The presence and prevalence of false positives is a serious concern in any scientific test. One of the reasons that multiple HIV tests exist is precisely to rule out false positives. However all the tests have a relatively low false positive rate (I admit I don’t know the actual numbers); the probability that two or three tests would read false positive is vanishingly small. The reported inaccuracy of polygraphs, virtually indistinguishable from chance, make them useless for any practical use, even as a backup. Hardcore has made a persuasive case that polygraphs are indeed useless from a scientific perspective.
A trivial point, which I will concede. However, if my job or freedom were in jeopardy, I would not accept either.
I really don’t understand the vehemence of your objections. The links that hardcore presented made fascinating and enlightening reading, and their conclusions seemed well-supported. Granted a polygraph might be a little better than flipping a coin, but that little difference does not seem to be worth expending the energy and passion you seem to be dedicating to this topic.
I have a vague recollection that polygraphs have a high success rate (with very low false-positives) on “guilty knowlege” tests, where strong correlations between physiological response and questions concerning details of the crime (such as naming different possibilities for the color of the victim’s clothes), but I am unable to locate the cite.
Perhaps if you could actually find such a cite, you might indeed advance your case. Otherwise, I must say I find your objections specious and inconsequential.
You did not answer the question. I didn’t ask what you would or would not do, I asked if an assertion made in the first message of a thread is different from an asserion made in the nth message of a thread.
.
You think “go to a library” supports your assertion? Do you think you could get away with that in scientific paper? Testimony befrore Congress? A presentation at a symposium? A debate? No. You say you don’t have several hours to spend wading through those websites (which would not take several hours), yet you think others should spend more time trying to locate and then wading through references?
The only logical conclusion is that you cannot provide references, wither because you are incapable or because those references don’t exist.
It doesn’t matter from where your assertion arose. You made an assertion, you support it.
Hardcore has provided evidence. You are free to assess that evidence as you see fit. You have provided no evidence for anyone to assess.
OK, I’m willing to listen; please define “provided”. Is “go to the library” providing evidence? Is posting links to scientific papers and reviews of scientific literature providing evidence? What, exactly, do you think is providing evidence?
I don’t understand the vehemence of his objections either; but he’s made it clear that he hasn’t seen any evidence because he hasn’t bothered to look at any evidence.
It seems reasonable to interpret The Ryan’s exhortation to “go to the library” as an ironic rejoinder and not literally as a attempt at support his position.
The claim was made that there has been shown that there is no correlation. CAn you see any place in which that has occurred?
I started out, in my very first post, saying that only very sensitive jobs should it be used, a position which hardcore disagreed with on the basis that it has no validity. If you concede the point that they do in have validity, then you agree with me and disagree with hardcore.
I just find it strange that hardcore has made a claim and provided absolutely no evidence for it, and everyone is acting like that’s okay.
I most cewrtainly did answer your question. Do you suffer from some sort of cognitive disorder?
Why do insist on this lie that hardcore has provided evidence? I have already made it clear that I’m not buying it.
How about actually quoting it? Is that too much to ask for? Or at the very least, provide a link to the evidence, rather than a link to the website which has the evidence.
No, you most certainly did not. You answered the question “would you start a thread with an assertion without evidence for that assertion?” My question was (re-stated) “Do you think that you can make assertions without evidence just becasue you didn’t start the thread?”. That question you have not answered.
No. There is evidence in this thread that leads me to believe that others do.
Because it is a fact that hardcore has provided evidence, whether or not you buy it as convincing evidence. So far you have erroneously complained that no evidence was provided (it was); somehow you have convinced yourself that evidence is not evidence unless you are convinced by it. You have complained about the form of presentation of this supposedly non-existant evidence. My complaint is that you have made no persentation whatsoever.
It’s a little obnoxious to ask for quotes of long and complex papers, but hardcore kindly did exactly that when it became obvious that you couldn’t be bothered. He invested far more time that it would have taken you to follow the links and discover the information for yourself; and you are probably the poorer for it, because you’ve missed the opportunity to learn things that hardcore found unnecessary to extract.
Is it too much to ask you to do something far less onerous; provide any evidence for your position, in the form of links or quotes or references?
