Telepathy

[King of the Americas]

I had suggested that I would replace the card and reshuffle the deck between each pick. I suggested doing this ten times over twenty minutes.

KotA wants each card to take 15 minutes, which would make ten trials take two and a half hours. He also wants a hit rate of 20% which would, to my understanding, require far more than 10 trials to confidently distinguish it from chance.

These are huge hurdles. I don't want to force him to do something he doesn't think he can pass. But the test has to be manageable for me.

I would appreciate any input regarding strengthening this protocol.

I'll do the test however you'd like, but I think 'passing' should be a matter of of overcoming statical probability, not demonstrating perfection.

One correct response would not be enough, but two, within a test of 10...I think that'd be impressive, 3 much much more so. 4 would be shocking.

5 or above, and I think Randi should be ponying up cash...

 

[theprestige]

First, the MDC has been withdrawn. There will be no cash..

Second, it's not a question of what you think is impressive. It's a question of what result actually beats the calculated odds.

 

[Startz]

I'll do the test however you'd like, but I think 'passing' should be a matter of of overcoming statical probability, not demonstrating perfection.

One correct response would not be enough, but two, within a test of 10...I think that'd be impressive, 3 much much more so. 4 would be shocking.

5 or above, and I think Randi should be ponying up cash...

Two might be mildly impressive, but it wouldn't be really strong evidence. Three or more has about one chance in a thousand. That's about what they used to use for Randi's preliminary test. So getting 3 or more would be a good test.

Would you feel comfortable going for Loss Leader's pick 10 out of of the deck, with success being three or more right?

 

[Startz]

I think 'passing' should be a matter of of overcoming statical probability, not demonstrating perfection.

You've mentioned that you are a graduate student. Are you taking a quantitative methods class? It might be fun to bring this problem in front of the class for discussion.

 

[arthwollipot]

Thanks for your interest, and no worries, I'm good to go.

What's the worsted thing that could happen?

The worsted thing that could happen is that you do the test, fail, and then complain that the test wasn't fair.

It's very important for you to agree that it is a fair test from the outset. If you don't think it's a fair test, let's get that out now, so that the test can be redesigned so that it is fair. Let's do that before you do the test.

 

[Loss Leader]

I'll do the test however you'd like, but I think 'passing' should be a matter of of overcoming statical probability, not demonstrating perfection.

One correct response would not be enough, but two, within a test of 10...I think that'd be impressive, 3 much much more so. 4 would be shocking.

5 or above, and I think Randi should be ponying up cash...

I absolutely don't want to ask you to perform to a standard that you don't think you can meet. If a card is picked at random from a 52 card deck, how often will you be able to guess it correctly?

I don't think I can spend two and a half hours staring at cards. Logistically, it's not possible for me to carve out such a large block of time. I'd like to find something you're comfortable with that has a very small success rate by pure chance. 2 cards out of 10 is not a hit rate high enough to declare success in any meaningful way from just 10 trials.

 

[jt512]

I absolutely don't want to ask you to perform to a standard that you don't think you can meet. If a card is picked at random from a 52 card deck, how often will you be able to guess it correctly?

Yes, we want to design the test so that it is fair to KotA, meaning that if s/he has the ability s/he claims, the test will have a high probability of confirming his/her claim. Simultaneously, the test must also have a high probability of not concluding that KotA has the claimed ability if he/she in fact does not. As I explained in detail in a post a couple pages back, the key piece of information needed to design the test to meet both goals is the answer to the highlighted question above.

 

[LongFuzzy]

If a card is picked at random from a 52 card deck, how often will you be able to guess identify it correctly?

He must not guess!

 

[Loss Leader]

Yes, we want to design the test so that it is fair to KotA, meaning that if s/he has the ability s/he claims, the test will have a high probability of confirming his/her claim. Simultaneously, the test must also have a high probability of not concluding that KotA has the claimed ability if he/she in fact does not. As I explained in detail in a post a couple pages back, the key piece of information needed to design the test to meet both goals is the answer to the highlighted question above.

I would be satisfied with a 95% confidence that the result is not the product of chance.

KotA has said that he thinks he can get 2 out of 10 (each card representing a distinct trial wherein it is returned to the deck and reshuffled).

 

[jt512]

I would be satisfied with a 95% confidence that the result is not the product of chance.

A 5% false positive rate (FPR) doesn't make for a very conclusive test if the result is positive; a 1-in-20 chance of a positive result being false is not even suggestive that the claim is real. To me, it would seem pointless to conduct the test.

IIRC, the FPR for preliminary tests for the MDC was set at a minimum of 1/5000. And preliminary tests conducted by the Independent Investigations Group have set the FPR orders of magnitude lower when doing so does not make the number of required trials too large.

From a Bayesian perspective, the lower the prior probability of a claim, the more likely it is for a positive test result to be a false positive. The prior probability for a claim of telepathy is very small. Thus, for a positive result to have a reasonable likelihood of indicating that the claim is true, the test's FPR must be made very small.

Finally, keep in mind that there is no harm to the claimant by setting the FPR to a small value, because for any FPR we can attain any desired power of the test by setting the number of trials n appropriately. Only if n becomes inordinately large is there any reason to even think about softening the FPR.

 

[King of the Americas]

First, the MDC has been withdrawn. There will be no cash..

Second, it's not a question of what you think is impressive. It's a question of what result actually beats the calculated odds.

