Just how accurate is the polygraph?

[lane99]

The above subject is listed as being included somewhere in this particular forum, but forgive me that I'm unsuccessful to find this info using the "search" function here.

Could someone direct me to where I can read about how accurate the polygraph is/isn't?

[anxietyguy]

Download the TLBTLD on the main antipolygraoh.org homepage. Just read some posts on here and that should give a brief idea of what the polygraph is all about. *tricks*

Anxiety

[dimas]

Flip a coin


In all seriousness the Poly has its uses and while not the most accurate machine in the world it has prevented some individuals with shady pasts from becoming LE officers.  Unfortunately, it has also sacrificed quite a few innocents along the way.

Download the manual/book and read it for yourself.  

[George W. Maschke]

See in particular Chapter 1 of The Lie Behind the Lie Detector, which covers the validity of polygraphy (or the lack thereof).

[lane99]

See in particular Chapter 1 of The Lie Behind the Lie Detector, which covers the validity of polygraphy (or the lack thereof).

Thanks for the referral.  I had seen in the NAS review their opinion that the polygraph seems to be able to discern lie from truth at a rate well above average in specific-incident testing.   And I was wondering how those of you with an "anti-polygraph" viewpoint would cope with that.  I've now seen in TLBTLD that this point is critiqued.  It is interesting reading.

I've a question, though, for clarification purpose:  do those of you who are anti-polygraphy advocates accept or acknowledge that under the right conditions (for example, no countermeasures being employed by the subject), for SPECIFIC-INCIDENT situations, polygraphs do have some validity and can indeed provide some incremental benefit for the task of discerning truth from lie?

I ask this question because the homepage of Antipolygraph.org says polygraphs have no more scientific validity than astrology or tarot cards.  But, this is overstating the case, it seems to me.  

[George W. Maschke]

I've a question, though, for clarification purpose:  do those of you who are anti-polygraphy advocates accept or acknowledge that under the right conditions (for example, no countermeasures being employed by the subject), for SPECIFIC-INCIDENT situations, polygraphs do have some validity and can indeed provide some incremental benefit for the task of discerning truth from lie?

Regarding the incremental validity of polygraphy, the National Academy of Sciences' Committee to Review the Scientific Evidence on the Polygraph notes at p. 214 of its report, The Polygraph and Lie Detection:

There is essentially no evidence on the incremental validity of polygraph testing, that is, its ability to add predictive value to that which can be achieved by other methods.

Because polygraphy lacks both standardization and control, no meaningful accuracy rate is knowable, and no valid inference may be drawn regarding any particular person's truthfulness based upon polygraph chart readings. I think this will become clearer if you read up on polygraph procedure, which is explained in Chapter 3 of TLBTLD. Note also, with regard to your above-cited question, that polygraphers have no reliable way of knowing whether or not a subject has employed countermeasures.

[lane99]

Don't you think the context of the quote from page 214 (which I had read prior to your post) is there is a practical difficulty to obtain evidence regarding efficacy/non-efficacy because in real-life situations there are many variables which are difficult to quantify and control.

However, the opinion is given that the polygraph indeed can ascertain truth from lie at much greater than chance under the right conditions.  Thus it seems to me this website is exaggerating when they equate polygraphs with astrology or tarot cards (I'm assuming that those two practises can NOT perform at a level much greater than chance).

[Twoblock]

lane99

I have done a lot of research, at my own costs, on the polygraph and I can assure you the contents of the "Lie Behind The Lie Detector" are factual.

I am a 74 year old guy that has the BP and heart rate of an athlete because I work hard at my mine and I mountain climb in the off season when I can.

I have a friend that is a MD. I bet him a steak dinner that I could raise my BP and pulse rate at will. After he hooked me up, I told him to say "when" at the time he wanted me to try. First time I used the mental method i.e., falling out of an airplane. Not extremely high but, significant enough to say that I was lieing. The second round I squeezed the old but muscle as tight as I could. He said "I don't know what you are doing but stop it". Then I told him we would do a third one and I would show him about half way between the significant and the dangerous. Which I did. You can do the same thing at a drug store that has a simple BP machine

I enjoyed my free steak.

