...Just not in the way we're led to believe. They don't detect lies (or at least the evidence isn't very good), but because people believe they work, a polygraph test (or threat of one) can still be useful. See: http://skepdic.com/polygrap.html
But Polygraphs Work...
- Thread starter RonZ
- Start date
-
- Tags
- polygraphs
Ladewig
I lost an avatar bet.
- Joined
- Dec 4, 2001
- Messages
- 28,828
If false positives ruin people's lives (Wen-Ho Lee) and false negatives give counter-espionage agents a mistaken sense of security (Aldrich Ames, Larry Wu-Tai Chin, Brian P. Regan, Robert Hanssen), then the question then becomes is it useful or accurate enough to use within the intelligence community. I say the answer is "no."
Also, some of the people who have committed crimes know that polygraphs don't work and have studied methods for beating the machine.
Also, some of the people who have committed crimes know that polygraphs don't work and have studied methods for beating the machine.
Segnosaur
Penultimate Amazing
Just curious...
I know that there are a lot of problems with polygraphs giving false positives and false negatives (and thus should never be trusted); however, I've never come across any 'statistics' on their accuracy. (Most of the arguments against them complain about how they can be wrong, and give examples of cases that show that.)
Do they give correct results more often than would be expected by 'chance' alone? (Not enough to say "polygraphs work", but enough to say "statistics show that its more accurate than random chance")
I know that there are a lot of problems with polygraphs giving false positives and false negatives (and thus should never be trusted); however, I've never come across any 'statistics' on their accuracy. (Most of the arguments against them complain about how they can be wrong, and give examples of cases that show that.)
Do they give correct results more often than would be expected by 'chance' alone? (Not enough to say "polygraphs work", but enough to say "statistics show that its more accurate than random chance")
When I was a Scientologist™, admitting that you'd "fooled" the e-meter (not at all difficult to do) was enough to get you slapped with an ethics order and declared an SP, because CoS doctrine maintains only a suppressive person would want to fool the e-meter and only and is capable of doing so. 
Beanbag
Illuminator
- Joined
- Jun 7, 2003
- Messages
- 3,468
Polygraph reliability
The only figure I've heard for the "reliability" of the polygraph determining whether a reply was true or false was 70 to 90%. This particular figure comes from an article in Reader's Digest (hey! at least it ain't the National Enquirer) I read many years ago. This was an optimistic figure provided by the polygraph industry. I'm sure the polygraph detractors would provide a much lower figure.
I filed that number away in my head because at that time my line of work might require me to be polygraphed (the joys of government security clearances). It was comforting to know that if I was asked ten questions, most likely they'd misinterpret the response for at least two of them. Think about it: if you had a car that wouldn't start one out of five times you went to work, you'd get rid of it.
Of course, there's the apocryphal story about the company who owned their own polygraph, complete with a supervisor trained to operate it. Materials were being pilfered regularly, so they instituted a series of questionings, and several employees were "let go" because of "untoward responses" detected by the operator. Thefts still continued, and were a regular occurrence until the very polygraph operator was caught red-handed in the act.
A polygraph gives a good indication of a person's state of agitation, nothing more. I can think of quite a few topics for questioning that would blow my agitation level off the map (child molesters, the way society exempts religions from most responsibilities, people who claim Rolexes are the most sophisticated watches in existance -- important stuff like that).
Call it a placebo effect -- if you believe a polygraph will catch you telling a lie, then it probably will. If you know it's a load of hooey, no more reliable than running a hot knife blade over your tongue and seeing if it blisters (they did that in the Middle Ages), then there's little or no effect to be measured. Of course, if I believed that God told me it was OK to lie about certain matters, that might take precedence over whatever fears I might have about the polygraph.
Regards;
Beanbag
The only figure I've heard for the "reliability" of the polygraph determining whether a reply was true or false was 70 to 90%. This particular figure comes from an article in Reader's Digest (hey! at least it ain't the National Enquirer) I read many years ago. This was an optimistic figure provided by the polygraph industry. I'm sure the polygraph detractors would provide a much lower figure.
I filed that number away in my head because at that time my line of work might require me to be polygraphed (the joys of government security clearances). It was comforting to know that if I was asked ten questions, most likely they'd misinterpret the response for at least two of them. Think about it: if you had a car that wouldn't start one out of five times you went to work, you'd get rid of it.
