Polygraphs: The evidence

[Belz...]

People,

You are missing the point.

You can, in fact, use polygraphs to tell if people lie or not. What you can't use polygraphs for is to tell if people lie or not by reading the output.

Readings from polygraphs do not show if people lie or not. Using polygraphs puts people in a state of stress so they get confused, so they mess up their confession. That's what interrogations also do: You ask the same questions, again and again, in order to find out if the suspect has his story straight. Using a polygraph, you merely introduce a technobabble factor, to impress the lesser informed.

A polygraph is a an intimidation tool.

I'll return to this:



Astrology does "work": You can, in fact, tell people's futures and reveal their lives from casting their horoscope. Not because the planets influence our lives, but because of cold reading.

If you want to argue that polygraphs work the way they are claimed to work, then:

Explain why the scientific community isn't convinced.

Explain what harm it would do to present the argument at TAM and/or write articles for skeptical/scientific journals.

Do they perform better than chance, yes or no ?

 

[drkitten]

You can, in fact, use polygraphs to tell if people lie or not.

Well, I'm glad you finally acknowledged what was obvious from reading the ROC curves.

What you can't use polygraphs for is to tell if people lie or not by reading the output.

This statement, however, continues to be incorrect, as a number of citations in this thread (taken from the NAS report) indicate. For example, Kircher's study : "The independent evaluator, who numerically scored the charts blindly, correctly diagnosed 87 percent of the subjects." in other words, just by reading the output, he was able to tell if people lie or not with 87 percent accuracy.

Given that he's documented as doing the impossible, I suggest your claims of impossibility are overstated.

Similarly, "different computer decision models, on the average, correctly identified 84.9 percent of subjects." Again, the "impossible" is apparently routine. (And, of course, computer technology is such that the computers couldn't even UNDERSTAND the human answers; the idea of a "lie" as we understand it is inexpressible to computers. But separating hyperplanes in a dataset -- now THAT they understand.)

Readings from polygraphs do not show if people lie or not.

Except that both people and computers can determine from those readings if people are lying, with better than 80% accuracy.

 

[Myshkin]

...What you can't use polygraphs for is to tell if people lie or not by reading the output.

Just to give this dead horse one more whack...the scientific data do not support this claim, and in fact the data generally support quite the opposite, with qualifications.

Why? I don't know. Maybe we are so conditioned to "always tell the truth" that, even when it is explicitly requested for the purposes of an experiment that you must lie, we get uncomfortable about it...as silly as that may be. People blush when they're embarrassed, sweat when they're frightened, and maybe have some subtle physical reactions when they lie.

Look at the data.

ETA - I see the rude and condescending tags have been removed. Thanks.

 

[digithead]

Now having read through this thread and since I do research into this area, I'd like to clear up some misconceptions...

Firstly, the polygraph itself measures what it measure accurately, i.e. physiological response to questions. What is controversial is the method in which questions are asked. There are two basic forms of polygraph test. The first, most widely used and certainly the most controversial is the Control Question Test. Please note in this context that "control" does not have the same meaning as a scientific control in that responses and questions are not standardizable in any fashion. The CQT was the form of polygraphy that the NAS reviewed.

What makes CQT polygraphy so controversial is that it is based on emotional response to questions rather than cognitive response. The NAS concluded that lie detector tests based on emotional response of a subject have no scientific basis. This emotional basis of polygraph lie detection makes it inherently flawed because the polygraph cannot accurately distinguish between the innocent but nervous and the guilty but nervous. In addition, little or no research has been performed on the effects of undiagnosed cardiovascular, endocrinological, and mental illnesses on so-called psychophysiological detection of deception but one can surmise that it's probably not in the direction of improved accuracy. In screening applications, this is especially important.

The reason that CQT polygraphy performs above chance but well below perfection for specific incidents is because specific incidents also involve cognitive responses from subjects which increases its accuracy, however minimal it might be. There is a cognitively based polygraph test called the Guilty Knowledge Test (GKT) or Concealed Information Test (CIT). In this type of test, a series of multiple choice questions are asked of a subject consisting where each question has an answer that only the guilty person would know. This test is based on cognitive response and does have a scientific basis in that it can be expected through orienting response and habituation that humans will have a common physiological reaction to what they know, i.e. cognition. Additionally, GKTs do not rely on conditioned response, emotions, or concerns which are all used in CQT polygraph tests to detect deception.

