Oldskeptic wrote:With such a small group to choose from I think it does become more an either or situation with 50/50 probabilities than a guessing game with 25% probability.

I don't see how, if the guesser has 4 possible choices they could make, all equally likely, the odds of any single guess being right can be other than 1 in 4.

Oldskeptic wrote:In Sheldrake's experiment when a 5 or 6 was rolled they rolled again instead of letting the phone ring with no sister on the line. We will never know from Sheldrake what would have happened if 5 and 6 had been left in.

We won't know because that would have been a different experiment, with effectively more callers.

I'm not sure someone can necessarily be blamed for not doing a different experiment.

Oldskeptic wrote:Well your math is all fine and good but it has already been done using probability theory showing different results. And I can say, from experience, that the probability matches reality. A professor of statistics/probability of mine 35 years ago, in a lecture hall holding around 100 students, picked the first person in the front row and asked her birth date. He went down the rows of students and before 50 he had a hit.

He did it again with the next student in the first row and had a hit well below 50.

He did it again with the third person in the front row with the same result.

I'll admit that this professor seemed to be something of a mad scientist at the time but he did impress on me that statistics are not always what they seem to be, and that there is always another way to look at them.

If you aren't misremembering the experiment, the birthdates were obviously very far from random or something seriously odd (or fixed) was happening.

Look at it this way.

A line of 50 people all happen to have different birthdays - what's the probability that one of them shares my birthday?

50/365, or 0.137 - it really can't be anything else since there must be 315/365 birthdays still 'free'

Now, in a line of 50 people, it is likely that some of them will share a birthday, and so the probability of any of them sharing a birthday with me is necessarily less than 0.137.

In fact, I make it 0.128.

For simple statistics, there isn't 'always another way to look at them', unless you include the wrong way.