Polygraph Tests--completely unreliable, or just not very reliable?

[Frylock]

Are polygraph tests non-admissible in US courts because they're not reliable at all, or merely because they're not reliable enough?

-FrL-

[mks57]

I'm sure some people would say that they are not reliable at all, but I think it's that they are not reliable enough. With training, people can manipulate the results. Some people don't respond as the theory predicts, and it isn't a lie detector, it measures physiological reactions to questions. There is the old joke about the delusional mental patient that flunks the lie detector test when asked if he is Napoleon Bonaparte and he says no.

[Campion]

The test for the admissibility of scientific evidence is, in many US jurisdictions, governed by a case called Daubert (which some people pronounce Dowbert and some pronounce Dowbear).

Prior to Daubert, scientific evidence could be admitted in court if it was "generally accepted" in the relevant field, which meant that each side would proffer a series of experts, each opining that the evidence either is or is not accepted, and even arguing over what the relevant field is. The leading case is Frye, which held that a polygraph was not admissible. Frye is still used today in some jurisdictions (including mine, California) as the relevant standard to judge the admissibility of scientific evidence. You can see, however, what it does to polygraphs: Frye held that polygraphs have "not yet gained such standing and scientific recognition among physiological and psychological authorities as would justify the courts in admitting expert testimony deduced from the discovery, development, and experiments thus far made." Consequently, under Frye, polygraphs were out as not generally accepted.

Under Daubert, however, the test for admissibility changed, in light of Federal Rule of Evidence 702, which states:

If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise, if (1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.

This article discusses Daubert's factors, which include (in a non-exhaustive list):

• Has the theory been tested?
• Has the theory been subjected to peer review?
• What's the error rate?
• Is the expert respected in the community?
• Can the jury understand it?

As the Daubert court explained,

The inquiry envisioned by Rule 702 is, we emphasize, a flexible one. [n.12] Its overarching subject is the scientific validity--and thus the evidentiary relevance and reliability--of the principles that underlie a proposed submission. The focus, of course, must be solely on principles and methodology, not on the conclusions that they generate.

With that as the background, the question of polygraphs comes in. This page, which should not be considered reliable, makes the facile statement that polygraphs are admissible if the judge admits them. Duh. The inquiry, instead, is whether courts generally admit polygraph results under Daubert. The answer is "sometimes."

This article links to some of the relevant cases; it seems that there is no outright bar on the admission of polygraph evidence, but that it is not generally favored. See, e.g., US v. Cordoba, holding that Daubert eliminated a "per se" rule of polygraph inadmissibility.

In Cordoba, the defendant took a polygraph (administered by the defense, not the prosecution), and sought to have the polygraph admitted to rehabilitate his credibility if the government attacked his credibility. Because the trial court believed that polygraph evidence was always inadmissible, the trial court excluded the polygraph. The Ninth Circuit ruled that polygraph evidence, like all scientific evidence, must instead be evaluated under Daubert, so it sent the case back to the trial court for an evaluation.

This is a Department of Justice pamphlet on the use of polygraphs; I find the alternate arguments to exclude polygraphs interesting and, in at least one case, relatively persuasive. If scientific evidence must be helpful to the jury, the article notes, one can argue against the admission of polygraph evidence because the jury doesn't need any help judging the credibility of a witness.

Originally Posted by Frylock

Are polygraph tests non-admissible in US courts because they're not reliable at all, or merely because they're not reliable enough?

With all that as background, the answer is that in Frye jurisdictions, they're generally out because they're not generally accepted in the relevant scientific community.

In Daubert jurisdictions, while polygraphs are not automatically inadmissible, they are generally not admitted for a variety of reasons. When you do see a polygraph admitted, it's generally because both sides have agreed to its admission.

Thus, the question really isn't solely one of reliability: it's verifiability, acceptance in the scientific community, and all the other Daubert factors.

[GiantRat]

I've always wondered if the admissibability of polys didn't really come down to the potential devastation of the judicial system. If we were to accept polygraphs as scientifically reliable (to an extent that they were generally accepted as being fairly free from error), we'd no longer have a need for juries and lawyers. A quick polygraph test and forwarding to the judge for sentencing would certainly speed up the legal process, but all those lawyers would have to start doing (gasp!) civil work!