Additionally, polygraphy is related to known psychological phenomenon called the "bogus pipeline." If you hook someone up to a machine (any machine) and convince the person that it can detect lies, then you get people to confess. If people are aware of the bogus nature of the machine, it does not have this effect.

Lastly, CQT polygraph itself does not measure deceit rather it measures only when the sympathetic autonomic nervous system has been activated via emotional response. If a subject shows a response during the polygraph examination, then the inference is made that the subject is deceptive. It requires a logical leap to assume that the response is due solely to deception because this response can also be attributed to fear of failing (rather than fear of detection), simple test anxiety, anger, or many medical or mental conditions. Simply put, nature did not equip us with a "Pinocchio's nose."

For those that want some reading material, see the following:

Faigman, D.L., S.E. Fienberg and P.C. Stern. 2003. The Limits of the Polygraph. Issues in Science and Technology 20(1):40.

Iacono, W.G. 2001. Forensic "Lie Detection": Procedures without Scientific Basis. Journal of Forensic Psychology Practice 1(1):75-86.

Jones, E.E., and H. Sigall. 1971. The Bogus Pipeline: A New Paradigm for Measuring Affect and Attitude. Psychological Bulletin 76:349-64.

Lykken, D. 1998. A Tremor in the Blood: Uses and Abuses of the Lie Detector, 2nd Edition. Reading, MA: Perseus.

Patrick, C.J. and W.G. Iacono. 1991. Validity of the Control Question Polygraph Test: The Problem of Sampling Bias. Journal of Applied Psychology 76(2):229-38.

 

[David Wong]

If you read the thread more closely, you'll see that all of your points have already been raised and addressed.

 

[digithead]

If you read the thread more closely, you'll see that all of your points have already been raised and addressed.

I read the thread fine. Some have been raised and partially (and sometimes incorrectly) addressed. Much of what is known about the polygraph is rooted in myth...

I also gave citations for people to follow up on...

 

[Reaver]

Additionally, polygraphy is related to known psychological phenomenon called the "bogus pipeline." If you hook someone up to a machine (any machine) and convince the person that it can detect lies, then you get people to confess. If people are aware of the bogus nature of the machine, it does not have this effect.

This reminds me of an episode of 'Balls of Steel':

 

[drkitten]

What makes CQT polygraphy so controversial is that it is based on emotional response to questions rather than cognitive response. The NAS concluded that lie detector tests based on emotional response of a subject have no scientific basis.

If they concluded it, I certainly didn't find it in the report. They were quite harsh on the voice stress analyzer (p. 168) : "Overall, this research and the few controlled tests conducted over the past decade offer little or no scientific basis for the use of the computer voice stress analyzer or similar voice measurement instruments as an alternative to the polygraph for the detection of deception. The practical performance of voice stress analysis for detecting deception has not been impressive. It is possible that research conducted in high-stakes situations would give better results, but we have not found reports of the accuracy of voice stress analysis in such situations."
I find no such statements in the NAS report.

Quite the contrary.

Marston (1917), Larson (1922), and Landis and Gullette (1925) all found elevated autonomic (blood pressure) responses when individuals engaged in deception. Marston (1917) described the underlying psychological state as fear; other writers have conceived it as arousal or excitement. The idea that fear or arousal is closely associated with deception provides the broad underlying rationale for the relevant-irrelevant test format. Subsequent research has confirmed that the polygraph instrument measures physiological reactions that may be associated with an examinee’s stress, fear, guilt, anger, excitement, or anxiety about detection or with an examinee’s orienting response to information (see below) that is especially relevant to some forbidden act. (p. 72)

You went on to write.

This emotional basis of polygraph lie detection makes it inherently flawed because the polygraph cannot accurately distinguish between the innocent but nervous and the guilty but nervous.

This statement is simply false and has been adequately addresed upthread.

