Dowsing trial for million dollar prize

Larry Lovage

Thinker
Joined
Oct 26, 2006
Messages
220
Eight experienced dowsers took part in 6 trials in which they were required to locate a bottle of water in a concealed container from a row of six such containers, the remaining five containing sand. Participants had to locate the water 4 or more times out of the 6 trials in order to pass the test. In line with previous controlled tests of dowsing ability, the results for each individual dowser were below chance expectations.
40 out of 60 would be significant. I don't see that 4 right answers out of 6 could remotely count as significant enough to qualify for the final test.

And surely performance less than chance is no less miraculous than more thanchance. If the dowsers are making the wrong decision from the information they are receiving such that they are moving away from the water instead of towards it, it simply sounds as if the experiment is not as completely rigorously controlled as it would need to be.

Can someone explain this to me? Reading the Million Dollar prize applications, I've frequently been disturbed by what seem to be lacklustre controls over the qualifying demonstrations, such that it begins to look like the psychics' (etc) criticisms of the Randi prize conditions could almost be justified.
 
Your first complaint (that 4 of 6 is not significant enough), even if legitimate, is a flaw in favor of the applicants, not against them.

Your second complaint (that below chance is just as miraculous as above chance) is not legitimate:

Chance would dictate 1 of 6. Therefore, below chance is 0 of 6. If you could extend the distribution curve to the left of zero, you would see that 0 of 6 is within expectations, just as 2 of 6 is.

In addition, and most importantly, the dowsers take part in designing the trials and agree to it. If they want to be tested for being wrong more than would be expected, they can claim that as an ability. None of them do so. And in the pre-trial, in which they know where the water is, they don't "make a wrong decision" from the information they are getting.

Their claim is that they get things right. That is what they are tested for.

If they want to claim that they get things wrong, then they need to design a test for that.
 
Garrette said:
Your first complaint (that 4 of 6 is not significant enough), even if legitimate, is a flaw in favor of the applicants, not against them.
Uhm. Yes, I know that. That's partly what I was complaining about!
 
Could you provide a link or a source for the description in the opening post?
 
Given that there's a one in three chance of missing all of them, I don't think that they performed below chance at all. That may be a misstatement. Is that from a Randi commentary?

Here are the odds of achieving a specific number of successes by chance in six trials at 5:1 odds:

Code:
0     33.5%
1     40.2%
2     20.1%
3      5.4%
4      0.8%
5      0.1%
6      0.0%

The odds of getting all six are less than 1 in 200,000.

26 out of 60 would be about 1 in a million odds. Microsoft Excel breaks if I plug in 40 out of 60 (It shows the probability as negative.)
 
The odds of getting all six are less than 1 in 200,000.

If the probability of getting each test right is 1/6, then the probability of getting all 6 right should be (1/6)6, i.e. 1/46656.

How did you calculate < 1/200k?
 
At any rate, my own objection to the 4 out of 6 trial is provisionally removed. It just struck me that it wouldn't be that unlucky for one of the 8 to hit the 1 in 6 jackpot four times out of six. But because the number of trials was the same as the number of test bottles, I think I got confused when initially thinking about it. Otherwise I wouldn't have posted a whole thread. However, 4 out of 6 is (1/6)4 which is 1/1296, divided by 8 for the number of dowsers, there's a 1/162 chance of having to actually go to the preliminary test. So, if it happened that there were so many dowsers needing testing that you did this test three times a week, someone would hit the jackpot sometime in the first year.
 
If the probability of getting each test right is 1/6, then the probability of getting all 6 right should be (1/6)6, i.e. 1/46656.

How did you calculate < 1/200k?


44656 is correct, which is 1/2.14 x 10-5, which I mentally converted to 1 in 200k incorrectly. :blush:

Thanks for the assist.
 