[yabob]

Originally Posted by GiantRat

I've always wondered if the admissibability of polys didn't really come down to the potential devastation of the judicial system. If we were to accept polygraphs as scientifically reliable (to an extent that they were generally accepted as being fairly free from error), we'd no longer have a need for juries and lawyers. A quick polygraph test and forwarding to the judge for sentencing would certainly speed up the legal process, but all those lawyers would have to start doing (gasp!) civil work!

A reliable lie detector might devastate the court system, but not in the way that you are thinking. It's hard to accept that any device would do anything but tell us whether somebody was conciously lying or not. I think we would find a disturbing amount of downright contradictory testimony that witnesses believed to be the truth, as confirmed by the machine. We would have some very fundamental questions raised concerning the value of eyewitness testimony, and the rate of self-delusion in the general population. Rather than sorting out who was lying, the legal system would wind up sorting out who was mistaken in their observations and/or deluding themselves.

[yabob]

Note also that I am speaking hypothetically about a reliable lie detector. I don't believe the polygraph comes anyplace close.

[Campion]

Originally Posted by GiantRat

I've always wondered if the admissibability of polys didn't really come down to the potential devastation of the judicial system. If we were to accept polygraphs as scientifically reliable (to an extent that they were generally accepted as being fairly free from error), we'd no longer have a need for juries and lawyers. A quick polygraph test and forwarding to the judge for sentencing would certainly speed up the legal process, but all those lawyers would have to start doing (gasp!) civil work!

Several problems with your notion, not least of which is that it puts the cart before the horse. You suggest that the reason that polygraphs may not be admitted is so that the criminal justice system can protect its survival. But, of course, polygraphs aren't accepted in the relevant scientific community, aren't verifiable, aren't reliable. But assuming that somehow, magically, they became so...

Polygraphs can be compared to DNA evidence: neither a polygraph result nor DNA evidence would be sufficient standing alone to support a conviction, but (assuming polygraphs were scientifically valid/accepted/reliable, etc.) either one is good evidence that the defendant is guilty. (And note that DNA evidence, which, in my opinion, is a much stronger form of evidence than a [reliable] polygraph, is routinely admitted at trial.)

The problem is that crimes are generally established by laws, and the laws establishing a crime lay out "elements" of each crime. The prosecution bears the burden of proving by a preponderance of the evidence that each element of the crime has been established. It isn't enough, for example, for the prosecution in a murder case to show that the defendant wielded the murder weapon. The prosecution must also show the appropriate level of intent. So neither DNA nor a polygraph would go far enough.

Then there's the whole "jury as trier of fact" thing and "fifth amendment" thing and "seventh amendment" thing. In short, I don't think it's fair to say that the reason the scientific community hasn't accepted polygraphs (the way they have DNA, for example) is to protect the survival of judges and lawyers who do criminal work.

[Constipated Mathematician]

A polygraph can, I believe, help as an indicator to the truth or deception. So it's not completely unreliable.

However, different people react to stressful situations in different ways. Therefore, if you can't rely on it 100%, it's best to leave it out of the legal system.

I knew a person who had to take a polygraph for their job. Now, this is just what I was told, so take it for what it's worth. They felt that if something wasn't the business of their employer, they didn't feel guilty about lying about it. This person passed the polygraph. That single story has always made me believe that depending on the mindset of the person in the chair, you may or may not get the truth.

It's not like DNA. Until it is, leave it out of court.

[Guinastasia]

Originally Posted by Constipated Mathematician

I knew a person who had to take a polygraph for their job. Now, this is just what I was told, so take it for what it's worth. They felt that if something wasn't the business of their employer, they didn't feel guilty about lying about it. This person passed the polygraph.

Really? Because I believe requiring polygraph tests for employment is illegal.

[Scuba_Ben]

Polygraph tests are permitted in limited circumstances, mostly involving national security or public safety positions.

[A.R. Cane]

I took a pre-employment PG in 1981. In the interview, before the test, I admitted to pot use, but told the interviewer that it had been several years since my last use, when in fact it had only been about a month. During the test he asked about use in the last week, month, etc., to which I replied 'no'. I was hired. There were a group of us there for training, staying at the same motel. One other guy smoked weed daily and made no secret of it. He took the PG right after I did and was also hired. There were plenty of applicants, so I seriously doubt that any failures were intentionally overlooked.
I really don't see the logic in taking a PG if you''re accused of a crime. Even if it indicates that you're being truthful, it isn't likely to help you very much. At best, it might divert police attention away from you, but if they already believe that you are a stong enough suspect, to ask you to take a PG, it's likely that they will still suspect you. If they later charge you, the PG can't be used in your defense w/o the prosecuter's approval, which is extremely unlikely.
I can understand an innocent person wanting to prove their truthfulness, but if you think it through, a PG isn't likely to be much help.

