Even if the polygraph did work, the use of “little white lie” questions would render it worse than useless. A great many people, for instance, have not, in fact, cheated on tests. If such a person answers “no” to that question, it will be the truth. But if the examiner assumes that it’s a lie, then when other questions lead to responses that look the same way, then the examiner would assume that those are lies, too. So the net effect will be that the polygraph test is skewing you towards hiring the people who cheat on tests, and away from those who don’t (or any of those other “test of character” questions).
Leo Bloom can be accused of argumentum ad chicken fucker, which is equally valid, in his rhetoric, as argumentum ad hockey player, which his esteemed interlocutor has chosen.
They are equal.
He sees no difference between his posts and that in the longer reply of his interlocutor.
Just tell the truth on the polygraph is all he’s saying. Ain’t America grand?
You are right that this is one of the reasons that lie detectors are worthless. In theory, if no one lies on any of the questions, there is no difference in the metabolic response between the non-probative neutral questions, the non-probative accusatory questions, and the probative questions. In theory, this should mean that the answers to all those questions have the exact same metabolic response and that the test should be “inconclusive.” There is no differential metabolic response to distinguish between honest answers and dishonest ones.
In the real world, a truly honest person won’t lie on any of the questions, but they will react more strongly to the “have you ever cheated on a test” non-probative question because it comes across as a nerve-inducing accusation, which might really bother an honest person. This will appear like a low level metabolic response to a lie-inducing non-probative question. They will react exactly the same way or maybe a lot more strongly to “Did you kill your husband?” Then the examiner will brand them a liar on the probative question even though it’s a completely bullshit interpretation of the data they are seeing.
There are many other ways the test is flawed. Why should the examiner be the one to interpret the readings? It should be a person who can’t hear the questions or answers but only has a time log of when questions were asked and when the answers were given. This would be at least a “single blind” test. That person should be able to readily distinguish between the lie-inducing non-probative questions and the non-lie inducing probative questions but I bet they would fail at that completely. If they couldn’t at least do that without knowing the questions, they would need to admit that their interpretation of the rest of the questions is worthless. As a practical matter, the examiner might need to see the polygraph data during the questioning because the subject’s metabolic processes have to return to his or her baseline before the next question gets asked, but that doesn’t mean the examiner must be the one to interpret the data. In fact, there is no reason that the person administering the exam needs any more than three pieces of information: (1) When was the question completely asked? (2) When did the subject complete his or her answer (which for yes/no questions doesn’t take long)? and (3) Has the subject’s metabolic response gone back to normal so they are ready for the next question? They don’t have to be in the same room as the subject to get this information. A neutral “proctor” could signal to a distant polygraph examiner when the first two things happen and the examiner can use those times and the polygraph data to conclude when the next question could begin. They could then signal the questioner to start the next question.
The examiner is supposed to use a neutral tone of voice to ask all the questions but the examiner could use a subtly accusatory tone when asking the damning probative questions and perhaps even betray a suspicious face. That would make the subject more nervous during those questions and elicit the polygraph appearance of a lie. Why not have a computer generated voice ask the questions? The examiner doesn’t need to be in the same room to get the data from the exam. The neutral proctor could watch for things like whether the subject is tampering with the polygraph sensors.
There is no reason to believe that people’s metabolic responses remain exactly the same over the course of a long interview, so the answers to early questions might appear more or less like lies than the answers to later questions. Maybe people get less nervous as time goes on. What if all the accusatory probative questions are asked early in the exam?
I have also seen suggestions that psychopaths don’t have the same metabolic responses to lying that the general population has. That means pyschopaths are essentially immune to polygraphs. Who do you think commits a disproportionate number of the crimes that people think lie detectors should be used for?
Lie detectors are well and truly worthless.
So you’re telling me they’re about as useful as an ADE 651bomb detector then?