From Cecil’s column that pravnik linked to:
Also from Cecil:
You’re the employer. You can screen employees however you like. However, a psychic and a polygraph are not equivalent, and anyone who suggests there are is being silly.
As I understand, polygraphs are a pseudoscience, but they’re part of a larger picture. You’re not just hooked up to a machine and asked questions in a sterile environment, you’re being interrogated. People who know they’re pseudoscience and beat it are also having to lie convincingly to the person interrogating them. Someone who is good at it can play a good-cop/bad-cop routine, act like they got some kind of a reading, whether it’s there or not, and drill on the guy.
Unfortunately, like any of that sort of interrogation, you can easily end up with people thinking at some level that it’s seeing something that’s not really there and end up confessing to things they didn’t really do. Worse, there are times where the actual pseudoscience itself is the flaw, whereby people will fail the test despite being completely honest. I’ve heard plenty of stories of people having jobs requiring a TS with poly and losing it because of failing the poly despite being sqeaky-clean honest people.
It really seems quite odd to me that polygraph tests are inadmissible as evidence in trial in most (all?) states but is a requirement for access to our nations most protected secrets.
I have been the subject of three polygraph exams over the years. They work. And by “they”, I mean the skilled examiner using the machine as a tool. My deceptions were detected by the examiner.
Yes, they can be beat by countermeasures and sociopaths. Yes, they can give false positives. They aren’t some “magical machine”; they are tool, and have proper and improper uses. Under the proper conditions, they can establish reasonable suspicion, but not guilt-beyond-a-reasonable-doubt.
My understanding is that they’re rarely used on suspects because A) they can’t force people to take them and B) the results are I admissible in court.
From what I hear, usually when used by the police it’s done to judge the veracity of witnesses they’re not %100 certain of. I.E. John Smith comes in and says his drinking buddy just confessed to a ten-year-old unsolved murder.
“Ok, let’s get Mr. Smith on a polygraph and if he passes it, let’s reopen the case.”
I remember reading about the most effective way to use a polygraph. Sorry, can’t cite the source, but it makes sense.
You set up like any ordinary polygraph test, with boring questions, etc., etc.
Then you casually disclose a key item or object from the crime scene, which only the guilty party would know is significant in the case. E.g., someone steps aside, revealing the log splitter, with no fanfare.
Of course, the guilty party sees the unexpected but hugely significant item. Do you think that might register? I do!
No doubt most interrogations don’t lend themselves to the ideal setup.
In any case, they’re not admissible as evidence in US courts. As Ibn W says, they’re more often used to help weed out the noise and get pointed on the right track. I also suspect that they’re more often demanded by accused parties (over their lawyer’s objections, no doubt) than by the police. (Why would lawyers object? They don’t like any question being asked that they don’t already know the answer to!)
I guess in a way it’s no different than not hiring you because you had a typo on your cover letter. Does that guarantee, 100%, that the candidate has poor attention to details, or perhaps he simply missed that one small typo?
He was 10 minutes late for the interview, so no chance for him because he is a slacker? Or perhaps his car battery really did die that morning and he needed to wake up his neighbor to get a jump start.
Neither of those two scenarios are guarantees, but they are good enough for many job interviewers.
Unlikely. Honestly they dont even bother with the readings. They have decided up front (and based upon the questioning, but not the polygraph) whether or not youre guilty. They then say “Well, the Polygraph sez youre lying- look here (whereupon they can point to any weird reading, even if it doesnt correspond to your answer at that time)- this sez youre lying.” Whereupon you confess.
Bullshit! busted this and so have many others. Those guys are just good cold readers and decent interigators- they use “the machine” to show you what they have decided, readings or no.
In fact, they often do a primitive card trick at the start of a multi-employee reading. Yes, they know who has the Queen of Hearts- *it’s a card trick. * If they have to resort to cheap card tricks, how much science can there be?
Whereas the Mythbusters confirmed it. Of the two shows, I’d give far more weight to the Mythbusters.
Above either, I’d go with Cecil, and above that is Google Scholar, where you can look through actual research which has been done on the subject.
Now if you look through Google scholar, you’ll find that there are (simplistically speaking) studies which prove that polygraphs work and there are studies which prove that they don’t. That could either mean that the results are indeterminate (i.e. sometimes you’ll get a positive result and sometimes you won’t) OR it means that the methodology of the studies causes different results.
That I’ve seen, there are two key findings in the literature:
-
Lie detectors are wholly ineffective unless a person fears the outcome of the test. Telling lies doesn’t cause people to become so ashamed of themselves that they demonstrate measurable signs of stress. Fear of getting caught and punished causes people to become stressed, whenever any question comes up which, if they answered honestly, would lead to them being punished. The vast majority of lie detector tests fail to find any results because they expect the machines to detect lies, rather than to detect stress. The name “lie detector” for a polygraph is a misnomer. It should be “stress detector”.
-
It is amazingly easy to beat a polygraph. Simple training in a few easily accomplished techniques can completely destroy the effectiveness of the polygraph.
In the real world, there is the real threat of punishment if you are guilty of the crime that you’re under the polygraph for, so while item #1 is relevant to the testing of polygraphs, it isn’t relevant to the usage of polygraphs as a tool.
And, in the real world, the grand majority of criminals are uneducated morons. They have almost certainly not researched methods for beating a polygraph, nor have they practiced them. So item #2 is also not relevant.
In the real world, it’s almost certain that polygraphs are highly effective.
Where they break down is with the following three classes of people (one can infer):
-
Anyone who feels no guilt in their crime or concern for their future - sociopaths, psychopaths, etc. - are liable to pass the test with no issue because they don’t get stressed out about the situation.