[t-bonham@scc.net]

Originally Posted by Frylock

Are polygraph tests non-admissible in US courts because they're not reliable at all, or merely because they're not reliable enough?

You've got it a bit backwards:
They are non-admissible because they have not been proven to be reliable and repeatable.

[DocCathode]

I spent a few years at a company that sold medical sensors and the related software. I've been hooked up to photoplathysmagraphs, pneumatrodes, gsr sensors, and thermistors.

As indicators of stress, they work great. As indicators of truthfullness, they suck. Most of the folklore methods to beat lie detectors will work. Clench and unclench your buttocks. Just convince yourself the machine is an unreliable piece of crap. Switch between vivid imaginings of eating a live cockroach and having your favorite celebrity bathe you with their tongue.

I can tell the examiner that I'm Henrietta R Hippo, a widowed black woman, that the year is 2065, and that I'm 70- and not have the machine detect any stress. It will detect great stress when I tell how my son died in the war with Canada.

My sister's ex failed a test when the machine indicated he had used marijuana. The man has many failings, but he was only ever near drugs when confiscating them. A friend had mentioned failing the same question. This made the man nervous, leading to a false positive.

[Diceman]

Doc, can you confirm or deny something I've heard about polygraphs? They say that sociopaths can pass lie detector tests easily, because they don't feel guilty when they lie. Is this true?

[Triskadecamus]

Long ago, I was peripherally involved in a military situation involving stolen stuff. The stuff was fairly important. They gave everyone lie detector tests. The guy doing the test provided the names of the three guys who were involved in the theft of the stuff. Two of of them were not on the base at the time the stuff was stolen. (Nor in one case was he even in the same country.) The third was, but had only very limited access to the area where the stuff was, and would have required the help of at least one , out of about five, people who had passed the test.

The matter was never brought up again.

Tris

[DocCathode]

Diceman The company I worked at was interested in polygraphs only as measures of stress levels and symptoms of health problems. The boss wasn't interested in detecting lies, but reducing stress, and demonstrating correct sensor placement, conductor gel use, etc.

[marshmallow]

Originally Posted by yabob

It's hard to accept that any device would do anything but tell us whether somebody was conciously lying or not.

A minor aside, but a brain scan can detect if someone is consciously lying by detecting activation in certain parts of the brain associated with deception or imagination as opposed to recall of facts or experiences. Of course, it's a rather cumbersome and not at all practical...but hey, maybe someday if you suspect your hubby is lying about taking out the trash you can simply pass a Lie Wand(TM) over the scumbag and see who gets to sleep on the couch.

[DrDeth]

There's a thread right here on Polygraphs:
http://boards.straightdope.com/sdmb/...d.php?t=379460

Here's what I said (and I have a lot of experience with Polygraph operators, and have talked with them a lot "off the record")

IMHO a "good" polygraph operator decides whether or not you are lying based upon mostly various intangibles he observes, adding his "squiggles" then telling you that the "machine says you are lying". Then since experienced investigators are quite good at determining truth (with or without a polygraph), many dudes collapse and confess, since "the machine can't be wrong".

However, many dudes who claim they are alien abductees can and do "pass" polygraph exams, and since I hope we all know that they have NOT been "abducted by aliens"- "the machine" is hardly infallible.

Let me go on to say that Polygraph operators (and any expereinced investigator) are pretty good at noting the rather clear "signs of guilt" most of humankind show. Where the problems lies are:

1. Some few dudes truely believe something that isn't really true. Thus, they will appear to be truthful

2. Some even rarer dudes are sociopaths and have no guilt for their crimes. Thus, again, they won't show signs of guilt even if guilty.

3. The signs are rather "in the eyes of the beholder" and even a great investigator can convince himself he is seeing signs of guilt where there are none, especially when said investigator is personally convinced of your guilt to start out with.

4. The most common is where you didn't really commit the crime, but "feel guilty about it". A Parent may be questioned about their child's death- that parent will often feel "it's their fault" (for letting the child go unsupervised, or the parent may thing they weren't "being a good mother"), and thus will exhibit signs of guilt, even those not guilty of the crime. This is why polygraphs are a BAD thing, as you can get a dude who have convinced himself he is responsible (even though he commited no crime) and then the Polygraph operator sez "the machine says you are lying- confess, you'll feel better" and then yes, the dudes will confess, and yes he'll feel better. Then he'll be convicted for a crime he never commited.

