Art,
It doesn't matter whether I assume dowsing to be ideomotor or not. For the test, all that matters is whether dowsing works, regardless of the mechanism. Either it works, or it doesn't. If it works, then we can study the mechanism by which it works. If it doesn't work, then we can study the mechanism by which dowsers are being deceived.
Chance guessing has nothing to do, per se, with flipping a coin, or 50% odds. Predicting the flip of a coin is just ONE example of chance guessing, and happens to have odds of 50% per flip. Predicting a die roll is another example of chance guessing, where each event has odds of 1-in-6. Guessing a card from a well-shuffled deck has odds of 1-in-52. PowerBall has odds very close to zero (1-in-146,000,000 right now, or 0.0000007%).
Do you understand now that the probabilities associated with guessing depend on how the experiment is set up? This is a very mature branch of mathematics, taught in every university on Earth, but you are welcomed to try and prove it all wrong.
If I therefore place a gold bar (secretly) under one of two paper plates and ask you to tell me which one, you have a 50% chance of being right just by guessing, each time we do this. Does that make sense?
If we keep doing this same experiment, where I hide the gold bar and you guess the location, then sometimes you'll guess right, and sometimes you'll guess wrong. Over the long run, you'll get about half right, and about half wrong. Sometimes you'll get a run of several right answers in a row*, and sometimes you'll get a run of several wrong answers in a row. Does that make sense?
Now bring in a pair of dowsing rods. We re-do the whole thing, but this time you get to use dowsing rods. Can you now locate the gold bar with better results than guessing? If your results with the dowsing rods end up the same as guessing, then dowsing doesn't appear to be very useful.
As I said, statistics is very mature, and for these simple shell-game tests we don't have to run the guessing part of the test to verify that statistics works. I can calculate the gaussian distribution for any such test, and tell you exactly what the odds are for any level of success (lotteries and casinos can do this, too). Now if you chose NOT to believe mathematics, then we can certainly run the guessing portion for your own satisfaction. Randi would, too, I'm sure. In fact, if several observers are standing around, you could run the guessing part WHILE you are dowsing.
You need to understand, though, that even guessing has some normal variation in the distribution. If, ferinstance, we ran our little 2-plate guessing experiment 20 times, you might guess correctly 10 times, or 9 times, or even 11 times. In fact, for 20 runs with just 2 plates, 11 correct guesses has about the same odds (16%) as does 10 correct guesses (17.6%). So 11-of-20 correct guesses is not a very meaningful "success".
So when we compare dowsing to guessing, we cannot set the threshold of what we consider "successful dowsing" at exactly the mean for chance guessing, i.e., 10-of-20 in this example. That's because guessing 1 or 2 over the mean would not be an unusual outcome. That's why Dell said he would like to make anything over the mean to be the threshold for dowsing success... it's almost certain to occur, even when guessing!
So we need to set the threshold of success somewhat higher to differentiate successful dowsing with guessing. How much higher depends on how many tests you're willing to run, and it also depends on how many target locations you use. Does all that make sense?
Now, I know you've read a definition for "double-blind" that talks about "groups". Yes, some DB tests have people who are in a "test" group, and some who are in a "control" group. But in shell-game-type tests, the "shells" are actually the groups. That is, if I hide a gold bar under 1-of-2 paper plates, then the plate with the gold bar under it is the "test group" and the empty plate is the "control group."
You, as the subject, don't know which group is which. That's the "blind" part of the experiment. Everyone who is watching you try to guess (or dowse) the gold target, including the proctor for the test, also does not know the gold bar's location. That's the "double blind" part of the experiment. The guy who actually hid the bar, left the scene without making contact with anyone else involved. This is analogous to drug testing, where some anonymous person randomizes the drug and the placebo in some dark room, so when the doctor hands the drug to the test subject, even he doesn't know whether it's the drug or the placebo.
Speaking of drug testing, do you understand why it has a test group and a control group, and why one group is compared to the other? It's because the placebo effect does not follow probabilities. That is, there is no mathematical predictability for how a placebo will affect a certain ailment. Heart disease and cancer, ferinstance, have almost no response to a placebo. But headaches and joint pains do. Since there is so much variation in the placebo effect, and no way to predict it, drugs have to be tested this way. When the test is over, and the results compared, the drug company hopes their new drug is significantly better than the placebo, and some are willing to cheat to make sure.
So, once again, I'll say that my test is double-blind, and makes a fair comparison of dowsing to the well-known statistical results of guessing. I welcome you to take my test procedure, as well as this explanation, to a university statistics professor and ask his opinion. I also continue to urge you to give Randi a shot.
- Carl
*Getting several right answers in a row is what often fools people into thinking they have a "gift".
