Background
Recent conversations in the forums remind me of a pivotal moment in my magical development, and an associated experiment that I did myself many years ago, and have instigated in other discussion forums over the years (Usenet, Myspace, etc.). I figured we could give it a shot. It's been a long time for me, and I'm due to give it another go anyhow.
Several reader/posters that frequent places like this tend to periodically express an ability to manipulate physical phenomena, on a large scale in some cases (weather, earthquakes, power outages, etc.), and often on a smaller scale (moving a pencil across a table, influencing dice, etc.). I admit to a personal history of making similar claims. At one point many years ago, a highly respected mentor of mine suggested that maybe, just maybe, my sense of power/ability to cause some such effects might be just a tad exaggerated. He suggested a simple test, which I did, and which I failed. Believe it or not, I count that as the most valuable instruction of my lengthy magical career.
The test is simply to see if you can beat the odds of rolling a selected number on a fairly-weighted 6-sided die, according to standard statistical reasoning (details below). For the record, I have not, to date, beaten this test personally, though I continue to assume that it is possible. I plan to try a magical method I have not used before this time around.
Participants and Methods
The experiment is open to anyone, and to any magick-related methods that participants choose, from straight up telekinesis, to spell-casting, to evoking spirits, etc.. Whatever one feels they can do to influence the outcome. Needless to say, any method that would be considered cheating at a casino craps table would be cheating here too (they don't count magick or telekinesis as cheating by the way).
Procedure
Use a fairly-weighted, single, 6-sided die. You can assume that any store-bought dice that were not sold as
trick dice are fair, but generally speaking, the less you pay, the greater the chance of getting defective dice, so try to get casino-grade dice if you can.
1. Select a number to be the target number for the duration of the experiment.
2. Roll the dice 100 times and record how many times your number comes up over the course of the session.
Note. Roll the die into an empty shoe-box or similar container, with enough force to to bounce the die off at least one side. You are allowed to look at the die and position it in your hand as desired, but you're not allowed to throw so that it slides but never rolls.
3. Repeat step 2 every day for 9 additional days, for a total of 10 samples of 100 rolls.
Results
The criterion for a successful sample will be hitting the target number 21 out of 100 rolls. (see Appendix for an explanation of the statistical reasoning), and the overall experiment will be judged a success if a participant has 7 or more successful samples out of 10.
Report
After the 10 days, post your results in the form of X number of successful samples.
Discussion
After you have posted your results, please share the technique you used, your experience during the experiment, insights, or other comments.
I would like to personally request that all participants and observers commit to accept any reports and claims made subsequent to this experiment at face value, at least as far as the public discussion goes. If someone lies about their results, well, shame on them. But no matter what gets posted, resist the temptation to assume someone is making false claims if that's your impression. I'm concerned that people might be reluctant to post wildly successful results, but I want people to feel free to post what happens without concern for what people might say, so please, let's all agree to take what is posted at face value for the sake of discussion.
Appendix
Statistical Reasoning
The probability of any one number coming up on a 6-sided die is 1/6 or 16.67% (16 to 17 times out of 100 rolls) each and every roll. However, chance expectation predicts that for each 100 rolls, the actual outcome will vary around that number, roughly equally above and below the computed probability. In other words, you can expect your number to come up somewhere near 17 times out of a hundred, but about half the samples will be higher than that, and about half the samples will be lower than that, such that over the long haul they will balance out to very close to 16-17%.
Now, when things are expected to vary equally above and below a value (which is the case here), we can estimate how much we expect the values to vary above and below. That's called the standard deviation, and it happens to be 3.73, computed by taking the square root of: the sample size multiplied by the probability of a hit multiplied by the probability of a miss. So, we expect that for any 100 rolls the target number will come up 17 plus or minus 3.73 times. It will be higher and lower than that in some samples, but the majority of samples (about 70%) will be within that estimation.
The statistical criterion for beating chance in most scientific studies is usually about 2 standard deviations, which for us would mean one would have to roll the target number 25 or more times out of 100. However, it's possible to have a consistent effect on the dice that is too small to meet that criteria, so instead, the experiment is set up with 10 replications, so that we have a chance of detecting small but consistent effects.
What I propose then is to count any 100-roll sample a success if it exceeds 1 standard deviation above the probability: 16.67 + 3.73 =20.4, so we'll say 21 rolls out of a hundred is a winning sample. Then, because we can further expect about 70% of the samples to be within 1 standard deviation above or below 17, I propose that we set at least 7 out of 10 successful samples as the criteria for beating the odds.