[Frylock]

[quote=DrDeth]However, many dudes who claim they are alien abductees can and do "pass" polygraph exams, and since I hope we all know that they have NOT been "abducted by aliens"- "the machine" is hardly infallible.
QUOTE]

There's an error in this portion of your post, though I suspect you meant to say the correct thing rather than the incorrect thing you actually said.

You say the machine is "hardly infallible," which taken literally is true of course, but it seems by this you mean "the machine often says things happened that did not happen." But no one supposes a lie detector functions to tell us what happened. It tells us, rather, whether a person is attempting to decieve us when he says it happened.

So if someone believes they were kidnapped by a UFO, and the poly registers a positive, this is not an indication that the polygraph has gone wrong in any way, despite the implication in your post that it would be such an indication.

-FrL-

[Guinastasia]

[quote=Frylock]

Originally Posted by DrDeth

However, many dudes who claim they are alien abductees can and do "pass" polygraph exams, and since I hope we all know that they have NOT been "abducted by aliens"- "the machine" is hardly infallible.
QUOTE]

There's an error in this portion of your post, though I suspect you meant to say the correct thing rather than the incorrect thing you actually said.

You say the machine is "hardly infallible," which taken literally is true of course, but it seems by this you mean "the machine often says things happened that did not happen." But no one supposes a lie detector functions to tell us what happened. It tells us, rather, whether a person is attempting to decieve us when he says it happened.

So if someone believes they were kidnapped by a UFO, and the poly registers a positive, this is not an indication that the polygraph has gone wrong in any way, despite the implication in your post that it would be such an indication.

-FrL-

No, he's saying that people lie deliberately, and the polygraph STILL records it as being true.

[Frylock]

[quote=Guinastasia]

Originally Posted by Frylock

No, he's saying that people lie deliberately, and the polygraph STILL records it as being true.

Oh shoot, you're right. For some reason I read him as talking about people who honestly believed they'd been kidnapped.

Sorry!

-FrL-

[muldoonthief]

Originally Posted by Guinastasia

Really? Because I believe requiring polygraph tests for employment is illegal.

Here's a summary of the Federal Law in pdf. You're basically right, but there are plenty of exceptions - Federal, State & Local governments are completely exempted from the law (hah), defense contractors, pharmaceutical companies, & security companies can polygraph employees as well.

Massachusetts has a polygraph law with none of the above exceptions. 10 years or so ago, if you worked for a defense contractor in MA and wanted to get a security clearance that required a polygraph test, they'd send you out of state to take the test. I notice in my link that that's illegal as well - I don't know if that's a new addition to the law or if they've taken a "live & let live" attitude to security clearances, since MA has plenty of defense contractors.

[chorpler]

Originally Posted by mstay

A minor aside, but a brain scan can detect if someone is consciously lying by detecting activation in certain parts of the brain associated with deception or imagination as opposed to recall of facts or experiences. Of course, it's a rather cumbersome and not at all practical...but hey, maybe someday if you suspect your hubby is lying about taking out the trash you can simply pass a Lie Wand(TM) over the scumbag and see who gets to sleep on the couch.

I asked this exact question in the above-mentioned other thread on polygraphs, and Excalibre specifically denied that there is any such brain pattern associated with deception. Clearly, the two of you are going to have to battle it out for cite supremacy!

[GiantRat]

Originally Posted by Campion

The problem is that crimes are generally established by laws, and the laws establishing a crime lay out "elements" of each crime. The prosecution bears the burden of proving by a preponderance of the evidence that each element of the crime has been established. It isn't enough, for example, for the prosecution in a murder case to show that the defendant wielded the murder weapon. The prosecution must also show the appropriate level of intent. So neither DNA nor a polygraph would go far enough.

Then there's the whole "jury as trier of fact" thing and "fifth amendment" thing and "seventh amendment" thing. In short, I don't think it's fair to say that the reason the scientific community hasn't accepted polygraphs (the way they have DNA, for example) is to protect the survival of judges and lawyers who do criminal work.

I'm responding to both Campion and Yabob here for brevity's sake. I didn't mean to imply that protection of the judicial system was the only reason polys aren't used in court (I've been polygraphed for work, and was accused of lying about my drug history and handling of Classified materials - both of which were absolutely NOT legitimate). I fully recognize that the polygraph is a scripted interrogation. The poly is not a solid tool; it is, however, useful when they polygrapher already has facts that the subject doesn not think they possess.

