He also once (
here) accepted an answer, apparently as a hit, that was wrong- for a number between 1 and 10, the answer he accepted was "1" (while the actual correct answer was 10), because 1 is the first digit of 10 and the answer "sounded rather friendly"; while he rejected for his statistical analysis an answer of "3" (because no comment accompanied the guess, and both the number 3 and the poster's name- Red Baron Farms- demonstrated "an element of aggressivity"). He also rejected for inclusion in his analysis an answer of "I don't know," because the comment that accompanied it ignored that Michel had explained how he selected the number. His premise is that
everyone in the world receives his telepathic transmissions; it seems to me that an answer that says "I don't know" would be just as statistically valid, for deciding whether there was such a transmission, as any right or wrong answer. Pollsters who want to see if an electorate is convinced of a particular political position- i.e., if that position has been successfully "transmitted"- don't reject "undecided" in their analysis, because that is, in fact, part of what they want to know. And if you just decide that everyone who answers wrongly or in a way that you don't like is lying, there doesn't seem to be much point in doing the survey at all.
In the example you mention, scarlettinlondon's answer ("1") wasn't a full hit, as I explained, her answer seemed more reliable and was also related to the correct answer, and I felt that this was of interest.
If people answer just "I don't know", it means "I don't know". I think that you should the right to say "I don't know" in all my tests, just like I have the right to consider the possibility you are lying. However, if people reply something like:
"I donnnt know",
the way it is written raises suspicion, the credibility is low.
We don't live in an ideal world, and I have to use the data you provide, try to extract the right conclusion from inevitably imperfect data. This is difficult, and the credibility analysis seems to be an essential tool for this work.
It is important to understand that, even if people are able to read my mind accurately, they will not necessarily fully cooperate in the test (this is a point which, I think, is widely misunderstood by skeptics).
It is necessary to "fight hard" to try to extract the truth from people who are not necessarily eager to reveal it, this is why the analyses seem sometimes somewhat complicated. There is an element of complexity that cannot be completely removed.
One fundamental observation that I made in my tests, conducted over many years and in several languages, was that, when people give the correct answer, they tend to be friendlier and more serious in their answers than when they do not. This aspect seems to be the key element for analysing results. On the other hand, if you ask people, "did I write 1, 2, 3 or 4" or "did I write 2, 3, 4 or 5"", it seems that people will give the correct answer about 25% of the time,
even when they know the correct answer. In other words, it seems that people usually refuse to fully cooperate (though they are some cases of good cooperation), and provide answers with an element of complexity. There is no way an analysis can be straightforward if the answers are complex, this is something which is not taught in books and that you have to accept.