In addition, little or no research has been performed on the effects of undiagnosed cardiovascular, endocrinological, and mental illnesses on so-called psychophysiological detection of deception but one can surmise that it's probably not in the direction of improved accuracy. In screening applications, this is especially important.

But not to the simple question of whether on not polygraphs work. We already know that field conditions make the problem more difficult. That doesn't mean that the problem is unsolvable. This argument is an irrelevancy at best.

The reason that CQT polygraphy performs above chance but well below perfection for specific incidents is because specific incidents also involve cognitive responses from subjects which increases its accuracy, however minimal it might be.

Well, I'm glad that you have achieved such a firm theoretical understanding of the reason that polygraphy works, something that apparently eluded the NAS in their extensive 2003 report. Is this understanding universal, or it is only your theory, and one not shared by the wider community?

But beyond that particular bit of sarcasm --- you've just admitted that "CQT polygraphy performs above chance but well below perfection." If you're arguing that CQT polygraphy doesn't work, why offer an explanation of why it does?

Lastly, CQT polygraph itself does not measure deceit rather it measures only when the sympathetic autonomic nervous system has been activated via emotional response. If a subject shows a response during the polygraph examination, then the inference is made that the subject is deceptive. It requires a logical leap to assume that the response is due solely to deception because this response can also be attributed to fear of failing (rather than fear of detection), simple test anxiety, anger, or many medical or mental conditions.

Yes. Similarly, when you catch someone entering your window at 2am with a prybar and an empty bag and wearing a ski mask, then an inference is made that the person is a burglar instead of a drunk on his way to a costume party. When a person with a badge and handcuffs shows up at your door in response to your call to 911, an inference is made that the person is a policeman and not a prostitute with a certain specialized clientele.

I have no problem with making grounded inferences, even ones that I know are not 100% certain. In fact, I'm not sure how you could infer with certainty that the cop was in fact a cop. The simple fact is that genuine cops are more common than the alternative.

The question is not technological, but statistical. If you assert that, say, 1% of the population will have an undiagnosed condition that guarantees they will fail a polygraph examination -- well, that doesn't mean much when I'm examining Kari, Tore, and Grant to see who took the wallet. It means much more when I'm examining all 10,000 employees of General Dynamics to see if any of them are spies. That's one reason that polygraph screening is ineffective -- not because there's anything wrong with the polygraph, but that it's simply hard to identify anything that rare.

 

[digithead]

If they concluded it, I certainly didn't find it in the report. They were quite harsh on the voice stress analyzer (p. 168) : "Overall, this research and the few controlled tests conducted over the past decade offer little or no scientific basis for the use of the computer voice stress analyzer or similar voice measurement instruments as an alternative to the polygraph for the detection of deception. The practical performance of voice stress analysis for detecting deception has not been impressive. It is possible that research conducted in high-stakes situations would give better results, but we have not found reports of the accuracy of voice stress analysis in such situations."
I find no such statements in the NAS report.

That's from Chapter 6 "Alternative Techniques and Technologies" and irrelevant to this conversation...

You cite the beginning of Chapter 3 "The Scientific Basis for Polygraph Testing." Read the rest of it...

Pg. 78 of the NAS report:

"Most comparison question testing formats face the difficult challenge of calibrating the emotional content of relevant and comparison questions to elicit the levels of response that are needed in order to correctly interpret the test results." http://www.nap.edu/openbook.php?record_id=10420&page=78

If they have to calibrate emotional content, then one can conclude that it is emotionally based...

Pg 83.

"The physiological responses measured by the polygraph do not all reflect a single underlying process such as arousal. Similarly, arousing stimuli do not produce consistent responses across these physiological indicators or across individuals. This knowledge implies that there is considerable lack of correspondence between the physiological data the polygraph provides and the underlying constructs that polygraph examiners believe them to measure. On theoretical grounds, it is therefore probable that any standard transformation of polygraph outputs (that is, scoring method) will correspond imperfectly with an underlying psychological state such as arousal and that the degree of correspondence will vary considerably across individuals. Little is known from basic physiological research about whether there are certain types of individuals for whom detection of arousal from polygraph measures is likely to be especially accurate—or especially inaccurate."

They go on in chapter 3 to list all of the endogenous and exogenous threats to polygraph accuracy.

You went on to write.

This emotional basis of polygraph lie detection makes it inherently flawed because the polygraph cannot accurately distinguish between the innocent but nervous and the guilty but nervous.

This statement is simply false and has been adequately addresed upthread.

I disagree, pg. 101:

Basic scientific knowledge of psychophysiology offers support for expecting polygraph testing to have some diagnostic value, at least among naive examinees. However, the science indicates that there is only limited correspondence between the physiological responses measured by the polygraph and the attendant psychological brain states believed to be associated with deception—in particular, that responses typically taken as indicating deception can have other causes.

Besides the NAS report, my opinion relies on a large bulk of research about the polygraph both pro and con. CQT polygraph is inherently unreliable.

In addition, little or no research has been performed on the effects of undiagnosed cardiovascular, endocrinological, and mental illnesses on so-called psychophysiological detection of deception but one can surmise that it's probably not in the direction of improved accuracy. In screening applications, this is especially important.

But not to the simple question of whether on not polygraphs work. We already know that field conditions make the problem more difficult. That doesn't mean that the problem is unsolvable. This argument is an irrelevancy at best.

I agree that the problem is not unsolvable but in my opinion CQT polygraph should be abandoned in favor of cognitively based tests that have internal, construct, and domain validity which the CQT does not have.

Additionally, it is imperative to research and understand the threats that undiagnosed conditions pose to any polygraph test, let alone the CQT.

Well, I'm glad that you have achieved such a firm theoretical understanding of the reason that polygraphy works, something that apparently eluded the NAS in their extensive 2003 report. Is this understanding universal, or it is only your theory, and one not shared by the wider community?

But beyond that particular bit of sarcasm --- you've just admitted that "CQT polygraphy performs above chance but well below perfection." If you're arguing that CQT polygraphy doesn't work, why offer an explanation of why it does?

Your sarcasm aside, this understanding is shared by most of the scientists in psychophysiological testing and its implementation.

My argument about CQT is that it a flawed procedure rife with errors that undermine its usefulness. I'm about to submit a journal article based on this, I'll PM you a copy when I finish it.

I explain why it "works" with specific incidents because it has a cognitive component (i.e., knowledge of the crime by the guilty party) which makes it more accurate.

The question is not technological, but statistical. If you assert that, say, 1% of the population will have an undiagnosed condition that guarantees they will fail a polygraph examination -- well, that doesn't mean much when I'm examining Kari, Tore, and Grant to see who took the wallet. It means much more when I'm examining all 10,000 employees of General Dynamics to see if any of them are spies. That's one reason that polygraph screening is ineffective -- not because there's anything wrong with the polygraph, but that it's simply hard to identify anything that rare.

The problem is that we don't know what conditions will unduly affect any polygraph test beyond rudimentary guesses. Most polygraphers will not test anyone if they have the common cold because they know it affects the test regardless of test type. If the common cold can affect the test, what does hypertension, hyper- or hypothyroidism, bipolar, or other illness do to it? How about general mood?

DrK, I'm not trying to be snarky. My research into CQT polygraph deals with its use in sex offender programs. Thirty-eight states rely on CQT to determine if sex offenders are complying with conditions of parole. Similar to the NAS conclusion that CQT polygraph is a threat to national security, I feel its use in post-conviction sex offender treatment poses a similar threat.

This is a matter of using an unscientific and nonstandardizable procedure to screen people for positions of authority and to monitor behavior. Shouldn't we expect more from it than "better than chance but well below perfection?"

Do me a favor, please read Dr. Lykken's (1998) book A Tremor in the Blood and let me know if your opinion still stands. The late Dr. Lykken was a well-respected psychologist from the University of Minnesota who did a ton of research on the polygraph. He was one the first to show that psychopaths have a lower fear arousal than normal people. It sparked his interest in the polygraph...

Regards...

 

[drkitten]

Your own citations disprove you.