Pardon? I don’t understand a distinction. The links provided went directly to the papers themselves, rather than to top-level pages requirng you to search. There is more detail in some of the links provided at the end of the papers; is it too much to ask you, if you are interested, to scroll down to the bottom and click; or do you only look at links if all possible links you might take from that page are posted in the original message? What kind of link would you find acceptable?
The claim was made that the trivial correlation is useless and the evidence hardcore offers supports his claim.
In my opinion based only on the evidence presented in this thread, the polygraph’s lack of accuracy does not support its use in any circumstances. Your opinion obviously differs.
Say that I have 10,000 employees of whom 100 are spies. I apply the polygraph to each employee and fire anyone caught lying. At a 10% false negative rate and a 47% false positive rate, I will fire 4,653 honest employees for every 90 spies. I must thus hire an additional 4,743 employees, of whom ~47 are spies (assuming a constant rate of 1/10 and a false negative rate of 10% on the initial screening). Additionaly, I must screen ~9739 people to do so. I have essentially spent half my training budget to remove 43 spies, and darkened the reputation of almost 5,000 absolutely honest people.
Let us look at the situation in game theoretical terms. You have two players with equal finite resources. It costs the spyer $X to train and attempt to introduce a spy. It costs the spyee $Y to train and introduce a new employee. “fp” represents the rate of false positives, and “fn” represents the rate of false negatives (0 < fp < 1 and 0 < fn < 1). “n” is the size of the spyee’s organization (0 < n). It is assumed that the outcome of a test is uncorrelated with performance on previous tests.
It costs (X / fn) to successfully introduce a spy. Once in place, it costs (Y * fp * n) to eliminate fn of the spies. Therefore, to use a technique with the expectation (fn,fp) is justified if and only if ((Y * fp * n) / fn) < (X / fn) or Y < (X / (fp * n)). Thus in our example, the polygraph is useful only if the cost to train an employee is less than 1/4300 of that necessary to attempt to introduce a spy.
Notice too how the false negative rate drops out of the equation.
The Ryan, denying the existence of evidence that has been suitably presented will not win you many supporters, I’m afraid. Keep in mind that I am challenging the scientific validity of lie detectors, so the impetus is on those who believe them to be effective to provide evidence of their claim. While I decry the lack of said evidence you declare
This is certainly true, but we often make decisions about credibility based on the absence of evidence. I suspect you regard the existence of ghosts, goblins, Bigfoot, and Santa Claus as unproven simply because their is no objective evidence for them. I take the same position regarding polygraph validity.
I have been to many libraries, and I have read a great many books. I have yet to see any credible evidence that polygraphs function effectively as lie detectors. Do you know of any?
When faced with statistics showing the polygraph’s accuracy to be barely above 50% and false positives ranging from 43-50%, you replied:
This is a strange point to hang your hat on, especially when you claim scientists consider the polygraph effective and you haven’t seen anything indicating the device doesn’t work. How much lower would the data have to go before you would agree it is ineffective? Would a drop in accuracy from 53% to 49% magically meet your standard?
On the subject of false positives you proclaimed:
I am relatively certain that if any of these tests had a false positive rate anywhere near the 43-50% range, they would be abandoned in favor of something better. Would you consent to a justice system where 50% of the innocent parties were declared guilty?
A particularly distressing problem with the high false positive rate is realized when you combine it with the usually low base rate of guilt. Let’s assume 10 of every 100 people tested are actually guilty. I’ll be generous and assign a 70% accuracy rate. This means that 7 of the 10 guilty will be deemed deceptive, and so will 27 of the innocent ones. This is not an impressive performance.
As you can see from the above example (with numbers skewed in your favor), 27 of the 90 truthful individuals would be called liars when they in fact are not. This doesn’t seem like a well-established positive correlation to me, especially since this outcome could have a profound effect on someone’s life.
At least with the coin flip or Magic Eight Ball, everyone is aware the results are random and not indicative of true guilt. Many people are under the mistaken impression that the polygraph has some legitimacy, and from this the true damage ensues.
Having said that, I likely would choose the polygraph under duress, but not for the same reason as you. It has been well documented that once one understands the polygraph doesn’t work, passing the test becomes quite easy to accomplish. Notorious spies such as Aldrich Ames passed repeated polygraphs without any difficulty.
On a side note, when questioned about the difference between statements made in a thread and statements in the OP, you claimed:
Yet when I complained that you did not grant me this same assumption, you sarcastically noted:
Perhaps you have incongruous guidelines for proper discourse, much like your reasoning with respect to acceptable evidence.
