On the other hand there isn’t anything which is 100% effective–so the relevant question is how does its effectiveness compare with the alternatives available. But definitely it should be used in combination with multiple other tools–and not something which is going to be the only thing relied on.
It sounds a bit like Voodoo in that the degree to which you believe it is the degree to which it works. If your worried that the Polygraph will catch you lying then you are going to be nervous about the questions you are lying about, which will show up on the test. Which is why in my understanding many polygraphers start out by doing a simple guess the card magic trick with the subject to “demonstrate” the effectiveness of the machine.
The problem that I see with them is the following. Suppose your wife accuses you of cheating and demands that you take a polygraph test. Even if you’re innocent any question that comes up related to her accusations of how you cheated are going to make you nervous because you know that those are the ones that count.
A lot of the supposed accuracy is dependent on having a skilled operator too, which I read as meaning the operator is making judgment calls based on his experience (meaning his gut.) That sort of thing is ripe with opportunity for discrimination. A few bucks cross hands, some undesirables fail their polygraphs, who’s gonna know? They weren’t hired, or they were fired, because they failed a lie detector test.
At least in theory, the idea is that the operator can tell the difference between “baseline no stress” “honest answer to stressful question” and “stress resulting from deceit.” That’s part of why they ask so many questions. By starting with baseline questions (to which the answers are known) the person giving the test should be able to see how your chart reacts.
The baseline questions do include some that generally produce stress responses even when you’re honest, for example questions like “Are you lying to me now?” or “Do you intend to lie to me?” produce stress that would be equivalent to the questions about infidelity you’re thinking of. Then they see how you react to additional questions where the answers are not known. If you show the same signals as the baseline question, then it can be assumed that you answered truthfully. Remember, too, that they’ll ask numerous questions in numerous ways.
It won’t just be one question about infidelity. They’ll ask if you’ve thought about infidelity, if you’ve kissed, if you’ve had oral sex, if you paid for sex, if you slept with co-workers, if you trawled bars, etc. If you’re honestly faithful, then all those questions will show the same baseline response (stressed but honest). If you show that baseline on every question except that, if your chart spikes only on questions about oral sex and co-workers… well, why is it that those were more stressful than other methods of infidelity? Maybe because you got a blow job from the secretary at last year’s Christmas part? The person asking questions will see those stress marks and ask follow-up questions to help clarify where the stress is strongest.
Obviously, the evidence shows that this is not a perfect science, with an least an 8% error rate even in the average population and several methods that can make it even less reliable. Still, 92% accuracy is not half bad if it’s used as a piece of an overall investigation and not just as a stand-alone test.
Reported.