At any rate, my own objection to the 4 out of 6 trial is provisionally removed. It just struck me that it wouldn't be that unlucky for one of the 8 to hit the 1 in 6 jackpot four times out of six. But because the number of trials was the same as the number of test bottles, I think I got confused when initially thinking about it. Otherwise I wouldn't have posted a whole thread. However, 4 out of 6 is (1/6)4 which is 1/1296, divided by 8 for the number of dowsers, there's a 1/162 chance of having to actually go to the preliminary test. So, if it happened that there were so many dowsers needing testing that you did this test three times a week, someone would hit the jackpot sometime in the first year.

Actually, it was almost a 7% chance that at least one would win regardless of ability. They individually had 1 in 115 odds.

1-(114/115)8 = 6.75%
 
I used (Ps)k * (1-Ps)(n-k) * COMBIN(n,k) = P(k) (the probability of k successes out of n trials), but for k>28 the actual prob is so small that the cumulative P is 1 (to the precision Excel is capable of).

After a bit of fiddling, I've got it cough up that P for 40+ out of 60 is 9.1 * 10-18, i.e. 1 in 1017. So IMO, any dowser getting 40 or more out of 60 in a properly controlled test has probably earned themselves the 1 million dollars.
 
Besides, I think a preliminary test is just supposed to be a coarse sort of thing. Remember, dowsers are notorious for being self-deluded. (Most of them probably think they'd be right all the time, and will be surprised that a guess is wrong ever.)

If you had someone getting a consistent 4 or better out of 6, all it means is that you do a more refined test. At the preliminary stage, we're not really proving a paranormal power. (If they pass the prelim, they don't get the $1 million.)
 
I guess I'm only assuming that this is a preliminary test. If someone had passed a preliminary test and was now in a final test for the million, I sort of assumed it would be big news.
 
Besides, I think a preliminary test is just supposed to be a coarse sort of thing. Remember, dowsers are notorious for being self-deluded.

From Bindamel's table you can see that 4+ out of 6 is about a 1% chance. So on average only about 1 in every 100 dowsers would ever be tested further.

I begin to see why researchers in parapsychology labs have such poor controls or low success criteria. If they didn't, they'd spend most of their time twiddling their fingers or playing solitaire
 
From Bindamel's table you can see that 4+ out of 6 is about a 1% chance. So on average only about 1 in every 100 dowsers would ever be tested further.

Meaning you'd have 1 false positive (Type 2 error) in 100 rounds of testing. Yes--exactly. The preliminary test is meant to save time and money.

If someone actually had the ability that dowsers claim to have, they would definitely get through the prelim and into seriously rigorous testing.

Again, since these guys are mostly believers (rather than out and out frauds), I don't think you've got to worry about every little detail to set up what should be an eye-opening trial.

To avoid all the post-hoc stuff (someone cell phone caused interference) you do the dry runs where they "dowse" with full knowledge of where the water (or whatever) is, and ask them once more if there's any condition that will prevent them from performing.

If someone can repeatedly "win" in the prelim, then you can refine the statistics, tighten up controls and look for cheating. Before that, it's just not worth it.
 
I wonder what the repercussions would be if a dowser actually won the preliminary trial.
If would be difficult to explain to the public why he has to sit again.
 
I wonder what the repercussions would be if a dowser actually won the preliminary trial.
If would be difficult to explain to the public why he has to sit again.

Why? Is a hundred-to-one fluke considered so unlikely that a significant proportion of the population would consider the matter settled?

Explain it as being akin to someone who got three numbers on last week's lottery claiming to be able to predict lottery numbers. Doing it once isn't proof he can, but not doing it once is a fairly convincing argument that he can't.

I do wonder how the dowser might explain it though. It would be an obvious step to spin himself as 'the dowser that beat Randi's challenge', and complain bitterly how JREF 'moved the goalposts' on him.
 
The sceptic....
Explain it as being akin to someone who got three numbers on last week's lottery claiming to be able to predict lottery numbers. Doing it once isn't proof he can, but not doing it once is a fairly convincing argument that he can't.
versus the winner denied the prize...

'the dowser that beat Randi's challenge', complain[ing] bitterly how JREF 'moved the goalposts' on him.
I wonder where the sentiments of the public at large would lie?
 

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom