Polygraphs: The evidence

[CFLarsen]

Can you please link to the post where you answered this question unambiguously?

You mean to say you jump in on this part of the discussion without bothering to find out where it started?

Don't waste my time, OK?

OK then assume that everything I think about how a test is handled it wrong. I know very little about how a test is handled and am therefore not basing my questions on anything about that. I am merely trying to obtain information from you.

If you know very little about how a test is handled, you shouldn't criticize those who do.

Then you shouldn't be arguing from the perspective of the underlying mechanism being wrong - as that has no bearing on whether the lie detection results are true. You first need to show us the flaws in the experiments that lead to a woo, no better than chance technique getting better than chance results. How about you dissect some of these experiments for us? If you can show why all the experiments were not fair tests, then people might actually see what you are getting at.

It isn't the experiments that lead to woo, it's the premise that is woo.

Yes, because changes from the baseline in the old "sweat, breath'n'blood pressure" are indicators that seem to be correlated with lying. Whether because of anxiety, excitement, orienting response or something else is very interesting and helpful for making the test more accurate, but excluding one (which I don't think you have by the way) doesn't make the resulting correlations disappear ito thin air.

That's what was once thought, yes. We know today that those can also be indicators of a lot of other things.

What is a hallmark of pseudoscience? Using physiological measures to indicate psychological processes?

No, continuing to base the claim that those can reveal if people lie or not.

Nobody is denying that there are significant error rates. Otherwise it would have a heckofalot better description than "better than chance". And that's why pretty much everyone here argues that although it is not pseudoscience, it should never be used in a real life situation where it would be regarded as "diagnostic" - ie admissable in court, sacking someone based only on the results etc. But that doesn't mean that it can't have any applications.

What are those?

There was no lie. You must be seriously humour impaired.

A liar, when found out, claims it was a joke....

Why is it so hard for you to point out my errors in describing how the challenge works? It is very simple: whether or not you can do preliminary tests, the only group that can actually accept a challenge is the JREF. Their opinion (and specifically Randi's) is the only one that counts. As you said in post #247 - "I can't say whether it would qualify or not. That is entirely up to Randi."

Why can't you accept that Skeptica can do preliminary tests?

First, you realize that you are insisting that I am arguing two different things in the bolded paragraphs, correct?

Second, my question does not rely on how polygraphs work. For the purpose of the challenge, all they have to do is work. And all I am asking you is if in your opinion (considering your stated stance that they are "pseudoscience") a successful test of the polygraph should qualify.

No, no, no. You were talking about how polygraphs work, e.g., when you argued that people don't get excited when they lie. Therefore, answer the question:

You argue that people don't get excited when they lie.

You argue that polygraphs detect changes in people when they lie.

Therefore, you must argue that people get more excited when they lie.

Therefore, you must argue that polygraphs work by detecting these excitements.

Is this correct, yes or no?

 

[Thanz]

A liar, when found out, claims it was a joke....

You must have never seen the movie Fight Club. Regardless, there was no lie. Feel free to actually post any evidence of a lie at your leisure.

Why can't you accept that Skeptica can do preliminary tests?

Why can't you accept that this has nothing to do with preliminary tests?

No, no, no. You were talking about how polygraphs work, e.g., when you argued that people don't get excited when they lie.

I did not argue this.

Therefore, answer the question:

Your questions are based on a faulty premise, as noted above. And again, I have repeatedly said that it matters not how polygraphs work for the purposes of qualifying for the challenge. Why can't you accept this?

 

[Professor Yaffle]

You mean to say you jump in on this part of the discussion without bothering to find out where it started?

Don't waste my time, OK?

I have read the whole thread and I don't recall you unambiguously stating that you couldn't talk about this one aspect - merely that there were some aspects you couldn't talk about. The fact that you have consistently avoided answering this questionleads me to believe that you are claiming that this is indeed one of the aspects you cannot talk about. How can I be wasting yur time on this matter when you could have cleared it up a while ago with a simple yes or no, but instead spent several posts obfustcating or saying that you had already answered. A simple yes or no would have involved so much less time and effort for you.

If you know very little about how a test is handled, you shouldn't criticize those who do.

I am not criticizing you. I am trying to get you to give a straight answer to a straight question.

It isn't the experiments that lead to woo, it's the premise that is woo.

Even if the premise were woo, it would not negate whether or not somethin works. I could give you a woo explanation of how antidepressants work, but thefact that I did that would not negate the fact that they work.

That's what was once thought, yes. We know today that those can also be indicators of a lot of other things.

That is why the polygraph works under controlled conditions where the other things can to some extent be controlled for. And that is why it is only a correlation, and why there is a significant error rate.

No, continuing to base the claim that those can reveal if people lie or not.

A hallmark of pseudoscience is something which is specific to only this thing which you claim to be pseudoscience? The claim that polygraphs can detect lies at a rate better than chance by lookng at physiological changes is based on the fact that under controlled conditions if you look at these physiological measures they correlate with lying and the test performs at a rate better than chance.

What are those?

It could be used to screen a set of suspects to help determine the order in which they are investigated more thoroughly in order to save resources by looking at the most likely suspects first.

Why can't you accept that Skeptica can do preliminary tests?

Talk about straw men! Nobody, to my knoledge has even questioned that. What they have questioned is whether this prohibits you from stating a personal opinion on whether something should be included in the challenge.

No, no, no. You were talking about how polygraphs work, e.g., when you argued that people don't get excited when they lie. Therefore, answer the question:

You argue that people don't get excited when they lie.

You argue that polygraphs detect changes in people when they lie.

Therefore, you must argue that people get more excited when they lie.

Therefore, you must argue that polygraphs work by detecting these excitements.

Is this correct, yes or no?

This is one of the worst pieces of reasoning I have read recently (never mind the fact that it starts with a false premise...)

 

[CFLarsen]

Why can't you accept that this has nothing to do with preliminary tests?

The reason I can do a preliminary test is because I am member of Skeptica.

Do you accept that Skeptica can do preliminary tests?

I did not argue this.

Oh, really?

If you agree that fear and intimidation are not present in lab conditions, and testing done in lab conditions shows an accuracy rate well above chance, then fear and intimidation cannot be the reason the polygraph 'works'.

When I asked: "How can you design a lab test that invokes the same fear reactions as a real life test would?"

you replied:

I wouldn't, because that is not what we are testing. What we are testing is if the theory could be sound - deception causes certain physiological responses, that can be measured, and then interpreted to reveal deception.
...
We don't want that in the lab. The theory behind the polygraph is not based on fear. Therefore, we would want to eliminate that factor in lab testing.

Oops.

Your questions are based on a faulty premise, as noted above. And again, I have repeatedly said that it matters not how polygraphs work for the purposes of qualifying for the challenge. Why can't you accept this?

Oops.

It could be used to screen a set of suspects to help determine the order in which they are investigated more thoroughly in order to save resources by looking at the most likely suspects first.

But that would require that the interrogator knows who is lying?

 

[Professor Yaffle]

But that would require that the interrogator knows who is lying?

Why? All I am suggesting is that those that the polygraph indicates are lying are investigated first (or earlier in the investigation - obviously there would be other preliminary evidence whch would come into the equation). Of course there is also the problem of whether you can "cheat" the test. But the polygraph operator could possibly be trained to spot signs of this and people who seemed to be trying to cheat would also be given priority.

 

[Belz...]

I am asking you about this:

How did you get to computers based on what I wrote? Did I mention computers in the quote?

Claus, did you already forget about computers coming to the same conclusion as human operators in laboratory tests that were NOT designed to extract confessions ?

Here:

No, YOU need to follow the conversation. It was explained to you that computers can determine if someone is lying or not and pretty much agree with the people operating the polygraph. How can a computer get those results through extracting a confession ?

That's one example of me already explaining it to you. Why are you ignoring the words I post and instead focus on the ones I don't ?

 

[Thanz]

The reason I can do a preliminary test is because I am member of Skeptica.

Bully for you. However, I did not ask that. It is irrelevant to my question, because preliminary tests are irrelevant to my question.

Do you accept that Skeptica can do preliminary tests?

I simply don't care.

Oh, really?

Really.

When you asked (in post 321):

Are you saying that when people lie, they can get more calm?

I answered (in post 323):

Not at all, and I don't see how you get that from my response.

Pretty clear.

When I asked: "How can you design a lab test that invokes the same fear reactions as a real life test would?"

you replied:

How do you think that means I said have argued "that people don't get excited when they lie"?

I was arguing against your premise that the only value to a polygraph is in intimidating suspects. I pointed out to you that this intimidation is not present in lab conditions, and there is still a high hit rate.

Oops.
Oops.

Unless you mean these to be your apologies for misstating what I have said, I don't understand your meaning. If you think they prove your point, you may be delusional.

 

[Lucky]

I'm just about to dash out the door, so I'm being lazy. Do you have the specific source for the actual "better than chance" quote?

I found these:

Even where the evidence seems to indicate that polygraph testing detects deceptive subjects better than chance (when using the control question technique in specific-incident criminal investigations), significant error rates are possible, and examiner and examinee differences and the use of countermeasures may further affect validity.

The preponderance of research evidence does indicate that, when the control question technique is used in specific-incident criminal investigations, the polygraph detects deception at a rate better than chance, but with error rates that could be considered significant.

Both with qualifiers. Was there a "better than chance" assertion without qualifiers?

RayG

RayG, I don't mean this in an unfriendly way at all, but please read the whole thread. Your questions have been answered many times. Some of us from other scientific areas have put a lot of time and effort into reading up on the subject, evaluating the evidence, gaining an understanding of the problems, and explaining and discussing the scientific issues with posters who are interested in learning about the subject (and CFLarsen).

I, for one, am getting tired of posting the same thing over and over again.

The main point we are trying to make at the moment is that polygraphy cannot be 'pseudoscience', because it can be shown to work considerably better than chance in controlled laboratory conditions. The NA report makes that quite clear. Whether the proponents' claims of its effectiveness in the field are true, and whether there actually are any valid applications, is another matter entirely (I'd say no to both questions).

To address your specific question briefly:
As with any technology, of course we need qualifiers to indicate the scope of validity. The qualifiers for polygraphy are certainly considerable – indicating (correctly) that it's a technology that works, but not all that well.

My contributions have been concerned with laboratory studies, rather than field ones, as this simplifies the basic issue of whether there's anything to polygraphy beyond intimidation and skilled observation by the operator.

This post of mine points to some data and conclusions about the laboratory studies included in the NA report. Here's a relevant part:

p122 to 124 of the National Academies report summarise the results from 52 datasets from laboratory studies (all the studies of sufficient quality from a literature search).

First, the data (and their errors of estimate; see Appendix H, Figure H-3) clearly fall above the diagonal line, which represents chance accuracy. Thus, we conclude that features of polygraph charts and the judgments made from them are correlated with deception in a variety of controlled situations involving naïve examinees untrained in countermeasures: for such examinees and test contexts, the polygraph has an accuracy greater than chance. Random variation and biases in study design are highly implausible explanations for these results, and no formal integrative hypothesis test seems necessary to demonstrate this point.
Second, with few exceptions, the points fall well below the upper left-hand corner of the figure indicative of perfect accuracy. No formal hypothesis test is needed or appropriate to demonstrate that errors are not infrequent in polygraph testing.
Third, variability of accuracy across studies is high. This variation is likely due to a combination of several factors: “sampling variation,” that is, random fluctuation due to small sample sizes; differences in polygraph performance across testing conditions and populations of subjects; and the varying methodological strengths and weaknesses of these diverse studies. The degree of variation in results is striking.
...
It is important to emphasize that these data and their descriptive statistics represent the accuracy of polygraph tests under controlled laboratory conditions with naïve examinees untrained in countermeasures, when the consequences of being judged deceptive are not serious. We discuss below what accuracy might be under more realistic conditions.

[qimg]http://www.internationalskeptics.com/forums/imagehosting/thum_2550477e34c28029c.jpg[/qimg]

Do you understand what the ROC figure is telling us? (If not, do a search on 'receiver operating'.) Considered solely as a test of whether the claimed effect is real, these results are extremely convincing. They are also well within the range of accuracy required in many real-world screening (though not diagnostic) tests. (If, for example, a pregnant woman has a routine serum screening test for fetal chromosomal anomaly, then the accuracy of that test is reasonably well represented by the better of the two ROC curves above.)

 

[CFLarsen]

Why? All I am suggesting is that those that the polygraph indicates are lying are investigated first (or earlier in the investigation - obviously there would be other preliminary evidence whch would come into the equation). Of course there is also the problem of whether you can "cheat" the test. But the polygraph operator could possibly be trained to spot signs of this and people who seemed to be trying to cheat would also be given priority.

Remind me: Isn't it claimed that the experiments show that you can't tell if the individual lies - but only if a certain percentage of a group lies?

Claus, did you already forget about computers coming to the same conclusion as human operators in laboratory tests that were NOT designed to extract confessions ?

Here:

That's one example of me already explaining it to you. Why are you ignoring the words I post and instead focus on the ones I don't ?

OK, OK: You messed up and don't want to admit it.

Bully for you. However, I did not ask that. It is irrelevant to my question, because preliminary tests are irrelevant to my question.

I simply don't care.

What is your gripe about me and the tests then??

Really.

When you asked (in post 321):
I answered (in post 323):
Pretty clear.

It wasn't about whether you had argued that people get more calm when they lie.

It was about whether you had argued that people don't get excited when they lie.

You claimed you hadn't. I showed that you had.

How do you think that means I said have argued "that people don't get excited when they lie"?

I was arguing against your premise that the only value to a polygraph is in intimidating suspects. I pointed out to you that this intimidation is not present in lab conditions, and there is still a high hit rate.

Unless you mean these to be your apologies for misstating what I have said, I don't understand your meaning. If you think they prove your point, you may be delusional.

Oh, how you wiggle...

 

[Professor Yaffle]

Remind me: Isn't it claimed that the experiments show that you can't tell if the individual lies - but only if a certain percentage of a group lies?

The experiments can show that if you compare a person who failed the polygraph test with a person who passed, the person who failed is more likely to have been lying than the one who passed. We can't say for sure whether either one was lying, but we can use the results as suggestive evidence to help decide which avenues of enquiry to pursue first.

 

[Professor Yaffle]

And Thanz's words which you quoted earlier in no way mean what you think they mean. As far as I followed it, the scenario being talked about was that the external fear and intimidation of the interview can cause the same physiological reading that polygraphers interpret as lying. The counter argument was that in controlled experimental settings, that external cause of fear and intimidation is no longer there, but that the physiological reactions to lying still appear. How you intepret that as arguing that people don't get excited (or anxious) when they lie is completely beyond me. Besides that faulty premise there is a huge hole in your logic where you equate changes in polygraph readings with excitement. As if they could be no other causes for changes in physiological responses (like the orienting response for example).

 

[NoZed Avenger]

OK, OK: You messed up and don't want to admit it.

It is extremely lucky that you do not believe in the vindictive, smiting gods of Irony.

 

[Belz...]

OK, OK: You messed up and don't want to admit it.

WHAT ? I've pointed out to you, three times now, that I asked you about how computers managed to identify the truth value of statements and agreed with human operators, even in non-stressful laboratory conditions.

YOU said that it was all about intimidation. That wouldn't work in the lab, and it certainly wouldn't work if a computer was doing the analysing.


-------

Seriously, when I subscribed to the JREF, I was under the impression that the posters with the most posts were the veteran skeptics, and that since nobody who is NOT a skeptic with a rational mind could possibly spend so much time on a forum like this one, those high-post members were probably good bets as far as opinions and expertise went. Of course, those like Hammegk, Interesting Ian and Iacchus proved me wrong.

