares those results to the questions the examiner is most interested in. These tests have been shown to product accurate results about 65% of the time (that's per person tested).

Professional polygraphers will claim their test works 96% of the time. Those claims are bald-faced lies. Regardless of that we can take a look at what happens if the test really did work 96% of the time.

Some employers have been known to hire polygraphers to identify which employee may have been involved in some inside theft (or similar situation). The employer asks the polygrapher to test 50 employees. The odds that the tests will be correct with all 50 employees is 0.96^50=13%. So there is an 87% chance the test will accuse an innocent person...and that assumes the test is correct 96% of the time. What invariably happens is the polygrapher 'discovers' the culprit after the first few tests, packs up his things, and goes home. He identifies the suspect so quickly because the test is only right 65% of the time. Whether the accuracy is 65% of 96% the test will still point to a suspect even if none of the employees did anything wrong.

Re:Accuracy as against usefulness

[Kandenshi]

There are policing agencies out there who already do similar things. Despite it's absense, they may explain that there's incontrevertible evidence that shows that the suspect is guitly, they just want a confession so that the trial goes faster and with less fuss/humiliation for others/etc...

Turns out that one can get a fairly large number of confessions that way, much like you apparently desire. The problem is, it's not all THAT uncommon for the confessions to be lies. Innocent people will lie and confess to horrible, horrible crimes. And a confession given to a jury is a really really good predictor of them finding the defendent guitly. Even if there's little to no other evidence. People tend to believe confessions, which is sort of confusing since they have to reconcile the idea that "this is a dangerous lunatic with no morals and a willingness to kill" against "this is an honest man, who will condemn himself to jail by giving a confession". Still, they manage it.

Feel free to read a bit more about the subject of false confessions here [psychologytoday.com], on some webnotes for a college class here [72.14.205.104] or even here [innocenceproject.org](this last one is perhaps more likely to cherrypick it's evidence, but what it says appears to be true).

False confessions are a rather worrying thing to me, as once a person confesses, the police have a tendency to cease looking for other potential guilty parties. While it's possible some other person will eventually be found guilty and you get released, it's not really something that The System tries for. Makes 'em look bad if they accidentally put someone in jail and gave 'em a whole bunch of publicity as a convicted rapist.

Let the Knee Jerk responses begin...

[DumbSwede]

In this case it is the DEFENSE offering the lie-detector evidence. Most sexual molestation/battery cases come down to he-said she-said. Lots of innocent people have been convicted under these circumstances. While lie-detectors are not perfect I think in this type of situation they are perhaps appropriate. I would not allow them for prosecution (which is what I think a lot of knee-jerk post here have assumed) and only in cases where the evidence comes down to what is being said by two people, which appears to be what the Judge has decided in this case. While lie-detectors are only about 70% accurate, that is better odds than deciding just on the demeanor of two people in court.

I can sympathize that women are outraged by the high number of men that get off scott-free with these type of charges, but that doesn't alter the fact that it really isn't fair to convict someone on nothing more than an accusation by one person without direct supporting evidence (bruises are not direct evidence). Yes direct evidence is hard to come by in these cases, they are usually executed in private without other witnesses, but I for one would rather see 10 guilty men free than send 1 innocent man to jail.

not a precedent

[delong]

This evidentiary order is not a "precedent". First, it's a mere evidentiary order. Second, the decisions of state district courts are not precedential. They aren't in any way binding on any other court. Third, this is almost certainly error and will almost certainly be reversed on appeal if it isn't harmless error. The federal rules of evidence and the rules of evidence of every state that I know of bars polygraph evidence as unreliable, and has been so held in state appellate courts. THAT is precedential.

Re:Nice, unbiased source.

[Kythe]

Antipolygraph.org makes no pretense as to being "unbiased"--they're an advocacy group. But the information they provide is scrupulously documented and referenced, and it comes from some of the most credentialed scientific sources. They've done their homework.

Even better story

[nbauman]

George W. Maschke, the founder of Antipolygraph, posted a nice statement of why he founded his web site. https://antipolygraph.org/statements/statement-003 .shtml [antipolygraph.org]

The guy sounds like a real straight arrow, super-patriotic American who worked with a Top Secret clearance for U.S. Army Intelligence and with the FBI on the first World Trade Center bombing, and who was particularly valuable because of his fluency in Arabic and Farsi. After doing exempliary work, he applied for a job as FBI special agent, but was rejected and blacklisted elsewhere because a polygraph examiner falsely decided he was lying and rejected him, and the FBI rejected all his appeals.

That's Maschke version, and I'd like to see any response by the FBI or anything to challenge his credibility. I couldn't find anything.

Re:Lie detectors are very unreliable

[gvc]

Does this link [nature.com] not work for you? I can attempt to paraphrase the paper but since it is only two pages long it would speak for itself much better. I was aware of this article and it does show in one particular laboratory experiment that polygraph results likely differ from chance. The article does not conclude that the results are reliable or transferrable to a diagnostic setting. The authors report on a single experiment in which 100 people are re-interviewed for theft cases that have previously been resolved. 50 are innocent (because somebody else confessed) and 50 are guilty (because they confessed). In this experiment, false positive rates of up to 50% and false negative rates of up to 36% were observed (depending on the interpreter of the charts). These findings are better than chance (p .05). The authors conclude, "Hence, we conclude that the validity and reliability of polygraphic interrogation have yet to be established."

Re:Accuracy as against usefulness

[SEMW]

The argument is not well served by taking figures like this from the air. If you care to cite a particular study, we can debate its methodology, statistical power, and freedom from confounds such as selective sampling or lack of blinding to the "true" result.

I didn't realise there was much of an argument to be had. The 70% figure was remembered from a chapter on polygraph testing in a book I read about 5 years ago, not any particular study; if you want to read the details of particular studies, there are a few hundred out there, and Google is your friend (for example this 2003 meta-study [nap.edu]). They all seem to broadly agree that polygraph testing, whilst significantly better than chance, still isn't very good (e.g. the meta-study linked to concluded that a polygraph test regarding a specific incident can discern the truth at "a level greater than chance, yet short of perfection"; not very specific, but I have no particular desire to pay $3 to read the conclusions in more detail). If you are either so convinced that this is incorrect, or desperately wish to pin down one particular specific figure for accuracy, that you wish to "debate [the] methodology, statistical power, and freedom from confounds such as selective sampling or lack of blinding to the 'true' result" for each study in turn, feel free; I personally don't really see much point. That's what meta-studies are for.

Re:Lie detectors are very unreliable

[SuperCharlie]

A long time ago, I worked in a pawn shop. Somebody got a wild hair and decided everyone had to take polys to remain employed. They were looking for previous thefts, nothing specific, just if we had ever taken anything in the past.

I needed the job badly and I was terrified, being only around 21 or so at the time. I never took anything and I failed miserably because when they asked if youve ever stolen anything I was so nervous about setting it off.. right.. I set it off apparently. I got fired, no recourse.

I came to realize polygraph tests are more about seeing what happens when they kick you than getting at the truth.

Re:Accuracy as against usefulness

[gvc]

I can't tell you how to spend your next $3.00 but I suggest that you might get that much value -- entertainment if nothing else -- from reading a primary source rather than relying on Wikipedia paraphrasals

The meta-study found only 57 studies that were carried out with sufficient rigor to be considered. Of those, some but not all showed that under laboratory conditions the polygraph showed better than chance results. The result specifically notes that these laboratory findings likely overestimate (i.e. are an upper bound on) the potential accuracy of polygraphs for investigative or screening purposes.

Re:Weight vs admissibility

[All_One_Mind]

I failed a polygraph when I was telling the truth. I was looking at 14 years in prison, so the pressure was intense and I was nervous as fuck. The end result: The polygraph said I was lying about not shooting some guy I had never met in the face.

I can't even imagine what would've happened if that would've been considered "evidence" admissible in court. I'd probably be in prison right now.