We’re getting into the nitty gritty details at this point which I think is going beyond the OP - I think we would all agree that a properly designed experiment will rule out any other method of a “hit” aside from (a) random guess and (b) actual dowsing ability.
The method you describe with buried pipes has in fact been done before. I’ve got a book describing one such test in great detail and showing photos of the experiment and listing the outcome. If I mention the book or the tester Peter Morris’s head will explode and since Involuntary Cranial Detonation is a lousy way to start the weekend I shan’t do so
Not enough to matter. If you’d rather use a random number generator But if you really care, take a coin. Flip it twice. If it’s head-head, or tail-tail, ignore it. If it’s head-tail, that’s 0. If it’s tail-head, that’s 1. There’s a fairly random bit of information.
Or take eleven ten-sided dice. As different as you can get. Roll ten of them in boxes marked 1-10. Roll the eleventh die to determine which numbered box you’ll use for the result.
Either of those will fix most problems with random number generation.
Or use a physical random number generator driven by radioactive decay, picking the last digit of the decimal number generated.
I believe that is in fact true, depending on the construction of the die (specifically how the faces are marked), how they are tossed and over extremely long runs of die rolls. Since we’re talking about doing a few tens of trials I don’t think it’s significant.
If you want to discuss ways to come up with truly random selections we can go to some lengths but there are plenty of simple methods that will suffice for the limited number of trials we’re doing. If things like dice, wheels and bingo ball widgets are good enough for large commercial concerns to allow them to be used where billions of dollars are being gambled on a routine basis I think we’ll be in good shape with some d10 and a die cup.
That sounds like the same test I’m referring to. Randi screwed it up through utter incompetence, twisted the results to make it seem like he won, then told a tale in which he is a hero. He’s just doing it to sell books. He’s only interested in your money, and doesn’t care about accurate information.
I’d just like to point out that the OP is not asking for a debate on what Randi does or doesn’t do.
At this point he isn’t really asking for anything (see his later comments) but let’s do it this way:
Peter Morris, how would you design an adequately controlled experiment to test for dowsing? Let’s take the OP’s conditions, that the testee says that concealed jugs of water are an acceptable target, and that he believes he achieves approximately 20% accuracy.
Just in case other posters aren’t aware, Peter Morris has a bit of a history on the subject of James Randi. Some time back, he interpreted Randi’s statement of “Prove me wrong and win a million dollars” to mean proving Randi wrong on any point whatsoever of anything Randi had ever said, rather than the obvious meaning of proving Randi wrong on the subject of the existence of paranormal abilities, and therefore thinks that Randi owes him a million dollars. This experience has colored his entire view of Randi, and he therefore now interprets everything Randi does as dishonest.
The incident Chronos refers to was my attempt to show how dishonest he is. The response was, essentially, “We know he’s a liar. We love his lies. We hate you for pointing them out”
Whatever method produces random numbers over the number of trials that are done is fine with me. Let’s stipulate that RNT are fine with you, me, the testee and everyone else involved.
How would you set up the rest of the experiment? Number of trials, what constitutes success, etc?
I notice you haven’t answered several of my questions.
Please answer the following, because it almost exactly reflects what actually happened on a specific Randi test. I think it’s my most important question.
Hypothetical case: In a particular test, a statistician says that a score of 30% or over is significant. A dowser claims to have an 80% success rate. In fact, he takes the test and scores 35%. Which of the following is true?
he passed, because he got more than the statistically significant mark
he failed because he got less than he said he would.
As far as thart goes, I wouldn’t presume to do so, because I am not trained in science or maths. Anything that I say would probably be wrong.
James Randi is not trained in science or maths either. I reject anything he does. I only accept tests by actual trained scientists, with proper peer review. I reject anything else.
It’s a well-designed test. No need to reinvent the wheel.