Of course, there's the apocryphal story about the company who owned their own polygraph, complete with a supervisor trained to operate it. Materials were being pilfered regularly, so they instituted a series of questionings, and several employees were "let go" because of "untoward responses" detected by the operator. Thefts still continued, and were a regular occurrence until the very polygraph operator was caught red-handed in the act.
A polygraph gives a good indication of a person's state of agitation, nothing more. I can think of quite a few topics for questioning that would blow my agitation level off the map (child molesters, the way society exempts religions from most responsibilities, people who claim Rolexes are the most sophisticated watches in existance -- important stuff like that).
Call it a placebo effect -- if you believe a polygraph will catch you telling a lie, then it probably will. If you know it's a load of hooey, no more reliable than running a hot knife blade over your tongue and seeing if it blisters (they did that in the Middle Ages), then there's little or no effect to be measured. Of course, if I believed that God told me it was OK to lie about certain matters, that might take precedence over whatever fears I might have about the polygraph.
Regards;
Beanbag
Ron,
If you are trying to detect 1 person out of a group of 100 people and the false positive rate is 10%, then 10 people are going to test falsely positive. Worse still, if the very person you are trying to detect is likely to have characteristics that make him good at "beating" a lie detector, then the false negative rate is likely to be so high that it is very likely that the one person you are trying to detect will not even be amongst those who test positive.
The seems to be the main problem with the polygraph as far as I can understand.
BillyJoe.
If you are trying to detect 1 person out of a group of 100 people and the false positive rate is 10%, then 10 people are going to test falsely positive. Worse still, if the very person you are trying to detect is likely to have characteristics that make him good at "beating" a lie detector, then the false negative rate is likely to be so high that it is very likely that the one person you are trying to detect will not even be amongst those who test positive.
The seems to be the main problem with the polygraph as far as I can understand.
BillyJoe.
George Maschke
Scholar
- Joined
- Aug 15, 2001
- Messages
- 54
Because polygraphy has no scientific basis (it lacks standardization and is fraught with uncontrolled -- and uncontrollable -- potentially confounding variables), no meaningful accuracy/error rate is knowable for it.
You'll find a brief discussion of the scientific status of polygraphy in Chapter 1 of AntiPolygraph.org's free e-book, The Lie Behind the Lie Detector, which may be downloaded as a 1 mb PDF file here:
http://antipolygraph.org/lie-behind-the-lie-detector.pdf
Chapter 3 describes in detail how the test "works" (and doesn't), providing a detailed expose of the trickery (not science) involved, and Chapter 4 explains how anyone can beat a polygraph "test." (You don't have to go to spy school.)
You'll find a brief discussion of the scientific status of polygraphy in Chapter 1 of AntiPolygraph.org's free e-book, The Lie Behind the Lie Detector, which may be downloaded as a 1 mb PDF file here:
http://antipolygraph.org/lie-behind-the-lie-detector.pdf
Chapter 3 describes in detail how the test "works" (and doesn't), providing a detailed expose of the trickery (not science) involved, and Chapter 4 explains how anyone can beat a polygraph "test." (You don't have to go to spy school.)
davefoc
Philosopher
I have read various discussions over the years about polygraphs but I don't ever recall seeing real numbers being published about the accuracy.
The numbers provided by the polygraph advocates are crap IMHO. How was the number arrived at? What does the number mean? Are other factors besides the examination taken into account by the examiner to determine whether the person being examined is telling the truth or not?
What I would find interesting are the results of any objective studies done on polygraph effectivity.
I think it is likely that polygraphs are highly unreliable in determining whether somebody is telling the truth or not, but exactly how unreliable are they? In general can they determine truth better than chance? Are there specific instances where they work well? Can they determine truth better than an insightful observer without the polygraph? My sense of it is that they are at least somewhat better than chance at determining whether a person being tested is telling the truth or not if that person believes that polygraphs work. Is this even true?
The numbers provided by the polygraph advocates are crap IMHO. How was the number arrived at? What does the number mean? Are other factors besides the examination taken into account by the examiner to determine whether the person being examined is telling the truth or not?
What I would find interesting are the results of any objective studies done on polygraph effectivity.
I think it is likely that polygraphs are highly unreliable in determining whether somebody is telling the truth or not, but exactly how unreliable are they? In general can they determine truth better than chance? Are there specific instances where they work well? Can they determine truth better than an insightful observer without the polygraph? My sense of it is that they are at least somewhat better than chance at determining whether a person being tested is telling the truth or not if that person believes that polygraphs work. Is this even true?