Re: Statistics Book

Timo A. Nieminen wrote:

On Fri, 7 Sep 2006, glhansen@xxxxxxx wrote:

I didn't mean to seem overly defensive. I'd welcome other input. I
want to consider what I've done so far to be a pilot study, and start
again with a more carefully designed procedure that would answer the
most likely objections. And if someone else wants to collect data
under the same procedure, to share, we can work it out.


I think you need to think about whether you are looking for
(a) telepathic communication of the result of the die roll from you to them, or
(b) remote reading of a die roll.

For the latter, roll the die under a cup, and don't look. For the former, you need to look.

My first thought was that I didn't care if the hits were due to mind reading, remote imaging, precognition of my verbal report, retropsycokinesis, or any other method. The null hypothesis is that it doesn't matter what psychic channel is tested because none of them work.

I will explain my nominal new procedure. All die rolls will be standardized, e.g. with the die in a cup and covered, shaken rapidly up and down seven times, and the die allowed to settle before the cup is returned to the table, because I don't want to have to bang it down before the die stops. Before I call a participant I will roll the die once to determined which experiment, and once to determine a number. So no numbers will be "saved" from a previous attempt (e.g. wasn't home last time I tried), and there'd be no selection bias (sometimes roll again, sometimes not). Each invitation to guess will be scripted. There will be three possible experiments:

1) Roll the die, look at it, and invite the participant to guess the number.

2) Roll the die, keep it covered, tell the participant that I am about to look at it, and invite him to wait a few moments and guess.

3) Roll the die, keep it covered, tell the participant that I will look after he guesses the number, and invite a guess.

And sets of control rolls will be done between calls as a test of die bias. All tests will be done with the same die.

Raw data will be date, time, participant, experiment type, number rolled, number guessed. Or time, date, and ten control rolls.

What can be tested? Comparison of 1 and 2 would be a test of unintentional verbal communication. Comparison of 2 and 3 would be a test of excluding mind reading and remote viewing of the present. Comparison of the die rolls during the trials versus the control runs could be a test of psychokinesis or something.

Two possible results of psi are recognized in the literature: more hits than expected, and fewer hits than expected. Combining them could give expected results if both are operating. Normal random numbers would sometimes give more hits than expected, and fewer hits than expected. But if enough people have abnormally more and fewer hits than expected, then the distribution about the mean would have more results in the wings.

The control rolls would be used to check for bias in the numbers, and I would look for preferred guesses (e.g. I suspect the number "4" would be guessed more often than 1/6 of the time), to remove that type of false significance.

I also want to look for a selection bias in the guesses, e.g. will two numbers be given as guesses two times in succession more or less often than 1/6 of the time? As an example, in the pilot study I rolled a 6 for a friend, and she guessed 6. The next time I called, a week later, I rolled a 6 and she guessed 5, and said that she had thought of 6 but she'd guessed that one last time. I don't know what to do about that as far as making corrections to a measure of "true" ESP and second-guessing oneself, but it seems like it should at least be looked at.

.

Relevant Pages

  • Re: Statistics Book
    ... I have to say that the prospect of unconsiously working out a communication protocal through a scripted request with a network of people seems pretty far out there. ... One where I go through the scripted request while looking at the die roll. ... One where I keep the die roll covered, and inform the friend that I am about to look at the number and think about it. ...
    (sci.physics)
  • Looking for feedback on coding style
    ... from more experienced programmers on a few issues regarding efficient, ... where you can triple your money every time you roll heads (and lose it ... The Roller class simulates rolling a die until all sides have come up ... I never instantiate a Roller ...
    (microsoft.public.dotnet.languages.csharp)
  • Re: ID Advocates - ID is Science, Not Religion??
    ... There is no such thing as random chance, ... You roll a die. ... In computer programming, ...
    (talk.origins)
  • Re: Wording of Question
    ... >observe a 3. ... Assuming that each side has probability 1/4, find a) P, ... >first I took a single trial to be n successive rolls of a die. ... first roll, while B says your first 3 comes on the second roll or later. ...
    (sci.math)
  • Re: I cant take it any more.
    ... you to bet on the roll of a die. ... give him 10 bucks but if it doesn't you give him 10 bucks. ... He hits his number I give him ...
    (rec.gambling.poker)