Hey guys. I’m venturing into divination, but I wondered if any of you could suggest a good methodology for me. The trouble is that I suffer from what I call the Curse of the Scientist. I automatically, both consciously and subconsciously, reject any finding that fails statistical significance.

In science, we generally set statistical significance at 1 in 20, i.e. 5% (p < 0.05). This is to say, any finding that had a greater than 5% probability of occurring by random chance is rejected; there’s a variety of reasons why we use this threshold, but I won’t get into that here.

So for example, if I was doing penny divination, I would need to flip the penny 6 times to achieve significance, as the chance of getting 6 heads or 6 tails by chance is 3.125%.

5 heads and a tail or 5 tails and a head would not meet this threshold. There are 14 permutations of this (HHHHHH, TTTTTT, THHHHH, HTHHHH, HHTHHH, HHHTHH, HHHHTH, HHHHHT, HTTTTT, THTTTT, TTHTTT, TTTHTT, TTTTHT, TTTTTH), and there are 2^6 = 64 total possible outcomes. That means that the probability of getting one of the ‘acceptable’ outcomes is 14/64 = 21.875%.

To be able to accept one “error run”, I’d need to flip 9 times. There would be 20 ‘acceptable’ outcomes (2*9+2) and 2^9 = 512 possible outcomes. As such, the probability of getting an ‘acceptable’ result by chance is 20/512 = 3.9%, which is below the maximum threshold of 5%.

Needless to say, it would be very hard for a spirit to push a coin towards either heads or tails consistently 8 out of 9 times in a trial. So, are there easier divination methods that still would fulfill the 5% statistical significance criterion?