I disagree, pg. 101:

Basic scientific knowledge of psychophysiology offers support for expecting polygraph testing to have some diagnostic value, at least among naive examinees. However, the science indicates that there is only limited correspondence between the physiological responses measured by the polygraph and the attendant psychological brain states believed to be associated with deception—in particular, that responses typically taken as indicating deception can have other causes.

You originally claimed that "The NAS concluded that lie detector tests based on emotional response of a subject have no scientific basis."

The actual words of the NAS report directly contradict this. They're not wildly enthusiastic about the technology, but they specifically acknowledge that it has a valid scientific basis (as well as their specific statements that it performs well).

Besides the NAS report, my opinion relies on a large bulk of research about the polygraph both pro and con. CQT polygraph is inherently unreliable.

Given how you've misrepresented the very quotation you offer in support, I see no reason to treat your representation of the "large bulk of research" with any confidence.

Especially since the "evidence" that you cite against the CQT polygraph was, by and large, also available to the NAS and was not sufficient to shift their opinion. The Lykken book that you recommend was, in fact, specifically cited by the NAS but they evidently found it entirely unpersuasive.

I explain why it "works" with specific incidents because it has a cognitive component (i.e., knowledge of the crime by the guilty party) which makes it more accurate.

Yeah. That is to say, it works, and you have a theory of why.

And then you turn around and tell me that it doesn't work. Mere sarcasm cannot convey my contempt for this particular rhetorical flourish.

The problem is that we don't know what conditions will unduly affect any polygraph test beyond rudimentary guesses.

Which does not make polygraphy unreliable or pseudoscience. You're making the merely "good" the enemy of the "best" here (a common failing among academics) -- just because we have the potential for technology that works better does not mean that the old technology should be abandoned; if that were the case, we would have had electric cars fifty years ago -- and we'd probably still rely on the railroads for most of our transportation, because the cars would neither have the range nor the cargo capacity to be useful. We still don't know how to calibrate antipsychotic medicines beyond rudimentary guesses, so I suppose we should stop using them. Heck, we can't even perform spam filtering successfully.


Case in point:

This is a matter of using an unscientific and nonstandardizable procedure to screen people for positions of authority and to monitor behavior.

It is neither unscientific nor nonstandardizable; the science is well documented, and there are arguably too many standards and protocols. There are certainly abuses possible, when (ill-informed) people expect more of a tool than the tool is capable of delivering. But that's not the tool's fault.

Shouldn't we expect more from it than "better than chance but well below perfection?"

Well, let me know when you get "perfection" out of your technology, then. You'll be the first, you know. I don't think we've got any other test -- forensic, criminological, or diagnostic -- that can achieve "perfection." But I'll be sure to point out that any level of success of your own technology renders it -- by your own admission -- "unscientific and nonstandardizable."

I'm sorry to be so harsh. But I certainly hope that your standards for treatment of previous research is more even-handed in your journal article, or alternatively, that I don't end up reviewing it. If I caught the statement "the NAS concluded that lie detector tests based on emotional response of a subject have no scientific basis" in your article, I would read no further.

 

[verisimilidude]

A cold can affect the test?

Have the physiological conditions of a cold which affect the test been identified? If so, are they unique to that condition?

Correct me if I'm wrong, but it appears that a number of variables, which cannot be reliably controlled, can affect the accuracy of the test. If so, it makes me wonder how its use can be justified in any context. Does anyone other than those looking for a simplistic solution to a complex problem (the feds, corps.,etc.) really advocate that an individual be sanctioned for "failing" this test?

 

[Skeptic Ginger]

I haven't had time to keep up with this thread but I intend to catch back up later. In the mean time, Claus, I don't need to do any more research on polygraphs, read the huge, detailed GAO report I linked to that you quoted in the OP. Criminy, how much research should a person need to post to support their conclusions. It contains a thorough review of many studies. In some studies, per the analysis I referred to initially, the majority of the results were consistent with the polygraph being at least reasonably reliable especially in a group of people new to it and not trying to deceive the test.

All of the sources you cited debunking polygraphs were looking at it from the standpoint of assessing one individual. We all agree, I believe, that it is not accurate enough to fire someone, label them a spy, or send someone to jail.

