Merged Telepathy test: which number did I write?

[Michel H]

...If you insist on rating answers ... for 'credibility', then you would need to introduce blinding so that you do not know which answers are right or wrong. If you want to do such a test, I am willing to be the person to whom everyone sends their answers, once a protocol is established...

Very good, I take good note of your suggestion, and of your kind availability, Agatha.

 

[Michel H]

Michael, I'm not quite clear on whether we are being asked if we know the answer or if we're being asked to guess the answer to see if we know it without being aware that we do.

I'm assuming it's the former since the latter would render the "don't know" option obsolete.

So my vote is "I don't know"

Nobody knows. We can't hear you. Telepathy is not real.

"You're not quite clear"...You mean ... "You don't clearly understand" ? Well, let me try to explain it to you, then. When you do such a test, you have to be careful about the words you use...I am not asking you if you know the number I wrote. I am asking you to guess it, to divine it. You may provide an answer, even if you have a doubt. And you also have the additional option to answer "I don't know", like you just did. As for me, I also think that I have the freedom to ask myself if you told me the truth, or if you lied, and tried to deceive me. These things do happen in many places, unfortunately. Look at what is happening with chemical weapons in Syria now...Who used them? Not everybody seems to agree.

You say "Safely Ignored" about yourself (under your name, and above your avatar). So, then, perhaps it would be safer for me to ignore this post of yours.

You sound a little aggressive, Jack (in this post)... But maybe I think that because I didn't take my medication today .

 

[abaddon]

"But it gives a 20% probability to each of the answers that you have defined as valid."

No, it doesn't. Each of the four numbers has a probability equal to .25 if there is no telepathy, and chance only is responsible for the results. This means that, if I get 40 answers, and there is no telepathy, I should expect about 10 (10=.25x40) ones, about 10 twos, about 10 threes and about 10 fours, the four numerical outcomes (1, 2, 3 and 4) are "equiprobable". Since p1=p2=p3=p4=p (the four numbers are "on the same footing" when there is no telepathy), and p1+p2+p3+p4=1, then, necessarily p=.25 (not .20). The "probability" of answering "I don't know" ,when there is no telepathy, is not known, it is probably equal to neither .25, nor .20 and will probably vary, from one forum to another.

"Unless you are shifting your goalposts and are now excluding "I don't know" from any results you may glean."

No, I don't want to do that. I want to read and study all your answers, expressed in a free way. Look at the analysis of my previous test (ninth line):
http://www.internationalskeptics.com/forums/showthread.php?postid=8607740#post8607740 . Did I ignore people who answered "I don't know"? No, I didn't.

Let me get this clear. You have defined five responses as valid, and are flat out stating that you will ignore one of these responses that you yourself defined? At best this is faulty analysis, but I can think of a lot less favourable descriptions.

 

[abaddon]

"But it gives a 20% probability to each of the answers that you have defined as valid."

NoYes, it doesn't. Each of the four five numbers answers has a probability equal to .25 .20 if there is no telepathy, and chance only is responsible for the results. This means that, if I get 40 answers, and there is no telepathy, I should expect about 10 (10=.25x40)8 (8 = .20x40) ones, about 10 8 twos, about 10 8 threes and about 10 8 fours, and 8 IDKs the four numerical outcomes (1, 2, 3 and 4)+ the one non numerical one you defined are "equiprobable". Since p1=p2=p3=p4=pIDK=p (the four numbers and the non numerical answer are "on the same footing" when there is no telepathy), and p1+p2+p3+p4+pIDK=1, then, necessarily p=.25 (not .20). The "probability" of answering "I don't know" ,when there is no telepathy, is not known, it is probably equal to neither .25, nor .20 and will probably vary, from one forum to another.is .20

"Unless you are shifting your goalposts and are now excluding "I don't know" from any results you may glean."

No, I don't want to do that.I am doing exactly that. I want to read and study all your answers, expressed in a free way. Look at the analysis of my previous test (ninth line):
http://www.internationalskeptics.com/forums/showthread.php?postid=8607740#post8607740 . Did I ignore people who answered "I don't know"? No, I didn't.No, you simply did not count them.

FTFY

 

[Michel H]

Let me get this clear. You have defined five responses as valid, and are flat out stating that you will ignore one of these responses that you yourself defined? At best this is faulty analysis, but I can think of a lot less favourable descriptions.

No, I don't want to ignore the "I don't know" answers (or any other answer) in this test, abaddon. I wonder what makes you think that, right after I said exactly the opposite. I insist that everybody can answer freely one of the five possibilities, in this test, including "I don't know".

Also, I wonder what makes you think that the probability of answering "I don't know" should be exactly equal to the probability of answering "1" for exemple, that's somewhat mysterious to me.

There is, however, one thing that you said which is true: in my analysis of the previous test, I didn't count all the "I don't know" answers, I just mentioned that that there were several of them (I did not try to assess their credibilities either). However, counting them, or trying to assess their credibilities, can still be done, if this is of any real interest.

