James Randi - sincere skeptic or money-grubbing fraud? (dowsing)

[patchbunny]

*Originally posted by peter morris *
**
In the first place, you have confused two seperate things, here, my comments about the lack of logic in cecil’s column, and my opinion about Randi.

Cecil’s column offers the theory of a dowser getting subliminal signals from the landscape causing the stick to move through ideomotor action. He then says that this proves dowsing is a fraud. That makes no sense at all.**

I agree. Thank God he actually says nothing like this.

**

IF the theory is true, then the stick twitches when he moves over water, which he is unable to detect conciously. That seems like dowsing to me. If it looks like a duck… Rather than Cecil saying that its a fraud, he should say that it works, with a simple scientific explaination. **

No, the stick moves because the dowser thinks he’s over water, whether there’s water present or not.

Once again, let’s review what Cecil actually says:

 “Needless to say, dowsing is entirely a fraud, although often 
  an unconscious one.  Innumerable experiments, beginning in 
 1641--that's right, 1641--have demonstrated that:

 (a) The presence of water has no discernible effect on a rod 
 held above it, whether the rod is made of wood, metal, or 
 anything else.

 (b) The success rate for diviners is about the same as that 
 for people who use the hit-and-miss method when looking for 
 water.

 (c) Geologists trained to recognize telltale surface clues 
 (certain kinds of rocks and 	plants, various topographical  
 features) will invariably far outdo dowsers in predicting 
 where water will be found, and at what depth.

 Nevertheless, belief in dowsing has persisted, partly because 
 most people secretly want 	to believe in magic, partly because 
 water is fairly easy to find in most parts of the inhabitable 
 world, and partly because the plunging-stick phenomenon 
 seems so convincing to untutored observers.“

I fail to see any sort of “lack of logic” in Cecil’s part. Dowsing doesn’t work for the reasons described above. Nowhere in the column does Cecil say, as you state, that because the stick moves through ideomotor action, dowsing if a fraud. That’s your claim, not his.

**

As to the part about “normal” versus “paranormal” the distinction is meaningless. What, exactly, is “paranormal” anyway? How edo you define what is paranormal?

Just as an example consider this: flowing water causes minute vibrations, someone might detect them subconciously, and the stick moves through ideomor action. Is that normal, or paranormal? What if flowing water creates a magnetic field, which some people might sense? Normal or paranormal? What if it makes a sound, below the level of concious hearing, but can be sensed subconciously? Normal or paranormal?

Paranormal merely means unidentified. Once you acknowledge the existence of something, it automatically becomes normal.

The issue should be whether dowsing works at all, by any method whatsover, with no distinction between “normal” and “paranormal”. **

That is the issue. That’s why dowsing has been tested so many times. So far no one’s been able to demonstrate that dowsing will help you find anything other than gullible people. It doesn’t matter if the rod is moving because of the ideomotor effect or if the dowser has Parkinson’s. What matters is whether or not when the dowser says “There’s water here,” if water is actually present. And what has been shown is that when they do, they’re success rate is no better than chance.

–Patch

[Mr.Miskatonic]

*Originally posted by Irishman *
**Princhester said:
Mr. Miskatonic, I did put two mentions in the original thread, and at least one other person referred to it. I can’t help it if you can’t read. **

Hey, I’m not blaming you, its Peter I’m blaming. First he continues in the AACC thread after everyone has bailed. Then he posts to the pit thread, then shifts to the GD forum.

Obsheesh: Sheesh!

[Princhester]

Hey peter, I’ve been watching this space for days now. My eyes are getting sore. Are we nearly there yet?

[patchbunny]

*Originally posted by Princhester *
**Hey peter, I’ve been watching this space for days now. My eyes are getting sore. Are we nearly there yet? **

He appears to have moved over to the sci.skeptic newsgroup at the moment, where he’s not faring any better.

–Patch

[IzzyR]

*Originally posted by RickJay *
peter, we WILL need evidence that supports your example of “Randi would ask for 4 out of 5 when a psychic can get 2 out of 10, where random chance would be 1 out of 10”?

Actually if you look at the link about dowsing it is pretty clear. Dowsers testing for water in pipes had a 22% success rate when a 10% rate would have been predicted. But Randi declared it a failure because they had predicted an 86% success rate.

One might argue, as you seem to have, that it is only fair to hold these dowsers to their claimed success rate. But while it may be fair in terms of Randi’s dealings with the dowsers, it is misleading in terms of demonstrating the inefficiency of dowsing. IOW Randi is being misleading by presenting his results as demonstrating that dowsing doesn’t work, when in actually he appears to have demonstrated that it DOES work but just not as well as it’s practitioners believe.

[Patty_O_Furniture]

I’d like to see any cite that says what chance predicts for finding underground water. How do you calculate that, anyway? Walk over a 100×100 square foot piece of land (which may or may not have water under it); how can you say that chance predicts any success rate at all?

I mean, it’s not like flipping a coin, where you know the number of sides on the coin; or guessing at cards where you know the number of cards in the deck, and how many suits there are.

[RickJay]

*Originally posted by IzzyR *
Actually if you look at the link about dowsing it is pretty clear. Dowsers testing for water in pipes had a 22% success rate when a 10% rate would have been predicted. But Randi declared it a failure because they had predicted an 86% success rate.

Izzy, what link are you talking about? There are multiple links in this thread, and the one that I see referring to a Randi test states that the dowsing success rate was 13.5% overall, not 22%. It was 22% in a subset of the tests (water pipes) and not all the tests, which included dowsing for brass and gold, which the dowsers had claimed they would be even more successful detecting.

I don’t think I need to remind anyone that if random chance would expect 5 out of 50 and you get 11 out of 50, that is statistically unremarkable. I don’t have the formula for calculating a binary probability in front of me, but if random chance says 11% and the dowsers are saying 90, and the results hit 13.5% (the actual success rate on the whole test) what do YOU think that proves?

And The dowsers agreed to this test and agreed to the success rate. If their methods only worked 20-25% of the time, sure, it would have been reasonable to simply run a great many trials to establish a strong confidence interval at 18-22% success. But they claimed 80-90% success, and scoring 13.5% in 111 tries when your success rate is normally 90% - well, it’s impossible. In 111 tries, if your probability is .9, the likelihood of success in only 15 out of 111 trials is astronomically low.

This is why the JREF asks the applicants to set their OWN performance criteria.

**One might argue, as you seem to have, that it is only fair to hold these dowsers to their claimed success rate. But while it may be fair in terms of Randi’s dealings with the dowsers, it is misleading in terms of demonstrating the inefficiency of dowsing. IOW Randi is being misleading by presenting his results as demonstrating that dowsing doesn’t work, when in actually he appears to have demonstrated that it DOES work but just not as well as it’s practitioners believe. **

13.5% success when random chance dictates 10%, in only 111 trials, or 22% in 50 trials, demonstrates very little. No reasonable person could draw a positive conclusion from that. Now, if you made it 10,000 trials and they were still running 22% in a truly well-designed double blind test, I’d be convinced. Note that the dowsers performed WORSE on detecting brass than random chance would suggest, so it works both ways.

OF course, one could argue Randi hasn’t proved dowsing DOESN’T work, and I agree. He hasn’t. You’ll also note in the link that he doesn’t actually say “this test proves dowsing is false.” But let’s be clear; you have to prove it DOES work, not that it doesn’t.

[IzzyR]

*Originally posted by RickJay *
There are multiple links in this thread, and the one that I see referring to a Randi test states that the dowsing success rate was 13.5% overall, not 22%. It was 22% in a subset of the tests (water pipes) and not all the tests, which included dowsing for brass and gold, which the dowsers had claimed they would be even more successful detecting.

I specifically mentioned water. I would consider dowsing for water as independent from dowsing for anything else, regardless of claims that it would be effective there as well.

I don’t think I need to remind anyone that if random chance would expect 5 out of 50 and you get 11 out of 50, that is statistically unremarkable.

Well its a good thing you didn’t remind anyone of this because you would have been quite wrong. The likelihood of random chance producing any number greater than ten on such a test is less than 1%.

And The dowsers agreed to this test and agreed to the success rate. If their methods only worked 20-25% of the time, sure, it would have been reasonable to simply run a great many trials to establish a strong confidence interval at 18-22% success. But they claimed 80-90% success, and scoring 13.5% in 111 tries when your success rate is normally 90% - well, it’s impossible. In 111 tries, if your probability is .9, the likelihood of success in only 15 out of 111 trials is astronomically low.

This is why the JREF asks the applicants to set their OWN performance criteria.

Again, I am using water in isolation of other materials. (BTW, the likelihood of random chance producing greater than 14 successes out of 111 tries is about 14%, also a pretty impressive number. Nonetheless, I prefer to focus on water, which is - as I understand it - the most common application of dowsing, and where the results were most conclusive.)

In general, it appears that you made a fatal error by not bothering to actually calculate the binomial probabilities before posting your assertions.

I also note again that Randi’s test might have done a fine job of demolishing the claims of the dowsers. What it did not do is a good job of establishing that dowsing does not work - to the contrary, in fact. The fact that the dowsers were claiming overly high success rates is irrelevant to the results of the test (other than the agreement between the parties, obviously).

[RickJay]

*Originally posted by IzzyR *
**I specifically mentioned water. I would consider dowsing for water as independent from dowsing for anything else, regardless of claims that it would be effective there as well.Well its a good thing you didn’t remind anyone of this because you would have been quite wrong. The likelihood of random chance producing any number greater than ten on such a test is less than 1%.Again, I am using water in isolation of other materials. (BTW, the likelihood of random chance producing greater than 14 successes out of 111 tries is about 14%, also a pretty impressive number. Nonetheless, I prefer to focus on water, which is - as I understand it - the most common application of dowsing, and where the results were most conclusive.)

In general, it appears that you made a fatal error by not bothering to actually calculate the binomial probabilities before posting your assertions.**

I’m sorry, but there’s no “Fatal error” here. i fgirued the probability was LOW - in my head I guess three or four percent - but 1-2% isn’t a big deal. 14% is totally unimpressive; a 1 in 7 shot isn’t proof of anything. 1% would be impressive if you could replicate it a few times; otherwise, there’s no reason to think it any more than a fluke, especially when it’s a subset of a less successful trial. (After all, the odds of doing unusually well on 1 of the 3 trial subsets are three times higher than on any one alone. I suspect that had they blown the water test but fluked out on the gold test, the dowsing crowd would be hailing that as a major victory, too.)

Again; the test does not prove, conclusively, that dowsing does not work. However, there is no question that it fails to prove it DOES, and it certainly proved the dowser’s claims were false.

In terms of the fairness of the JREF challenge, a 14% fluke result that falls well below the stated claim is obviously not good enough to hand out a million bucks. I mean, not to point out the obvious, but if that was the standard you’d just have to spend a year lining up a dozen conspirators to take the same test, and in all likelihood you’d win your million bucks just by sheer force of numbers. To hand out the actual pot, JREF is perfectly justified in expecting results that go well beyond “unlikely chance” and well into “almost certain,” just because otherwise they’d be handing out a million bones every three or four years to people who were just the luckiest 1%.

I also note again that Randi’s test might have done a fine job of demolishing the claims of the dowsers. What it did not do is a good job of establishing that dowsing does not work

And Randi never claimed it did, that I can see. He makes the assertion dowsing is phony, but there’s more than this one experiment to say it’s phony.

[IzzyR]

*Originally posted by RickJay *
I’m sorry, but there’s no “Fatal error” here. i fgirued the probability was LOW - in my head I guess three or four percent - but 1-2% isn’t a big deal. 14% is totally unimpressive; a 1 in 7 shot isn’t proof of anything. 1% would be impressive if you could replicate it a few times; otherwise, there’s no reason to think it any more than a fluke, especially when it’s a subset of a less successful trial.

This is ignorant and silly. Standard practice in statistical testing is to use a 5% likelihood as a benchmark, some use 10%. Less than 1% is quite significant by any measure. In fact, I’m frankly a bit skeptical as to your claim that when you said “statistically unremarkable”, and “No reasonable person could draw a positive conclusion from that”, you really had in mind a 3-4% likelihood. But we’ll leave that aside for now. In any event, less than 1% is a pretty conclusive result.

14% is not considered statistically significant, but it is at least suggestive. Not “totally unimpressive”.

Hey, no one can force you to believe in dowsing if you don’t want to. You can believe that the test represented a 1 in 100 fluke, and no one can stop you. In fact, you can believe all sorts of strange things on that basis. But don’t be distorting the results of the test. Under accepted statistical criteria it strongly supported some level of for dowsing, and to represent otherwise is misleading.

(After all, the odds of doing unusually well on 1 of the 3 trial subsets are three times higher than on any one alone. I suspect that had they blown the water test but fluked out on the gold test, the dowsing crowd would be hailing that as a major victory, too.)

This assumes that the three tests are testing the same thing - they are not, as noted above.

In terms of the fairness of the JREF challenge, a 14% fluke result that falls well below the stated claim is obviously not good enough to hand out a million bucks. I mean, not to point out the obvious, but if that was the standard you’d just have to spend a year lining up a dozen conspirators to take the same test, and in all likelihood you’d win your million bucks just by sheer force of numbers. To hand out the actual pot, JREF is perfectly justified in expecting results that go well beyond “unlikely chance” and well into “almost certain,” just because otherwise they’d be handing out a million bones every three or four years to people who were just the luckiest 1%.

I’ve already made clear a number of times that I am NOT claiming that any of the dowsers should get a million bucks based on the results of the test. Enough times for you to have picked up on it by now. Please stop harping on the subject in your responses to my posts - it confuses the issue.

And Randi never claimed it did, that I can see. He makes the assertion dowsing is phony, but there’s more than this one experiment to say it’s phony.

Of course there’s more to it. But Randi clearly implies (in his “Results and Conclusions” section) that these tests support this claim. They actually support the opposite.

[Spoke]

*Originally posted by Podkayne *
**I was under the impression that, in most areas, pretty much anywhere you dig, if you go deep enough, you’ll find water. Is there any evidence that dowsers really do better than chance? (And, just to head off a lot of pointless posts, let’s avoid the anecdotes and stick with actual studies.) **

Podkayne, you seem to be a voice crying in the wilderness, so I’ll step in to back you up.

Why is “finding” water a big deal? You dig deep enough in any spot and you’ll find water, no? Hell, any dowser who doesn’t have a 100% success rate ought to turn in his stick.

[RickJay]

*Originally posted by IzzyR *
**This is ignorant and silly. Standard practice in statistical testing is to use a 5% likelihood as a benchmark, some use 10%. Less than 1% is quite significant by any measure. **

Izzy, THAT’S ignorant and silly, and you know it is. There is no “standard practice in statistical testing” for a level of confidence, because different tests and measures requires wildly different confidence intervals. If you’re conducting a poll to guess next week’s election, 95% confidence is the traditional benchmark. If you’re measuring weights in a pharmaceutical plant, your confidence interval could be 99.75%. If you want to sell intelligent people that dowsing is real, 99% in one experiment is woefully short.

What you’re asking us to buy here - or, more precisely, what the dowsers are asking us to buy - is that a 99% confidence interval, in one test, is enough to provide significant evidence of the existence of a supernatural ability that has completely failed many other double blind tests. Let’s not wear blinders and think this is the one and only definitive dowsing test. 99% confidence would be fairly interesting if this were the only study done on dowsing, but it’s one of scores of studies, so it’s really not at all remarkable it happened sooner or later, and it’s silly to pick on Randi because it happened to be a JREF challenge. I don’t see peter morris explaining the results of the Munich experiments. Sooner or later you’re going to have someone hit a 99% confidence interval.

Let me put it this way; based on this experiment, and knowing other experiments have turned up goose eggs, do you believe dowsing is for real? I sure don’t. Now, if they’d hit 42 out of 50, or could consistently hit 11 out of 50 in many studies, I’d start to buy it.

[Mr.Miskatonic]

*Originally posted by patchbunny *
**He appears to have moved over to the sci.skeptic newsgroup at the moment, where he’s not faring any better.

–Patch **

Thanks for the heads up, Oh man…he’s begging for leads.

[Irishman]

Well, gee, he should ask me. I know that Ed Mitchell (the Apollo astronaut) has mentioned on his own bulletin board something to the effect of knowing people with bad experiences with Randi. He could start there. Heck, I’ve debated asking Mitchell in an email or message post if he would give me the information.

spoke- said:

Why is “finding” water a big deal? You dig deep enough in any spot and you’ll find water, no? Hell, any dowser who doesn’t have a 100% success rate ought to turn in his stick.

This seems suspicious to me. Why then is well-digging such an expensive operation, and why are geologists so important in helping find where to dig?

[Princhester]

i think spoke- overstates the position. I don’t think (and I have never heard Randi allege) that water can be found in 100% of places. I think it varies but I think 80% isn’t an unrealistic figure.

Additionally, water may be too far down to be practical, or may be at too low a flow to be useful.

Which is why geologists etc may be required (ie for finding useful water at a reasonable depth).

However, the fact remains that dowsers can get away with their claims because water (at somelevel, in some amount) is exceedingly ubiquitous.

[IzzyR]

*Originally posted by RickJay *
Izzy, THAT’S ignorant and silly, and you know it is. There is no “standard practice in statistical testing” for a level of confidence, because different tests and measures requires wildly different confidence intervals. If you’re conducting a poll to guess next week’s election, 95% confidence is the traditional benchmark. If you’re measuring weights in a pharmaceutical plant, your confidence interval could be 99.75%. If you want to sell intelligent people that dowsing is real, 99% in one experiment is woefully short.

Well I’m sure there are instances in which you would ask for a greater than 95% certainty, but this is merely because in some cases the consequences of being wrong are severe. 95% is widely used as a level of statistical significance. Your original statement was that this result was “statistically unremarkable”. This is a false statement. You subsequently said that (without replication) “there’s no reason to think it any more than a fluke”. This is ignorant and silly.

In sum, the tests are statistically significant, and there is good reason to believe they are not a fluke. As for whether this settles the issue of dowsing, see below.

What you’re asking us to buy here - or, more precisely, what the dowsers are asking us to buy - is that a 99% confidence interval, in one test, is enough to provide significant evidence of the existence of a supernatural ability that has completely failed many other double blind tests. Let’s not wear blinders and think this is the one and only definitive dowsing test. 99% confidence would be fairly interesting if this were the only study done on dowsing, but it’s one of scores of studies, so it’s really not at all remarkable it happened sooner or later, and it’s silly to pick on Randi because it happened to be a JREF challenge. I don’t see peter morris explaining the results of the Munich experiments. Sooner or later you’re going to have someone hit a 99% confidence interval.

Let me put it this way; based on this experiment, and knowing other experiments have turned up goose eggs, do you believe dowsing is for real? I sure don’t. Now, if they’d hit 42 out of 50, or could consistently hit 11 out of 50 in many studies, I’d start to buy it.

What you are saying here is - for the most part - valid. One test alone does not settle the dowsing issue. But what you’ve lost sight of here is that this thread is not about dowsing. In fact it was specifically started as a departure from a thread in CCC about dowsing. See the title and OP. Your comments would be appropriate in the other thread. But what we are discussing here is whether or not James Randi is a fraud. And it is in this context that I’ve noted that in this particular instance Randi has taken a test that strongly supports the validity of dowsing and misrepresented it as a debunking of dowsing. This calls his credibility into question.

And the reason this came up, if I might remind you, is because peter morris specifically cited this exact technique as being one that Randi uses. You asked for an example and I provided this one.

So once again, to be clear. I am not saying that the Randi test conclusively proves that dowsing is a real paranormal phenomena. The entire body of evidence would have to be considered for that. But I am saying that this particular test strongly supports dowsing, by commonly used statistical criteria, and to claim that this particular test debunks dowsing is a misrepresentation. If you are merely claiming that based on other evidence or theory you personally remain unconvinced, that is another story. But you have to allow that these tests count against your position.

This is just one instance, of course. And I am no Randi expert. But if he does this type of thing often then he is indeed a fraud.

[Philster]

I’m sorry, but the parameters of all tests are agreed upon. How can you defraud someone who agrees to a certain test and the measurement?

The disgruntled people complain about negative energy, or complain the parameters weren’t right after they fail. If “x” = passing and “x” isn’t achieved, and you agree, then you can’t be defrauded.

You cannot be defrauded if you agree to the whole setup and measurement.

[Princhester]

IzzyR and RickJay, I have to say that I agree with almost everything that IzzyR says. Certainly all the statistical stuff. You can’t look at a 99% result and pretend it has no meaning at all. To do so is to lay yourself open to precisely the type of wilful blindness charges that skeptics lay so often at the feet of dowsers etc.

The meaning such a test has is that further testing is warranted. Of course, Randi and others have done such further testing on numerous occasions, and overall nothing significant has ever been found. I think that this knowledge colours Randi’s comments. Which is why I second Philster when I say that to suggest (as a result of this) that Randi is in any sense a fraud for failing to acknowledge the meaning of this result is just way, way, too strong. For several reasons.

Firstly, as Philster says, fraud suggests deliberate misleading. The dowser applicants said what they could do, the test was prepared accordingly, the applicants failed. Absolutely nothing fraudulent in that at all. To throw around allegations of “liar” and “fraud” willy nilly as you-know-who has done against that background of openness is utterly unjustified.