There are other uses such as tricking someone into confessing in a police interview.

The context of my hypothetical application to the thread on the 100 atheists had absolutely nothing to do with how one might actually use a polygraph. The only thing I would change about my original statement, "you could assume most of the people who were supposedly deceitful were lying", is adding a caveat "in the ideal testing circumstances". The urban myth claiming there actually were polygraph results asking 100 atheists if they didn't believe god existed was as far as I can tell completely made up. If you can make up the claim, I can add "ideal testing circumstances" to my entirely hypothetical statement.

I have said repeatedly the polygraph studies showed wide range of accuracy of results. The statistical assessment of 100 test results was a valid concept. I posted an extremely detailed analysis of the research on polygraphs.

You, Claus are fighting Cyrano's windmills. You seem to think I've made claims about polygraphs which in the context you place them in, I did not make. The conclusions in the GAO report seemed adequate enough for me. If you want to argue, then argue against the GAO report. I'm not aware I disagree with anything in it.

 

[Skeptic Ginger]

A cold can affect the test?

Have the physiological conditions of a cold which affect the test been identified? If so, are they unique to that condition? ...

Yes, heart rate, skin temperature, respiratory rate can all be affected by a respiratory infection.

As for, does anyone advocate using the polygraph to sanction someone, I don't believe so in the skeptical community but it is used in businesses to do so.

 

[digithead]

Your own citations disprove you.

No, they don't...

You originally claimed that "The NAS concluded that lie detector tests based on emotional response of a subject have no scientific basis."

The actual words of the NAS report directly contradict this. They're not wildly enthusiastic about the technology, but they specifically acknowledge that it has a valid scientific basis (as well as their specific statements that it performs well).

Pg. 81

To the extent that these principles do not hold universally, an examiner’s rapport with the examinee, the desired understanding of the polygraph examination and questions, and the clinical skill in determining the person’s veracity (i.e., detection of deception from demeanor) are all important in distinguishing among individuals who have physiological responses not indicative of deception (e.g., anxiety or anger regarding relevant questions, insufficient emotionality about the comparison questions), those who have physiological responses indicative of relatively innocuous transgressions, and those who have physiological responses indicative of significant transgressions. These distinctions are made on the basis of clinical judgment, which, though sometimes accurate, does not stand on a good foundation of theory or empirical evidence. There is little basis for relying on the accuracy of clinical judgments, especially in individual cases, without such a foundation. http://www.nap.edu/openbook.php?record_id=10420&page=81

Empasis is mine. No, they don't acknowledge that it has a scientific basis. In fact, they say that it does not have any foundation in which to believe CQT polygraph could have any accuracy.

Given how you've misrepresented the very quotation you offer in support, I see no reason to treat your representation of the "large bulk of research" with any confidence.

I've misrepresented nothing...

Especially since the "evidence" that you cite against the CQT polygraph was, by and large, also available to the NAS and was not sufficient to shift their opinion. The Lykken book that you recommend was, in fact, specifically cited by the NAS but they evidently found it entirely unpersuasive.

No, they found Lykken's book to be very persuasive.

Additionally, Stephen E. Feinberg, one of the main authors of the NAS polygraph study in his testimony to Congress stated:

"Let me just conclude by reminding you that polygraph testing rests on weak scientic underpinnings despite nearly a century of study. Much of the available evidence for judging its validity lacks scientic rigor, and our committee sifted that evidence and the report makes clear the limitations of the polygraph for the present context."

From Fienberg, S.E. and P.C. Stern. 2005. In Search of the Magic Lasso: The Truth About the Polygraph. Statistical Science 20(3):249–60.

Yeah. That is to say, it works, and you have a theory of why.

As I stated, when you have a specific incident, cognition comes into play regardless of the emotional basis of the comparison question. It is in this situation that the CQT becomes more like its relative the GKT...

And then you turn around and tell me that it doesn't work. Mere sarcasm cannot convey my contempt for this particular rhetorical flourish.