But you're a separate case. You are so utterly incapable of admitting to being wrong on ANY minor point that you will distort other people's words and lie to "win".

38,000 posts of belligerance, Claus. Think about it.

 

[CFLarsen]

The experiments can show that if you compare a person who failed the polygraph test with a person who passed, the person who failed is more likely to have been lying than the one who passed. We can't say for sure whether either one was lying, but we can use the results as suggestive evidence to help decide which avenues of enquiry to pursue first.

Let's scroll back and see:

If you test 100 people and the results are expected to be correct 80% of the time then you can draw a conclusion about the group that you cannot draw about an individual.

In the worst-performing study, about 40% of the liars would be caught by a test that caught only 10% false positives. In the best-performing one, 100% of the liars were caught at that level.

So out of a group of 200 people, half liars, we would expect (from a high-performing case) to get about 95 of the 100 liars, and 10 of the truth-tellers; of the 105 people who failed the test, better than 90% were in fact liars. That's pretty good. (Actually, from the best performing case, we'd get all the liars, but ceiling effects enter here and you can't trust that particular number to hold up.)

Perhaps I wasn't clear, I was asking for any hypothetical situation he cared to give that involved interpreting the results of a group of polygraph answers as opposed to a single answer. I am talking about what a polygraph can tell you about a group of tests and Claus is talking about what a polygraph can tell you about a single test. I agree with him on the single test. But his one track mind has stopped him from ever getting to the group of tests question.

Let me sum up the point of the thread for you. I made a statement in another thread that even though you cannot use a polygraph result to determine if an individual is deceptive, you can use it to draw a conclusion about a group of people being tested. If the results are that 100 people are being deceptive per the results of the polygraph, and the polygraph is accurate 80% of the time, then you can say that at least 80% of those people are likely being deceptive.

...

But to say, you cannot use any figure to draw a conclusion about the group test results because an individual test result is not reliable is not a correct statement.[/B]

Why, yes! It was claimed that the experiments show that you can't tell if the individual lies - but only if a certain percentage of a group lies.

Do you agree with the above?

 

[Aoidoi]

Apropo of nothing, and having skipped several pages of the usual CFLarsening of the thread... watched a true crime show the other day on Court TV (now Tru TV, as they remind you every 5 minutes). A law student was attacked and shot at by 2 men who came to his apartment, then had his car go up in flames upon turning on the ignition. The show had the amusing fact included (though not particularly called out) that they administered a lie detector test to the prime suspect, who passed as being truthful about not wanting to harm the victim. They then gave a polygraph to the victim who they suspected was inventing the attacks for attention, and he came across as lying.

Turns out the suspect had in fact hired several people to try to kill the victim. They had pretty much incontrovertable physical and eye witness testimony about it. So in the same case they had both a false positive and false negative. They did mention that the suspect had practiced on how to beat the test, though they didn't specify the method he used. No comment at all on how the victim's one was wrong.

Amusing anecdote, anyway. Thanks to those who did answer my questions a few pages back, I did learn more about the methodology and testing of these things than I previously knew, though I think it still leaves me believing that they aren't accurate enough in real world applications to trust. I guess it's better than rubber hoses in the police station basement.

 

[Thanz]

What is your gripe about me and the tests then??

My 'gripe', if I have one, is that you are using them to hide behind as an excuse for not answering my question regarding whether polygraphs should even qualify for the challenge.

It wasn't about whether you had argued that people get more calm when they lie.

It was about whether you had argued that people don't get excited when they lie.

You claimed you hadn't. I showed that you had.

You showed no such thing. It is at points like these that I really start to question you english comprehension skills. I don't think that you have accurately stated my position on anything in this thread.

Oh, how you wiggle...

Not wiggling. Explaining. For the Nth time.

 

[CFLarsen]

My 'gripe', if I have one, is that you are using them to hide behind as an excuse for not answering my question regarding whether polygraphs should even qualify for the challenge.

And it's been explained to you that you are not in a position to decide which aspects of the challenge that can be discussed in public.

You showed no such thing. It is at points like these that I really start to question you english comprehension skills. I don't think that you have accurately stated my position on anything in this thread.