Secondly, there is a very good discussion of the statistical issues involved in the sci.skeptics newsgroup, to which I don’t have access from here (at work). There is a post by an Eric somebody which makes a very good point, which I will attempt to make although his post is probably clearer.

If you come to Randi and say that you can achieve 80-90%, then he sets his test accordingly. Because of the very high rate you say you can achieve Randi sets the test at only 10 iterations.

If you had said in the first place that you could consistently achieve only slightly better than chance (eg that you could achieve 22% on a 10% chance test) then, to draw out a reliable result, Randi would have set the test at 100 or 500 or whatever iterations.

To illustrate, say someone said to you they could flip heads 99% of the time. You might say “OK, 5 bucks says you can’t flip heads 6 times in a row”. You would set the iterations so low because achieving 99% is so hard to do, and very easy to disprove. Say the guy then gets 5 out of 6. You then say, “Sorry you fail, and you haven’t demonstrated jack shit with any degree of conclusivity.” He then turns on you and says “chance would have predicted that I only get 3 heads, but I got 5, so I do have magical powers” and when you fail to recognise this, he says “you are a fraud”.

This is peter’s argument. He has made it several times in the sci.skeptic newsgroup. It is crap because he is taking an isolated test, designed to test a very strong claim (80-90%) and then extrapolating from a strong but not very strong result, to suggest that the person setting the test is a fraud for failing to recognise a result that the test was not designed to test.

The third point (and one where I do object to IzzyR’s position) is that Randi does not anywhere on the quoted page say that the 22% result for water dowsing is not significant or that it debunks all water dowsing or anything of the sort. Randi was doing a test of all sorts of dowsing (water, bronze, gold). The results for the day were 13.5% which means jack shit, and Randi goes on to discuss what really is the most interesting aspect which is the enormous disparity between what dowsers think they can do, and what they can actually do. Nowhere does he say “none of the results have any statistical significance”.

At the absolute worst, you could accuse Randi of a slight omission in failing to acknowledge the strong result for water.

Against that background, IzzyR, to say:

Randi has taken a test that strongly supports the validity of dowsing and misrepresented it as a debunking of dowsing. This calls his credibility into question.

is, of itself, a misrepresentation.

And to say, IzzyR:

This is just one instance, of course. And I am no Randi expert. But if he does this type of thing often then he is indeed a fraud.

is the worst sort of careless mudslinging. Because there are no other instances, and even the one instance you cite is pissweak.

There are plenty of dowsers out there who say they can achieve 80-90% on a pipe test. All they have to do, if their claims are genuine, is approach Randi and say that they can achieve 22% on a 10% test, leaving themselves a huge margin for error[sup]*[/sup]. If/when Randi designs a test with a suitable number of iterations to test that claim, and if/when the dowser passes but Randi fails to pay then he’ll be a fraud.

Alternatively, if Randi was to notice that water dowsers had a pattern (over a long period of testing) of achieving better than chance (but less than their own claims) I would be disappointed in him if he failed to report that.

But there is no evidence of either of these things.

[sup]*[/sup][sub]It’s kinda funny that no dowser has ever done this. I wonder why that could be…[/sub]

[Musicat]

Originally posted by IzzyR
Randi has taken a test that strongly supports the validity of dowsing and misrepresented it as a debunking of dowsing. This calls his credibility into question.

To boil down what Princhester is saying,
[ul][li]Claimant agreed in advance that to do X would be a success, and NOT do X would be a failure.[/li][li]Claimant did NOT do X.[/li][li]Therefore he failed. Q.E.D.[/ul][/li]
If anyone’s credibility is in question, it would be the claimant’s.

[Princhester]

Actually Musicat you miss the crux of my post, which is that while testing dowsers on their claims, incidentally Randi’s Australian testing threw up a significant (but far from conclusive) result in favour of water dowsing, and Randi does not mention this. There is no point in pretending otherwise. However:

[ul][li]Randi’s test was designed to test and debunk the dowsers specific claims, which it did, spectacularly. That was his focus[/li]
[li]There have been lots of other dowsing tests which do not show favourable results consistently and against that background Randi probably did not think the favourable but not conclusive result of this particular test worth mentioning.[/li]
[li]Randi did not say that his testing disproved dowsing, or that the favourable water testing results were insignificant.[/ul][/li]
Therefore, to suggest that Randi was being a liar, or fraudulent, is ridiculously over the top.