When I say it doesn't work, I mean that it doesn't work consistently. You have to separate out utility of the polygraph from its validity and accuracy. The NAS said that the CQT polygraph has utility in that it might deter spies and possibly catch a few but it has little validity and its accuracy cannot be reliably estimated. They actually refused to state an accuracy number. The one they use in their calculations (80%) is a hypothetical and they do not believe that CQT polygraph has very high accuracy.

Which does not make polygraphy unreliable or pseudoscience. You're making the merely "good" the enemy of the "best" here (a common failing among academics) -- just because we have the potential for technology that works better does not mean that the old technology should be abandoned; if that were the case, we would have had electric cars fifty years ago -- and we'd probably still rely on the railroads for most of our transportation, because the cars would neither have the range nor the cargo capacity to be useful. We still don't know how to calibrate antipsychotic medicines beyond rudimentary guesses, so I suppose we should stop using them. Heck, we can't even perform spam filtering successfully.

Except that we do have a better type of test, GKT, that does not have the same pitfalls as CQT. GKT also has been shown to have a sound theoretical basis (orienting response) that the NAS rejected as plausible for CQT.

It is neither unscientific nor nonstandardizable; the science is well documented, and there are arguably too many standards and protocols. There are certainly abuses possible, when (ill-informed) people expect more of a tool than the tool is capable of delivering. But that's not the tool's fault.

Uniformity of process is not the same as validation nor is it standardizable in a psychological sense. Every person in a CQT polygraph serves as their own baseline, hence it is an unstandardizable test. Again to quote the NAS, pg. 213:

Despite efforts to create standardized polygraph testing procedures, each test with each individual has significant unique features. http://www.nap.edu/openbook.php?record_id=10420&page=213

That sounds like unstandardizable to me.

Well, let me know when you get "perfection" out of your technology, then. You'll be the first, you know. I don't think we've got any other test -- forensic, criminological, or diagnostic -- that can achieve "perfection." But I'll be sure to point out that any level of success of your own technology renders it -- by your own admission -- "unscientific and nonstandardizable."

I don't expect perfection but I expect it to have validity, reliablity, and accuracy, all of which the CQT polygraph does not have.

I'm sorry to be so harsh. But I certainly hope that your standards for treatment of previous research is more even-handed in your journal article, or alternatively, that I don't end up reviewing it. If I caught the statement "the NAS concluded that lie detector tests based on emotional response of a subject have no scientific basis" in your article, I would read no further.

That's your perogative. I'm submitting it to an offender treatment journal though, are you a reviewer for any of those?

You and I have read the same NAS report and have come to differing conclusions about its accuracy, validity, and utility. That's fine. I really don't understand how anyone can read the NAS report and come away thinking that CQT polygraph was useful, accurate or grounded in science but then nothing much surprises me anymore when it comes to human behavior.

However, you can't argue that the NAS concluded that reliance on CQT polygraph in screening applications is unwise and poses a threat to national security. If the CQT polygraph was so useful, why do you think they draw such a conclusion?

I'm all for future research and efforts to further ground polygraphy in science. But CQT polygraph to me is a dead end. Efforts should be made towards GKT polygraph.

Finally, given what you've read so far, which do you think is better, GKT or CQT?

DrKitten, regards...

 

[digithead]

Yes, heart rate, skin temperature, respiratory rate can all be affected by a respiratory infection.

As for, does anyone advocate using the polygraph to sanction someone, I don't believe so in the skeptical community but it is used in businesses to do so.

Private businesses and most non-Law Enforcement public agencies cannot use the polygraph or force an employee to take a polygraph under the 1988 Federal Polygraph Protection Act.

The FPPA excluded law enforcement and national security agencies. So these agencies can still use the polygraph on their employees and potential employees. The controvery of its use in the DOE and the Wen Ho Lee case predicated the NAS review.

Also, polygraph can be used to sanction some people. CQT polygraph is used in what is called the Containment Method, a form of Post Conviction Sex Offender Treatment. Thirty eight states use the CQT polygraph to screen sex offenders to see if they are complying with their treatment and supervision protocols. Failing a polygraph and refusing to take a polygraph are grounds for revocation.