Not even when I quote you directly...

Not wiggling. Explaining. For the Nth time.

Wiggle, wiggle, wiggle...

 

[Professor Yaffle]

Why, yes! It was claimed that the experiments show that you can't tell if the individual lies - but only if a certain percentage of a group lies.

Do you agree with the above?

Oh for goodness sake Claus. Now I know you are either being deliberately obtuse, or you are extremely unintelligent (I favour the first explanation at the moment). I have already agreed that you cannot tell if a particular person is lying, but you can use the overall statistics from testing on groups of people to give a PROBABILITY that a particular person is lying. If in tests the polygraph can detect lies with eg 70% accuracy, then by logical deduction you can also say that a person who the test indicates is lying is MORE LIKELY to be lying than a person who the test indicates is telling the truth.

It is like something like a prenatal screening test for Down's Syndrome (although i will simplify slightly because the DS screening test doesn't give a yes/no answer but a probablility level above which a person is deemed at risk of carrying a foetus with DS). When a person gets a positive screening test for DS we cannot say anything about whether their baby will have DS. But we can say that a person with a positive result is MORE LIKELY to have a DS baby than a person who gets a negative result. Those with a positive result can then be offered a more definitive test(amniocentesis or CVS). This test isn't offered to all because of the cost, discomfort and risk of miscarriage.

Now I am not saying that the polygraph would definitely have utility in a crime detection scenario for prioritising the order of investigating suspects. there may be too many other factors that would make it worthless, for example it may not work as well outside of lab situations, the rate of people being able to cheat the test may be too high to make it worthwhile etc. I am just saying that it MAY have a possible utility in a situation like this.

 

[NoZed Avenger]

And it's been explained to you that you are not in a position to decide which aspects of the challenge that can be discussed in public.

(1) Which aspects of the Challenge cannot be discussed in public by you, if any?

(2) Why is that?

 

[CFLarsen]

Oh for goodness sake Claus. Now I know you are either being deliberately obtuse, or you are extremely unintelligent (I favour the first explanation at the moment). I have already agreed that you cannot tell if a particular person is lying, but you can use the overall statistics from testing on groups of people to give a PROBABILITY that a particular person is lying. If in tests the polygraph can detect lies with eg 70% accuracy, then by logical deduction you can also say that a person who the test indicates is lying is MORE LIKELY to be lying than a person who the test indicates is telling the truth.

It is like something like a prenatal screening test for Down's Syndrome (although i will simplify slightly because the DS screening test doesn't give a yes/no answer but a probablility level above which a person is deemed at risk of carrying a foetus with DS). When a person gets a positive screening test for DS we cannot say anything about whether their baby will have DS. But we can say that a person with a positive result is MORE LIKELY to have a DS baby than a person who gets a negative result. Those with a positive result can then be offered a more definitive test(amniocentesis or CVS). This test isn't offered to all because of the cost, discomfort and risk of miscarriage.

Now I am not saying that the polygraph would definitely have utility in a crime detection scenario for prioritising the order of investigating suspects. there may be too many other factors that would make it worthless, for example it may not work as well outside of lab situations, the rate of people being able to cheat the test may be too high to make it worthwhile etc. I am just saying that it MAY have a possible utility in a situation like this.

Maybe.

But here's the problem.

You have one large group of suspects. You split them up in two smaller ones and test them separately.

What will you ask each group in order to find out whether one group is more of a "liar" group than the other?

If you ask them about things you don't know, e.g. who did it - well, that's not what polygraphs are for, is it? Because if you could spot when they lie about what you don't know, you wouldn't need the groups but could focus on the individual.

So, you have to ask them about things you already know in order to find out whether each group lies or not. But again, if you already know things about the case, you wouldn't need to put them to the test. So you have to ask them about things you already know but which are unrelated to the case.

In short, you appoint one group of suspects as being more relevant to investigate first, because they lie about things unrelated to the case.

You gain nothing from that kind of use. All you do is waste your time on focusing on a group based on a false premise.