[Mr. Truth]

If you are an examinee and told the truth and were "judged" to be deceptive, zero percent reliable sounds about right.

If you're a polygrapher and you get someone to confess a la bluffing, the test says you're a liar, let's pray to God and ask for his help today during test, then you (as a polygrapher) would claim something much higher.

Either way, it is a piece of crap.

[George W. Maschke]

Don't you think the context of the quote from page 214 (which I had read prior to your post) is there is a practical difficulty to obtain evidence regarding efficacy/non-efficacy because in real-life situations there are many variables which are difficult to quantify and control.

The plethora of uncontrolled (and, indeed, uncontrollable) potentially confounding variables that beset CQT polygraphy certainly complicate any attempt to determine a meaningful accuracy rate for the procedure. (So, too, does CQT polygraphy's lack of standardization.) But the quoted passage from the NAS report does not offer such difficulties as an explanation for the virtual absence of any evidence on the incremental validity of polygraph "testing." This is an area in which research could be done, but (for reasons that are not too hard to guess) hasn't. For example, a study could be designed comparing the accuracy of decisions made by interrogators using a polygraph versus those made by 1) interrogators without a polygraph, or 2) interrogators using a non-functional polygraph strictly as an interrogational prop.

However, the opinion is given that the polygraph indeed can ascertain truth from lie at much greater than chance under the right conditions.

But it is virtually impossible to know whether "the right conditions" exist with regard to any particular polygraph examination administered to any particular person on any particular day.

Thus it seems to me this website is exaggerating when they equate polygraphs with astrology or tarot cards (I'm assuming that those two practises can NOT perform at a level much greater than chance).

That which is stated on the AntiPolygraph.org home page is: "The simplistic methodology used in polygraph testing has no grounding in the scientific method: it is no more scientific than astrology or tarot cards." This statement is concerned with scientific underpinnings, not potential accuracy "under the right conditions." CQT polygraphy has no scientific basis as it lacks both standardization and control (regarding which, see Dr. Drew C. Richardson's remarks to the NAS polygraph review committee) and is not supported by any theory that explains known facts. It is in this sense that the comparison with astrology and tarot card reading is made.

While a polygrapher might, under the "right" conditions (again, we can't truly know whether such conditions have been met), "discriminate lying from truth telling at rates well above chance, though well below perfection," so, too, might an interrogator who is not using a polygraph. Again, there is no evidence that the former makes determinations that are more accurate than the latter.

[lane99]

Although I might quibble with the wording, and the emphasis given to some of it, fundamentally I don't disagree with much of what others have posted here.  In case anyone is wondering, I'm not trying to defend polygraphy.  As a matter of fact, I have heretofore assumed that is had absolutely NO validity (i.e. I speculated that a polygraph would be unable to discern truth from lie at anything other than chance...no better than flipping a coin).

It was to my surprise, then, that I saw the NAS had said polygraphs DO seem to be able to distinguish truth from lie under certain circumstances (yes, admittedly, very limited and difficult to achieve circumstances.  But circumstances that nevertheless might be achievable).

I should admit then, shouldn't I, that the polygraph sometimes DOES work?  And not just in the sense that flipping a coin might be said to "work"-since that would just be random chance, and wouldn't actually be "working" at all.  

p.s. I'll point out here that when I previously spoke of "incremental benefits" I had been interpreting that at meaning incremental over and above the rate of chance.  I now can see I misinterpreted the context of those words in the report, and that they mean (as they said, and I didn't read clearly) incremental over that which "can be acheived by other methods" (other methods which themselves, presumably, are able to perform at greater than chance).  If I had interpreted this correctly in the first place it would have rung true to me.  It's my assumption that a polygraph couldn't perform more accurately than say human intuition and logical assessment of known facts and circumstances.

[anythingformoney]

No one can really tell you EXACTLY how accurate the polygraph is or is not.  Lab studies can't replicate the real world.  The polygraph, even by the NAS' admission, is much better than chance.  It's the best tool available, so it will be used until something better comes along.  I think I can safely predict that if you live another 25 years you will see something better come along.  Until then, we should all be thankful for what we already have.