Psychics and Random Chance.

[JohnB]

I was reading some of the threads at the James Randi Foundation when I came across this one.

 

In the second post down, Kramer says "The applicant would need to perform significantly above CHANCE results in order to pass the preliminary testing phase of the Challenge. To date, no applicant has passed preliminary testing, and most often, results far below CHANCE are what we actually see."

 

As I understand the concept, if there is no psi ability, then in a sufficiently large sampling we should get a correlation to random chance. As psi ability increases, the correctly perceived results should climb above chance until they theoretically approach unity.

 

Even without any psi ability, we should get ( with a large enough sample ) some readings that are well above chance and some that are well below, with most results clustered around the random chance value. I'm thinking of the standard bell curve here obviously.

 

Kramer says that "most often, results far below CHANCE are what we actually see." How does that work out? It's like the result of a negative psi ability.

 

As we don't know how many people have actually been tested, could the explanation be simply that the sample is not yet large enough to show a full distribution curve, so the results are skewed a bit in one direction?

 

Please note, this isn't about the existence or not of psi ability, I just thought the results were interesting. I would have expected most to come out around the random chance value.

[swansont]

Kramer says that "most often' date=' results far below CHANCE are what we actually see." How does that work out?

[/quote']

 

It may be related to the "dice have no memory" phenomenon. People think that e.g. if a 4 hasn't come up on the dice in a while, that it's "due" and somehow has a higher chance of occurring. That, of course, is false, and one reason why casinos make wads and wads of cash. It may be that the would-be psychics are similarly out-thinking themselves.

[danny8522003]

I think it's exaggerated. If he says that most applicants recieve scores that are below chance, then when someone (through probability) gets a score above chance, their psi ability becomes all the more believable.

[JohnB]

Maybe. To be honest, if you took a thousand test subjects I would expect some of them to show much higher than chance results. Some would be well below and most around the middle.

 

It would not be until after the same individual gets a high score in the second and third round that I would go "Hmmmmm."

 

The other thought is that if most of the results they are getting are well below what they figure as the random probability level, then I think they may have that level set too high.

 

A list of 20 words chosen at random (but sufficiently dissimilar that they can't be confused) out of the entire English language. I'd consider getting even one word right to be bloody amazing.

[CPL.Luke]

couldn't you technically model a dice with memory tendancy?

 

for instance in 6 rounds it is almost assured that you will get a 4, unfortunatly I think that violates the whole independant event thing, but it might work

[alibabba]

couldn't you technically model a dice with memory tendancy?

 

for instance in 6 rounds it is almost assured that you will get a 4' date=' unfortunatly I think that violates the whole independant event thing, but it might work[/quote']

 

Are you referring to one die, or a pair of dice?

 

On one die the chances of rolling a 4 once in six tries is even, isn't it? There are six sides, one of which has a 4 on it, so wouldn't the chances be 50/50? 1:6 pr. roll?

 

with 2 die, there are 3 ways to roll a 4: 2/2--1/3--3/1. Since there are 36 possible combinations, it would seem that there is a 1:12 chance of rolling a 4 with 2 die pr. roll. It would appear that the chances of rolling a 4 in 6 passes with 2 die is only 1/2 that of rolling a 4 with 1 die.

 

Are these numbers correct?

[CPL.Luke]

I'm pretty sure the probability at a high school level goes to 100% , but that there is a better way to do the probability than that, because it is possible to roll a dice an infinite number of times and never hit a 4 (unlikely though)

 

I was reffering to one dice

 

I think my idea violates the whole independant events in statistics though

[swansont]

I was referring to two dice - gambling, i.e shooting craps. 3 ways to get it out of 36, or 0.0833 probability.

 

Die is singular

 

Dice is plural

[Bubba]

On one die the chances of rolling a 4 once in six tries is even, isn't it? There are six sides, one of which has a 4 on it, so wouldn't the chances be 50/50? 1:6 pr. roll?

 

Your spot on with this but this is the expected chance, if six million people each rolled a die you would expect about 1 million to get a four.

The key idea here is that if 5 million people roll and do not throw a 4 then we can't expect the next million to all throw fours (we expect 250,000 to throw fours). The probability dose not depend on what has happened, hence "dice have no memory".

[YT2095]

further to that point, if 5 million peple rolled a die and non produced a 4, then I`de expect a similar result if they did it again with the same die, point being it wouldn`t Increase the odds of a 4, it would show that 4 was Very unlikely!

[Bubba]

Thats not fair, now your introducing a variable, the "uneven die". I suppose I should have specified the dice as all being equal and unweighted.