How Are Paranormal Test Odds Calculated?

[bmillsap]

Another possible difficulty with that test: Over the course of nearly 100,000,000 flips, testers might begin to slip up in the protocols.

Nah, you could give them plenty of rest by only working them 8 hours a day and doing a flip every 3 seconds. Only take about 29 years that way.

 

[T'ai Chi]

Luckily, computers don't get tired.

 

[TeaBag420]

In the Paranormal portion of this forum, a test of a young Russian girl by the name of Natasha Demkina was brought up. And this led to me to wonder how Randi, and others, calculate the odds of success for various paranormal tests. Some I can figure out, but the one with Natasha has me stymied.

Basically, there were seven patients. Six had medical conditions which could be verified by an X-ray. The girl was told of the six medical conditions and had to tell which patient had which condition just by looking at them. She got four right.

Were any of the conditions ovarian or testicular cancer? Broken bone or dislocation? Gunshot wound?

 

[JAK]

Unrelated to the X-ray girl:

In another thread, when someone said the Randi test couldn't work for "weak" psi,
...

Has there been any indication of what a "weak psi" actually is? How badly do you need to beat the odds? If it were 2% (like for Natasha), would 4% be sufficient? 10%? 50%? Or is Randi looking for something truly dramatic - like 70% reliability or better (at least 5 out of 7 for Natasha) even with a meager 2% probability?

 

[Vim Razz]

Has there been any indication of what a "weak psi" actually is? How badly do you need to beat the odds? If it were 2% (like for Natasha), would 4% be sufficient? 10%? 50%? Or is Randi looking for something truly dramatic - like 70% reliability or better (at least 5 out of 7 for Natasha) even with a meager 2% probability?

The problem with this kind of reasoning is that by the time you start calculating odds you're already assuming that psi is not an issue. Weak or strong.

In Natasha's case, for example, the test conditions were well within the range of her stated claims. Therefore, if she could actualy do what she said she could do, her "odds" of succeding would not have been 1 in 50 -- they would've been 1 in 1.

If she had gotten 4 correct and left the other 3 blank (i.e.: if she couldn't see with enouch detail to positivly identify certain conditions, and said as much) then that would have been a very promising result and the mater would worth further investigation. If it was simply an issue of "weak psi" then she may get fewer correct, but one would still expect her to identify those she could not determine.

By the time someone appeals to the odds in order to plea their case, that's already a big strike against them.

 

[Vim Razz]

Unrelated to the X-ray girl:

In another thread, when someone said the Randi test couldn't work for "weak" psi, I gave a hypothetical example: Someone who can make a coin come up heads 50.0001% of the time (I imagine correctly predicting the coin flip with a similar percentage would also do) and said that that could be tested, though it'd probably take a while to go through the requisite number of trials.

Anyone know how to calculate the number of trials it would take to test that claim?

I can make a coin come up heads (or tails) about 75% of the time when tossing it repeatedly. It's simply a mater of tossing it with a consistent spin and height.

My point (other then bragging ) is that coin tosses are pseudo-random. No amount of trials are going to be able justify a deviation as slim as 0.0001%. While the idea of using coins is purely hypothetical here it still illustrates, I think, that one needs to be careful -- there is always the possibilty that the result can be influenced by unknown, though purely natural, variables.