While discussing the comparison between Magic 8-balls and the polygraph, you asserted:
I’m not sure what claim you think I didn’t support, but I did not claim they were completely equal. I said “These devices likely have the same effectiveness in determining trustworthiness”. Perhaps you are quibbling about a couple of percentage points. If so, I submit that you are guilty of thinking like a statistician instead of a scientist.
Hey! Even a statistician wouldn’t accept such lousy numbers!
Although there was the time I was out hunting with a couple of friends. I shot at a deer and missed by 10’ to the left. My friend shot at the same deer and missed by 10’ to the right. The third guy (a statistician) jumped up and yell, “We got him!”
You asked a yes-or-no question. I responded with the word “yes”. Just what part of this are you having trouble following?
I simply do not consider a bunch of quotes from people saying that they don’t believe in the effectiveness of polgraph tests to be “evidence”. I could give you just as many quotes from people who do think they are effective, but I’m not going to because I don’t think that that is a valid argument. Truth is not decided by a democratic vote.
It’s obnoxious to ask that someone actually tell me what the evidence is, instead of just telling me where to look for it?
The first link I followed went to a site that actually supported my position. The second went to the title page of what looks to be at least several dozen pages long. When people cite something in a footnote, do they say “this is the book I found it in, find it yourself”? No, of course not.
I would find acceptable a link entitled “study of polygraph tests” and which linked directly to a page dedicated to study which showed the tests not to be effective.
If you want to present the idea that the correlation is useless as an opinion, I will not take issue with that (I disagree, but I won’t take issue). What I do take issue with is (a) the presentation of this as a fact (b) the claim that evidence has been provided (it hasn’t).
First of all, I don’t think it is necessary to “darken the reputation” of those that fail. Secondly, your numbers give a ratio of about 1% (I assume that you will agree that the use of only one significant digit is appropiate). Assuming that these numbers are correct, and assuming that there is a cost of, say, $1000 for each of rejected candidate, this means that $100,000 will be spent on each spy. Do really think that it is worth less than $100,000 to get rid of a spy? I don’t.
Let us look at the situation in game theoretical terms. You have two players with equal finite resources. It costs the spyer $X to train and attempt to introduce a spy. It costs the spyee $Y to train and introduce a new employee. “fp” represents the rate of false positives, and “fn” represents the rate of false negatives (0 < fp < 1 and 0 < fn < 1). “n” is the size of the spyee’s organization (0 < n). It is assumed that the outcome of a test is uncorrelated with performance on previous tests.
First of all, you seem to be assuming that the agency would train its employees, then test them. Why would it do that?
You also seem to be saying that if it cost more to get rid of a spy than it does for another country to get the spy in place, then you shouldn’t do it (let me know if I’m misunderstanding you, I’m not really clear on what yuou’re saying). That’s absurd. If it costs more to get rid of a bomb than it does to make a bomb, should the police just leave bombs laying around? What is important is not what it cost the other country to introduce the spy, but what it would cost our country to have a spy.
If you are simply stating that polygraphs are not proven to your satisfaction, then I have misinterpreted your position and I apologize. However, if your position is that polygraphs have been proven to not work, then I expect evidence for that claim (and by “evidence”, I mean more than hearsay)
It would be subject to a cost-benefit analysis.
But would they be said to “not work”? And in the case of polygraphs, is there “something better”?
No, because the cost of a false postive is much higher than the cost of a false negative. As I said in my first post, if the cost of a false positive is much lower than the cost of a flase negative, then a low effectiveness is still acceptable.
Assuming that the number of applicants exceeds the number of positions, and assuming that there are no other ways of determining truthfulness, it follows that with a polygraph test, 27% of the truthful candidates will be rejected, while 30% would be rejected without the test. Sure, it isn’t a huge improvement, but it’s better than nothing.
Evidently, you don’t know what the term “correlation” means. Mind telling me what you think it means?
While I think you should have evidence, that doesn’t mean that I expect you to have it. If I leave my wallet in the middle of a street, it should (in a perfect world) be there when I come back. That doesn’t mean i expect it to be.
OK, you are correct; you did answer the question. I saw the period after “Yes” as a comma. I am having trouble with the part where you amplified your answer and made your reply unclear. Completing the quote of your reply:
OK, I now understand that the second sentence does not have anything to do with your reply, and you believe that the original poster must support his/her assertions but nobody else must. Why do you believe this? It seems prima facie ludicrous to me.
a
That’s reasonable, bur you hav mis-characterized the posted evidence. The posts include quotes from people who have studied polygraph testing directly and/or reviewed the scientific literature, presentations. and formal reports, some of them with discussions of the reasons for reaching those conclusions, data that supports those conclusions, and references to more detailed dscussions and further data.
When the evidence is long and complex, when there is significant effort involved in extracting the data and presenting it in a coherent form, when it’s far easier for you to just click and read for yourself; yes, it’s obnoxious to aks what you have asked.
Which link was that?
In informal writing, yes they do. In formal writing,w hen the reference is to the body of the work instead of to some specific portion, yes they do.
When someone claims that some assertion is true, do they refuse to provide any hint of evidence? Not if they want to be taken seriously. When someone claims that a particular reference supports their position, do they refer to it as “he first link I followed”? No, they state which link and why they think it supports their position.
Sadly, that’s not likely to happen until more scientific journals start making their articles available on the Web.
Did I ever say this? My point was that hardcore had apparently started the thread with the intent of convincing people that polygraphs are completely ineffective. I responded tyo this claim with the claim that they were effective, with the intent of not letting what I considered to be a flase statement go unchallenged. As my goal was more modest, i believe that I have less of a burden of proof.
Just to be sure I went over hardcore’s post again, and I was unable to find any refereence to any evidence on which the people have reached their opinions. Can you quote such an instance? If these people did base their opinions on solid evidence, why did hardcore not quote the actual evidence, rather than people who have supposedly seen the evidence?
First of all, if someone wants to prove something, it is their obligation to prove it, nor their readers’. Second of all, it was not easier for me to read it than for him to quote it.
Which link was that?
[/quote]
The APA link.
Your link “Scientific Validity of Polygraph Testing: A Research Review and Evaluation” had such titles as:
Introduction
Federal Polygraph Use
Federal Polygraph Policy Changes
Are you really claiming that the entire paper was relevant, and not a specific portion?
I didn’t think that this claim would be subject to debate. Do you really doubt my claim? If so, follow the APA link. It should be obvious why i think that it supports my position.
OK, I’m a physicist. I want to see numbers. How well does a polygraph work, quantitatively? The only number I see is posted by hardcore, that number being that polygraphs are “barely better than 50 percent”. This supports The Ryan’s point: The polygraph is better than random. So far as I can see, all parties concede that it’s not much better than random… But it is better. Whether it is enough better is a completely different question.
Your answer was “Yes”, and you did not contest my assertion that you had not provided support. That’s not much on its own; but you have provided evidence that you believe that you have no responsibility to support your assertions, because you have consistently refused to provide any support other than “go to the library”.
Ah, I’ve got it now … way back there. The link was, of course, not to the APA; you actually had to follow another link to get to the APA! Horrors! {grin}. I posted that because it was the best pro-polygraph link I could find in a reasonably short time, and I thought it was fair to include some pro-polygraph information, but it’s the best of a bad lot. It is indeed pro-polygraph, but I would think that someone with as high a standard of proof as you would be ashamed to claim that one as support.
At that link, the question of “how accurate is a polygraph” is answered without data, without references, and with a claim that people who find the polygraph inaccurate are misinterpreting the data in a particular way. A little investigation of the other links will demonstrate that this particular misinterpretation, though possible, is not a major contributor to the disagreements over accuracy. In other words, they ducked the question.
I can do better; I can quote two.
“we conducted mail surveys to obtain the opinions of 2 groups of scientists from relevant disciplines: members of the Society for Psychophysiological Research and Fellows of the American Psychological Association’s Division 1 (General Psychology). Survey return rates were high (91% and 74%, respectively). Most of the respondents believed that polygraphic lie detection is not theoretically sound”.
“The internationally respected scientific journal Nature published its conclusions last year. Their aggregated findings were based on the polygraph charts of 207 criminal suspects, which 14 polygraphers scored independently. On the average, they erroneously diagnosed 43% of innocent suspects as deceptive. Such errors, called false positives, ranged as high as 50%.”