Honestly, I don't believe there ever was any actual cash offer...

Beating calculated odds is exactly what I propose should be a win, not a demand for perfection.

 

[King of the Americas]

Two might be mildly impressive, but it wouldn't be really strong evidence. Three or more has about one chance in a thousand. That's about what they used to use for Randi's preliminary test. So getting 3 or more would be a good test.

Would you feel comfortable going for Loss Leader's pick 10 out of of the deck, with success being three or more right?

A complete test would include 10 cards, I think. But I'm in favor of breaking that up, 3 each go...4 on the last day.

 

[King of the Americas]

You've mentioned that you are a graduate student. Are you taking a quantitative methods class? It might be fun to bring this problem in front of the class for discussion.

I took one on research methods and surveys. We covered randomizing your participants, bell curves, the missing p-hat, and both qualitative and quantitative measuring. A slightly different rose perhaps.

 

[King of the Americas]

Yes, we want to design the test so that it is fair to KotA, meaning that if s/he has the ability s/he claims, the test will have a high probability of confirming his/her claim. Simultaneously, the test must also have a high probability of not concluding that KotA has the claimed ability if he/she in fact does not. As I explained in detail in a post a couple pages back, the key piece of information needed to design the test to meet both goals is the answer to the highlighted question above.

0= no ability

1= A little luck, or tiny bit of ability

2= Really lucky, or something bumping the lowest end of measurability is happening

3= Go to Vegas

4 or more= ...ummm...

Ya know, suddenly, I think this is a really bad idea.

I mean, 'thinking' this is a possibility, is one thing, but verifying it as a real thing on a public forum seems..."poorly considered"...

This is not me backing out, I just think it's a bad idea, from which absolutely nothing good could come.

I mean, if I were to get 10 out of 10...that would be totally horked up. Such a capability is X-Men comic book stuff. I don't want that as my life, in part or whole. I think I'm really okay with not having any such ability.

Will someone talk me out of this horrible idea, please?

 

[Pixel42]

You've indicated that you don't have much free time, KotA. Whilst using some of it to repeat an experiment that's been done many times before may well be a good use of it, is there not a more interesting one you could try for yourself? Newton's experiment of splitting a light beam with a prism, say.

It's just that your chances of getting a different result to the ones obtained by everyone else who has ever done an experiment is so small that you might as well pick one whose results are at least pretty.

 

[King of the Americas]

You've indicated that you don't have much free time, KotA. Whilst using some of it to repeat an experiment that's been done many times before may well be a good use of it, is there not a more interesting one you could try for yourself? Newton's experiment of splitting a light beam with a prism, say.

It's just that your chances of getting a different result to the ones obtained by everyone else who has ever done an experiment is so small that you might as well pick one whose results are at least pretty.

Beauty is in who's eyes...?

 

[Startz]

0= no ability

Will someone talk me out of this horrible idea, please?

King,

It's put up or shut up time. If you think you have an ability, tell us what you think a fair test would be. (If you need help with the statistics, several people here can help you.) If you are another along the long list of people who talk about being tested, and then start changing the topic when someone steps up to test them...well that's your privilege I guess.

 

[Loss Leader]

A 5% false positive rate (FPR) doesn't make for a very conclusive test if the result is positive; a 1-in-20 chance of a positive result being false is not even suggestive that the claim is real. To me, it would seem pointless to conduct the test.

IIRC, the FPR for preliminary tests for the MDC was set at a minimum of 1/5000.

Well, don't I feel like a fool. If you say 1/5000 then 1/5000 it is.

0= no ability

1= A little luck, or tiny bit of ability

2= Really lucky, or something bumping the lowest end of measurability is happening

3= Go to Vegas

4 or more= ...ummm...

This does not seem to be statistically sound.

Ya know, suddenly, I think this is a really bad idea.

I mean, 'thinking' this is a possibility, is one thing, but verifying it as a real thing on a public forum seems..."poorly considered"...

This is not me backing out, I just think it's a bad idea, from which absolutely nothing good could come.

I mean, if I were to get 10 out of 10...that would be totally horked up. Such a capability is X-Men comic book stuff. I don't want that as my life, in part or whole. I think I'm really okay with not having any such ability.

Will someone talk me out of this horrible idea, please?

Let me make this completely clear: If you do not want to do a test, then you shouldn't do a test. I wouldn't want you to perform in any way if you have no desire to succeed. As far as I'm concerned, no test should be conducted until you agree that it is something you can do and that you want to do.

Be advised, however, that nobody on this forum will allow you to claim you have any sort of extrasensory power without reminding you that you refused to test it. You are backing out, whether you call it that or not. And something that lacks evidence simply cannot be said to exist.

I wish you the best.

 

[Dave Rogers]

Actually, it may be much better.

[analysis snipped]

If I've done this right, then there is a 40% or better chance of picking a card in the selected set, with or without replacement.

Yeah, you're right; I realised I was solving the wrong problem. Still, 40% is more then 25%, so I've got an excuse

Dave

 

[King of the Americas]

King,

It's put up or shut up time. If you think you have an ability, tell us what you think a fair test would be. (If you need help with the statistics, several people here can help you.) If you are another along the long list of people who talk about being tested, and then start changing the topic when someone steps up to test them...well that's your privilege I guess.

I said I would so I will, but I don't want to anymore...

If I do well, or bad, the ultimate result will be unpleasant, especially the better I do!