 

[CFLarsen]

Except that both people and computers can determine from those readings if people are lying, with better than 80% accuracy.

Present that at TAM, and/or write an article for Skeptic Magazine.

The context of my hypothetical application to the thread on the 100 atheists had absolutely nothing to do with how one might actually use a polygraph. The only thing I would change about my original statement, "you could assume most of the people who were supposedly deceitful were lying", is adding a caveat "in the ideal testing circumstances". The urban myth claiming there actually were polygraph results asking 100 atheists if they didn't believe god existed was as far as I can tell completely made up. If you can make up the claim, I can add "ideal testing circumstances" to my entirely hypothetical statement.

I have said repeatedly the polygraph studies showed wide range of accuracy of results. The statistical assessment of 100 test results was a valid concept. I posted an extremely detailed analysis of the research on polygraphs.

You, Claus are fighting Cyrano's windmills. You seem to think I've made claims about polygraphs which in the context you place them in, I did not make. The conclusions in the GAO report seemed adequate enough for me. If you want to argue, then argue against the GAO report. I'm not aware I disagree with anything in it.

I think it is extremely important that you present a paper at TAM.

It won't have to be a huge presentation. Just a few slides, a brief paragraph or two, and some references. I suspect that the questions that invariably will follow might take some time. But that's what you can expect, when you make such a claim.

This is big, skeptigirl. Scientific evidence that polygraphs work! This is exactly what makes TAM so great!

You know what to expect. You have months to prepare. Should be a walk in the park.

Will you do it?

 

[Pythra]

CFL, if you disagree with skeptigirl and dr kitten's claims (as you clearly do), will you please refute them and stop this ridiculous and frustrating dance of cowardice?

Count me among the readers of this thread who are fed up with your argumentative "technique" and would like to see some evidence.

 

[athon]

CFL, if you disagree with skeptigirl and dr kitten's claims (as you clearly do), will you please refute them and stop this ridiculous and frustrating dance of cowardice?

Count me among the readers of this thread who are fed up with your argumentative "technique" and would like to see some evidence.

Probably best do what most people do here and just ignore Claus's silliness. He's like the mad old uncle off his med's who nobody really takes seriously anymore. His 'present at TAM' tactic has become kind of annoying, I admit, but just as pointless as most of his points.

That said, the thread has been nicely saved by the interesting discussion between Dr. Kitten and digithead. Thanks folks. I can't claim a lot of knowledge on the topic, but have formed the opinion that it could well offer a better than chance ability to detect whether somebody is stating something that they know is a lie, but the confidence in this outcome is not sufficient that any drastic action, such as conviction of a crime, could be balanced on it.

Athon

 

[CFLarsen]

Probably best do what most people do here and just ignore Claus's silliness. He's like the mad old uncle off his med's who nobody really takes seriously anymore. His 'present at TAM' tactic has become kind of annoying, I admit, but just as pointless as most of his points.

It is anything but pointless.

Do you attend TAM solely to hear what you already agree with? Wouldn't you be absolutely thrilled to hear evidence of something considered a pseudoscience?

If Gary Schwartz would attend TAM, would you not challenge his claims?

 

[articulett]

Because polygraph tests are based on stress responses in the body, it can be fooled. I don't know if someone has mentioned this yet, but I remember Shermer talking about how contracting the anal sphincter to create a false stress response during the unstressful questions and relaxing it during the stressful ones can make for inconclusive readings.

I think it does help to know whether someone is willing to take them. If you were not guilty of a loved ones disappearance, and you were eager for them to stop suspecting you so that the right person could be located, then you would probably be eager to clear yourself... and if you avoided doing so, I think it's a valid red flag--

I've heard that sociopaths can often fool polygraphs because they don't have the stress response to lying that most people do.

I've heard of more recent brain imaging techniques where a suspect is shown an item that should look familiar and register a certain way in the brain if he recognizes it... (various pix are shown... some involving things only the perpetrator would find memorable)--and the brain reacts similarly to all the pix if there is no recognition-- I think this will be an interesting tool... but I think this will always be more useful in clearing suspects rather than in convicting them. It can help focus a search, however.