 

[abaddon]

No, I don't want to ignore the "I don't know" answers (or any other answer) in this test, abaddon. I wonder what makes you think that, right after I said exactly the opposite. I insist that everybody can answer freely one of the five possibilities, in this test, including "I don't know".

If there are five equi-probable answers then the probability of each is .20 and the only way you can get .25 is by eliminating the IDK answer from your "analysis".

 

[Michel H]

If there are five equi-probable answers then the probability of each is .20 and the only way you can get .25 is by eliminating the IDK answer from your "analysis".

abbadon, I have now edited my post #385, you might want to take a look.

The five answers (yes, there are five of them) are not equi-probable, only the numbers are (equi-probable).

 

[Aridas]

abbadon, I have now edited my post #385, you might want to take a look.

The five answers (yes, there are five of them) are not equi-probable, only the numbers are (equi-probable).

In short, 4x + y = 100%. In that case, characterizing x as either 20% or 25% would be a mistake, in short.

I'm a bit tired, so I'm not going to directly offer input, other than that I find this particular tangent to be rather boring.

 

[Jack by the hedge]

"You're not quite clear"...You mean ... "You don't clearly understand" ?

Yes. I mean that.

Well, let me try to explain it to you, then. When you do such a test, you have to be careful about the words you use...I am not asking you if you know the number I wrote. I am asking you to guess it, to divine it. You may provide an answer, even if you have a doubt. And you also have the additional option to answer "I don't know", like you just did. As for me, I also think that I have the freedom to ask myself if you told me the truth, or if you lied, and tried to deceive me. These things do happen in many places, unfortunately. Look at what is happening with chemical weapons in Syria now...Who used them? Not everybody seems to agree.

So you are trying to test to see if we may in fact know the answer subconsciously while being unaware of it. In that case what is the purpose of allowing a "don't know" response when you don't care if we know or not?

Had I appreciated that "don't know" would merely be treated as a refusal to guess, I would have guessed a number instead.

So I change my guess to 3.

You say "Safely Ignored" about yourself (under your name, and above your avatar). So, then, perhaps it would be safer for me to ignore this post of yours.

What if it's a bluff? It might be a trick. Or it might just be a joke. A smartass remark that once seemed cleverer than it is? An attempt to make you pause and think about whether you might actually be underestimating what I wrote, or alternatively a humble acknowledgement that it's usually nonsense?

Might be safer if you didn't ignore it just because it says you should.

You sound a little aggressive, Jack (in this post)... But maybe I think that because I didn't take my medication today .

Not intended.

The last line was somewhat on the assertive side, but no aggression was intended.

In all seriousness though, if you want to test this idea properly then you need to allow us to vote blindy, without seeing what everyone else voted, and you need to consider those votes blindly, so you don't fool yourself into rejecting or accepting votes for invalid reasons.

As it stands at the moment, the great majority of respondents are just horsing around because they don't take your test remotely seriously. If you take it more seriously yourself, that might change.

 

[dlorde]

I'm curious about how the broadcast telepathy of this claim might work; members of this forum could be anywhere on the planet - so is the telepathy broadcast to everyone on the planet, or somehow targeted to the people reading this thread?

If the former, then I could expect to ask anyone in the street to guess a number between 1-4 and get a statistically significant bias towards the broadcast number; i.e. for the duration of the broadcast, the whole planet is biased to guess the broadcast number. Yes?

If the latter, can broadcast telepathy be sent to any subgroup on the planet without knowing who or where or how many they are? how could that possibly work?

Finally - what is the duration of this telepathic broadcast?

 

[Michel H]

In short, 4x + y = 100%. In that case, characterizing x as either 20% or 25% would be a mistake, in short.

I'm a bit tired, so I'm not going to directly offer input, other than that I find this particular tangent to be rather boring.

You don't have to worry about y, Aridas. One important thing to understand is that, if there are, for exemple, 40 numerical answers (1, 2, 3 or 4), we can expect about 10 1's, 10 2's, 10 3's and 10 4's, when there is no telepathy. If there are strong telepathic effects, these numbers may change, and a probability calculation (calculation of the so-called p-value) might be done.

 

[Agatha]

abbadon, I have now edited my post #385, you might want to take a look.

The five answers (yes, there are five of them) are not equi-probable, only the numbers are (equi-probable).

The numbers are not equi-probable; as has been explained above people have an inherent bias towards choosing numbers in the middle of a range and avoiding the top and bottom of the range, plus people tend to choose odd numbers rather than even.

 

[AdMan]

So, are you going to attempt a proper blind test, Michel?

Your current protocol is pretty much worthless.

 

[Aridas]

You don't have to worry about y, Aridas. One important thing to understand is that, if there are, for exemple, 40 numerical answers (1, 2, 3 or 4), we can expect about 10 1's, 10 2's, 10 3's and 10 4's, when there is no telepathy. If there are strong telepathic effects, these numbers may change, and a probability calculation (calculation of the so-called p-value) might be done.

I'm not worrying, but I would suggest not including y in the first place if you're trying to deal with telepathy on a subconscious level.

That said, this may start to be reasonable if blinds are applied AND there's a sufficiently large number of participants AND there's a number of tests, though. A form like this, where posters just post the answers and can see each other's responses isn't a particularly viable way to handle it.

 

[Michel H]

...
A form like this, where posters just post the answers and can see each other's responses isn't a particularly viable way to handle it.

Well, Aridas, in scientific research, it is very (very, very, very) important to be able to make the "right approximation". If you keep insisting on no less than the perfect calculation, the practical consequence on this is that you (or me) might remain an idiot all your life (or my life) . Which I don't want to do, if I have a choice. Don't demand too much .

 

[elbe]

As far as I know, there was no "confirmation bias" in my analysis of September 2012, although there was a small(?) difficulty, related to the fact that I knew if the answers were numerically correct (or not) when I assigned their various CR's. But I explained why I chose these credibilities. I didn't say:"Well, my number to guess was 2, and this person answered 2, so this person is obviously nice and very good, and I can only give her a high credibility rating." Such an approach would not be serious.
The "confirmation bias" is something different, it's a tendency by some people to pay attention to events that seem to confirm their ideas, and to ignore events that do not seem to confirm their ideas, see http://en.wikipedia.org/wiki/Confirmation_bias . I did not ignore incorrect answers, I analyzed them like the others, I studied all 13 valid answers, not just the 3 which were correct. If I had retained in my study of the results only the three correct answers, and added: "Well, as to the nine other answers, they were not good, so I don't want to discuss these, feeling a little tired now, it's late- bye bye- end of this analysis, that would have been a confirmation bias, I suppose, but I didn't do that.

Your previous knowledge of which answer "correct" is entirely likely to influence your analysis of the results to reach the conclusion you're already biased to. You may not be consciously aware of it but everyone reading your "analysis" can see it clear as day. There is a reason that proper testing protocols are blinded and you had member offering to assist you in making your little test better. Strangely you appear to have ignored their advice and all the complaints about your protocol and just repeated your highly flawed test.

 

[Michel H]

The numbers are not equi-probable; as has been explained above people have an inherent bias towards choosing numbers in the middle of a range and avoiding the top and bottom of the range, plus people tend to choose odd numbers rather than even.

There is certainly some truth in what you say here, Agatha. Actually, assuming that all four numbers have probabilities equal to 1/4 (when there is no telepathy) is only a working approximation, a working assumption, which might, however, be considered reasonable (it's not clear what else one could assume).

 

[elbe]

There is certainly some truth in what you say here, Agatha. Actually, assuming that all four numbers have probabilities equal to 1/4 (when there is no telepathy) is only a working approximation, a working assumption, which might, however, be considered reasonable (it's not clear what else one could assume).

I don't think you can get around that humans are really, really bad at picking random numbers and we'll always learn heavily towards certain numbers. If you really wanted to use numbers I might suggest choosing X number of non-random, non sequential numbers to have the subjects choose from (i.e. {28, 48, 68, 88}), or, what I'd do, choose something else, like colors (red, blue, green), objects (ball, lamp, cup), or letters (J, K, L). Until your test is properly blinded, however, it's not a valid study.

 

[Aridas]

Well, Aridas, in scientific research, it is very (very, very, very) important to be able to make the "right approximation".

Statistically speaking?

If you keep insisting on no less than the perfect calculation,

I'm really not, though. Not even close. There are lots of potentially significant interfering factors, though, regardless of whether you can do what you think you can do, which I haven't asked you to account for.

the practical consequence on this is that you (or me) might remain an idiot all your life (or my life) .

No. Not accepting everything is not the same thing as being an idiot, even if that thing actually is the case. Why it is or is not accepted does matter for whether one can reasonably be called an idiot.

Which I don't want to do, if I have a choice. Don't demand too much .

I'm not demanding anything. I'm pointing out some of the base necessities required for a reasonable conclusion.

 

[dlorde]

Well, Aridas, in scientific research, it is very (very, very, very) important to be able to make the "right approximation". If you keep insisting on no less than the perfect calculation, the practical consequence on this is that you (or me) might remain an idiot all your life (or my life) . Which I don't want to do, if I have a choice. Don't demand too much .

It's not a question of the 'perfect' calculation, it's a question of avoiding obvious sources of bias that will render any result meaningless. For example, the biases mentioned by Agatha, and the biases that come when people see other people's choices and who chooses what. Whatever the effect of these biases may be, they will invalidate the results because they are unquantified and unquantifiable. Only an idiot would expect a meaningful result from such an exercise.