Again, "scripted interrogation." But useful. And yet, not concrete.

[Campion]

Originally Posted by GiantRat

Again, "scripted interrogation." But useful. And yet, not concrete.

In that sense, the polygraph is merely an accessory in a good cop-bad cop routine. Imagine how much more devastating it would be if it were believed to be accurate.

[DrDeth]

[quote=Frylock]

Originally Posted by Guinastasia

Oh shoot, you're right. For some reason I read him as talking about people who honestly believed they'd been kidnapped.

Sorry!

-FrL-

Both. What matters is that we KNOW they aren't telling the truth. Right? If they are lying for the sake of getting cash or honsetly believe that they were kidnapped- the LD has said they aren't lying and thus the LD is wrong. (Please tell me we are all 100% in line with no real alien abductions, right?)

[t-bonham@scc.net]

Originally Posted by Campion

In that sense, the polygraph is merely an accessory in a good cop-bad cop routine. Imagine how much more devastating it would be if it were believed to be accurate.

When I was working at the County IT Dept., we had a District Court Judge who installed an obsolete, non-working mainframe terminal on the corner of his bench. Then he would ask defendants questions like 'do you have any other charges pending?' and say 'I'm going to check this on the computer right here, you know.' Sometimes he would tap a few keys on the dead terminal, and then ask 'what about this one in that other county?'.

He ended up having a lot of them admit to past convictions or current pending charges in other counties; which entered into the sentences they received.

[DocCathode]

Originally Posted by DrDeth

(Please tell me we are all 100% in line with no real alien abductions, right?)

I was once abducted by a group of aliens who wanted my seed. I know they were aliens because they showed me their green cards.

[DrDeth]

Originally Posted by DocCathode

I was once abducted by a group of aliens who wanted my seed. I know they were aliens because they showed me their green cards.

Hmm, maybe the Resident Alien Work permit really is green in the case of an intergalactic alien. ("Green cards" haven't been green in over a decade).

[js_africanus]

Originally Posted by t-bonham@scc.net

When I was working at the County IT Dept.,...

He ended up having a lot of them admit to past convictions or current pending charges in other counties; which entered into the sentences they received.

Heh. I've argued with our current and previous community police officers about the polygraph, and what they describe, in terms of its effectiveness, matches pretty well what Aldrich Ames had to say about polygraphs:

You didn't mention one of the intriguing elements of the interrogations of Dr. Lee which is in fact quite common -- the false representation to the subject of the polygraph results. Because interrogations are intended to coerce confessions of one sort or another, interrogators feel themselves entirely justified in using their coercive means as flexibly as possible to extract them. Consistency regarding the particular technique is not important; inducing anxiety and fear is the point.

Polygraphers are fond of the technique used by psychics called cold reading, as a slightly less dramatic practice than actually lying to the subject about the results. In this sort of cold reading, the interrogator will suggest to the subject that there may be a potential problem, an ambiguous result, to one of the questions and inquire whether the subject knows of anything that might help clear it up, etc, etc.

They think the machine itself is effective because they're confusing the source of the results they say they see. They describe a process of cold reading to get information, perhaps something akin to what John Edward perform, to get people to give up information. And just as some people believe John Edwards can talk to the dead, some people believe the polygraph is effective. One officer described how people can learn to detect lies, and then describes how one investigator uses his skills to get good deals on a car, but the process he describes is not one where the salesman makes a statement and the investigator detects truth or dishonesty; rather, he describes a process of cold reading.

Indeed, when it comes to criminal investigation, the method sounds more like hot reading, where other methods are used in conjunction with cold reading. One officer spoke of how he knew a woman was lying based on her statements & behavior; however, when I asked why he interviewed her in the first place, he said that he already knew she was guilty based on other evidence. With witness reports, evidence, and outright lies, he was very well prepared to read the subject person. He describes the process as being the same as what the polygraphers do, the difference being they're using the machine in addition to all the other props available. At no point in the discussion with these officers does the polygraph machine's operation actually come into play.

Another mistake they make has to do with conditional probabilities. They look at evidence to make guesses about who is guilty before applying techniques. They've already pared down the set of possible criminals based on evidence, and so it becomes much easier to use manipulation and cold (or hot) reading to guess where the person is lying and focus on those areas.

Regardless, that's a small sample of advocates, so caveat lector.