-
Anyone who is simply a worrywart is liable to fail the test because, even though they didn’t do anything wrong, they know which questions would reveal guilt and become stressed just by hearing such questions. They’re worried what their body might show that they’re lying even when they’re not, since they can’t control their body. And of course, that’s just what happens.
-
Anyone who is reasonably intelligent could probably come up with a few potential methods to fool the polygraph, if not research a couple beforehand.
In prosecuting crime, everything should come down to odds. I should be able to say that, unless someone has utilized a method to beat the polygraph, the test will give a valid result ~85% of the time. (The other 15% of the time, the person is either a sociopath or a worrywart.) Though if they have an IQ higher than 100, there’s a 50% chance that they’ve utilized a method to beat it, and if they’re above 120 IQ then they almost certainly would have. In all of those cases, a polygraph showing someone as guilty is only false in the case of unintelligent worrywarts, which might only be 4% of all cases tested. The other 96% of guilty results, the person is genuinely guilty.
Compared to fingerprint evidence or DNA evidence, polygraph results aren’t very good, but paired with even one more item of evidence, and the odds that the accused isn’t guilty become impressively low.
You might ask this about any “science”
My mother has been diagnoses over her life with schizophrenia, bipolar, unipolor, anxiety disorder, she was also classed by John Hopkins Hospital as a basically a malingerer.
And these are not cheap run of the mill places she’s been taken to. If you can get medical doctors that can’t even figure out if a person is mentally ill or not, well basically it means, at least to me, there is a lot of room for interpretation of data.
As others have said, it doesn’t have to work as long as others think it does. The same way you can con a confession out of someone. It’s the oldest trick in the world to pretend you know something you don’t and the person will feed you the rest, even though you knew nothing to begin with.
Yet despite how old this is, it more often than not works.
The problem is that most Polygraph specialists are basically the equivilant of Psychic charletans. But instead of using a crystal ball or tarot cards they use a machine. The Mythbusters are scientists- scientists are easily fooled by conmen and such (Ask the Amazing Randi) . P&T are stage magicians themselves, they see thru the scam.
Let me make this clear one more time. Polygraph operators are usually not reading the data on the machines and making their determinations that way. They either have decided in advance who is guilty (They are told “We think this guy is guilty”) or they do a cold reading combined with pretty good interrogation procedures. I can tell you they are usually very good at both (I am a trained questioner myself, and those dudes are very good).
Then they say “The machine says you lied here”. They can point to any spike, even one where the meters slipped or the suspect sneezed.* It doesnt matter. *
Then you confess. Often they are right, you are guilty. But even if not, a innocent person will quite often confess if coaxed properly. Watch how the “operator” does it in the Bullshit! episode.
There is absolutely no science at all involved. The collander hooked up to the copier gag is just as scientific. It’s all a scam.
Now, mind you- there could be science involved. Those readings can often tell if someone is stressed. But dudes get stressed when they are asked incriminating questions, and when they are hooked up to a so-called “lie detector”. Which is why the operators have learned to pretty much ignore the readings entirely.
If that was true:
-
The computer wouldn’t be able to spit out a confidence rating. So far as I have seen, modern polygraphs do so.
-
There wouldn’t be any difference between testing people when there is punishment involved or no punishment involved. Interrogation is mostly based on confusing people and trying to catch them in lies, so what happens to you after the interrogation shouldn’t have an effect.
And even if it was true, so what? If it’s an effective technique, then it’s an effective technique.
I want to recommend two excellent books here.
One was recommended by a guy who I think is a doper, he certainly used to be (cause I went to a course with him and he recommended this message board cause of goat felching) called Ian Rowland. He wrote a book about cold reading (which I also recommend, but it’s not so relevant here) and it recommends a book called something like “police criminal interviews and interrogations” and goes through the Reid technique etc. That seems to have been alluded to above but thought I should spell it out. By the way the Reid technique is totally illegal in the UK under PACE…
The other book I recommend is called something like “Detecting Lies and Deceit” and has an author whose either first name or surname is Vrij. That goes through a lot of the evidence pro and anti polygraph and in any case is well worth a read for a number of other reasons!
My understanding is that polygraphs have little to do with “detecting” lies in a scientific manner. It has to do with using human psychology to get someone who is being interrogated to slip up and reveal something that’s being hidden.
In the most simple form of this interaction, it can take the form of the interrogator implying that the machine has detected a lie and thus tricking the interrogatee into confessing.
What it all really comes down to is — never talk to the cops.
Please note that never talk to the cops should be modified in the UK to never talk to the cops without legal advice. There are plenty of situations where it is very much in your interest to talk to the cops, and quite possibly even coming clean. Of course you shouldn’t ever do that without having first had your solicitor say so. For example, you cannot get given a caution without admitting the offence to police before they take you to court. Furthermore, while you have an absolute right to remain silent, the court is permitted to draw a negative inference if you do not mention when charged something that you later rely on in your defence.
If someone confesses to something he didnt do- then it’s wrong. Or if the “operator” decides someone is lying when they really arent, they could lose their job.
I found them on Amazon if anyone is interested:
His doper name is Ianzin, and he can often be found commenting on magic tricks, cold readers, stc.
Both of these statements are true; however my unimpeachable sources (like Forensic Files) suggest that they get a fair amount of use by cops to help eliminate suspects, and to confirm who needs further investigation.
Many years ago I had to take a polygraph (for a low-paying summer security guard job) and out of general nervousness I swear I must have “blown ink all over the walls” (as the saying goes). I think the needle was jumping with everything they asked me (“is your address correct?” “are you breathing?”) because I was offered the job anyway.
If there’s a 90% confidence rating, then 10% of the time, it’s wrong. Humans are stupid and take 90% to mean “guilty”. That’s a problem with how humans work, not with the machine.