That is data. Whether or not you find it convincing, it is data. There is much more at the links, but it’s too much to reasonably expect someone to quote in a message. Finding the data is left as an excercise for the interested reader {grin}.
I think it’s time to step back and recapitulate. I have stated my belief that polygraphs may have a positive correlation with lying, but I believe that correlation is tenuous at best and not sufficient to validate the use of the polygraph in any employment decisions. I’m willing to put some effort into evaluating claims and data that could lead me to believe otherwise. However, you have consistently stated that you are not willing to put in the slightest effort; you have made claims without any support; you have refused to provide any support when asked multiple times; you have claimed that evidence is not evidence because it doesn’t meet your elevated standards; you have claimed that data is not data (apparently for the same reason); and your complaints make it clear that you want to hold others to a much higher standard of proof than you are willing even approach. I agree that hardcore has a responsibility to support his position. He’s made an effort; you have not. I’m willing to listen and read and learn; but there’s nothing to listen to. That makes this entire “debate” very frustrating.
I asked for a reference to evidence on which people are basing their opinions. This only mentions what people’s opnions are; it does not include any such reference to evidence on which those opinions are based.
My claim was that the rate of true positives exceeds that rate of false postives. Any attempt to disprove that assertion must include not just the number of false positives, but also the number of true positives. WAhat’s so difficult to underrstand about that?
Oh, well, whoop-tee-do. The human body contains around 200 bones. The book of Leviticus contains around 600 laws. Whether or not you agree that this data supports my position, it is data.
Seeeing as how that’s pretty much my position, why do continue to disagree with me?
I am wiolling to make some effort.
“elevated standards”? Refusing to accept hearsay is an “elevated” standard? “Elevated” above what?
You said “I was unable to find any refereence to any evidence on which the people have reached their opinions.” The links provided include obviously relevant evidence on which people’s opinions are based. Your examples are obviously irrelevant.
If the number of false positives is unacceptable, then the number of true positives does not matter. Therefore, it is possible to discuss false positives and whether their number is unacceptable without considering the number of true positives. You have chosen to discuss the number of true positives relative to the number of false positives, which is another valid way of approaching the issue; but you have not presented any numbers. It is impossible to discuss your particular claim without the numbers, which you have refused to provide.
Because it’s not exactly your position, and I think the difference is crucial; you have made it clear, from your first post in this thread and onwards, that you think that polygraphs are useful in some employment decisions; I think they are not useful in any employment decisions.
OK, now you have stated that you are willing to make some effort; how about actually making some effort? Post some evidence to support your assertions!
No. However, the evidence provided is certainly not all hearsay, and it is arguable that none of it is hearsay (although the “APA link” that “supports your position” could well be hearsay). Some of the evidence is written by the people who did the scientific work, and most of the rest is written by people who have reviewd the scientific work and is based on that review; this is an accepted technique in scientific literature.
“Just to be sure I went over hardcore’s post again, and I was unable to find any refereence to any evidence on which the people have reached their opinions.” Ok, I used the word data and you used the word evidence, but I think they’re interchangable in this particular situation.
This discussion has definitely degenerated. I suggest we get back on topic and discuss polygraphy. If you want to discuss what is data, what is hearsay, what is evidence, or what must be contained in an original post in Great Debates, I suggest you start another thread. I may or may not choose to participate if you do … but I’ll definitely read your OP to see if you practice what you preach!
The purpose of my examples was to show that “data” and “evidence” are two very different things.
How you can look at a survey of scientists of what they think on a particular subject and not consider it hearsay is beyond me. Seeing as how not a single person quoted mentioned anything that they personally had done, it is safe to say that, as it currently stands, all of the “evidence” was hearsay.
As for the rest of your points, they are matters which I have already gone over. If you think that this discussion is degenerating, perhaps you should stop going over old ground.
I have already pointed out that if the links truly do contain evidence, it would be more convincing if that evidence were actually quoted, rather than simply declared to exist.
What is at issue is not whether the polygraph has acceptable effectiveness, but whether it has any at all.
It is also possible to discuss whether the Dallas Cowboys will win the next Super Bowl. Neither has any relevance to my claim.
I believe that I have also made it clear that the issue of whether polygraphs are effective and whether they should be used are two different issues. I do not understand why your disagreement on one issue has led you to oppose me on the other.