ESP test and chance

As reported in my other thread in this folder, I've been doing this ESP test. Both myself and some other people have got significant results. In order to confirm that there isn't some problem with the website, I made a quick-and-dirty automated test (in Java, using Selenium webdriver for Firefox) to test colors in precognitive mode.

The attached Java code (rename .txt to .java) runs with colors, and does 30 runs of 25 trials. It uses Java's ThreadLocalRandom class to generate a number between 0 and 4, which is then used to select a color.

The raw data is in the attached text file. The highest score was 9, the lowest was 0, and collectively the results were slightly below chance (137/750, on a binomial distribution p=0.89). This is expected as it's just randomly generated guesses.

So for some reason, humans on this site seem to be getting consistently higher scores than a random number generator. I believe this is evidence of a real psi effect, as the null hypothesis is unlikely given the statistical difference between human experimenters and a computer.

I will rerun the automated test in clairvoyant mode to see if there are higher results with that, and also to see if the test ever reaches 11 or 12 (which several humans on this site have reached). But to me, as my first semi-scientific parapsychology experiment, this is exciting evidence!

As reported in my other thread in this folder, I've been doing this ESP test. Both myself and some other people have got significant results. In order to confirm that there isn't some problem with the website, I made a quick-and-dirty automated test (in Java, using Selenium webdriver for Firefox) to test colors in precognitive mode.

The attached Java code (rename .txt to .java) runs with colors, and does 30 runs of 25 trials. It uses Java's ThreadLocalRandom class to generate a number between 0 and 4, which is then used to select a color.

The raw data is in the attached text file. The highest score was 9, the lowest was 0, and collectively the results were slightly below chance (137/750, on a binomial distribution p=0.89). This is expected as it's just randomly generated guesses.

So for some reason, humans on this site seem to be getting consistently higher scores than a random number generator. I believe this is evidence of a real psi effect, as the null hypothesis is unlikely given the statistical difference between human experimenters and a computer.

I will rerun the automated test in clairvoyant mode to see if there are higher results with that, and also to see if the test ever reaches 11 or 12 (which several humans on this site have reached). But to me, as my first semi-scientific parapsychology experiment, this is exciting evidence!

Great stuff Oliver. Glad to see you do this!

It is interesting evidence indeed. I was once a participant in an actual Telekinesis event, which made me question ALOT of things! I find that it's very exciting and energising to be actively involved in PSI in a general sense, especially when other people are witnessing/participating/sharing that journey with you. I've found that there is a world of difference between reading about it, and being part of it. 

Good luck with the testing (not that you'll need it!   ).

Card Test outcome: 13 hits in 25 trials, for odds of 2708.7 to 1. About time I did something right.

   

   

What about repeatability? If this is the first and only attempt of the day, then it's reasonable. On the other hand if it's just a selected highlight, it is maybe a bit like ups and downs of following a football team - we celebrate the wins, and don't talk so much about the losses.

What about repeatability? If this is the first and only attempt of the day, then it's reasonable. On the other hand if it's just a selected highlight, it is maybe a bit like ups and downs of following a football team - we celebrate the wins, and don't talk so much about the losses.

A point that should be stressed. Posting only good results may be fun, but the results are worthless in isolation. Only the total results matter.

In posting my good results for last night's session (I quit after getting that high score, BTW), I wanted to highlight the fact that the GotPsi tests (the name I and a great many veteran users know them by) are unique in maintaining records of all user trials. Naturally, this feature can be valuable for data analysis.

In my case, I'd like one day to examine my data for run-length position effects (a la Thouless, The Pattern of ESP). I've done millions of trials since the early 2000s, and all but a handful were performed via runs of exactly 25 trials, so my data is really homogeneous in that respect. I'd also like to examine the data for displacement effects, where cards beside the one I choose may reveal the target more often than one might expect by chance.

There are a though in my mind say the correct answer but i avoid it and fail. 
 In a moments star in my mind i chose wave and yes star is the correct answer.
When i chose Square it wave, my my say wave but English is not my language so i Confused it into squares. It's weird