He didn’t establish significance at all. That’s the point. That’s why I object.
Trying to find the data online, but not sure exactly what to search for. Is this the test in which it was said: " Even Arthur C. Clarke now understands this; he had been misinformed."???
If so, you need to provide the raw data immediately, because it’s starting to look like you’re making shit up.
Care for an answer? All right
At an alpha of 0.01, you would expect to see the results observed about one time in a hundred by pure chance. There are, at last count, about seven billion people on the earth.
If all of them tried to dowse tomorrow, how many would you expect to get the results you are making such a big deal about by pure chance?
I’m trying to pin a certain person down to exactly what he claims. He keepos on wriggling and dodging.
As soon as he says what he means I will post my cite. Until then, he has lost.
Wait, wait, wait, after an mainly fruitless internet search, I found the following:
“Searching for gold and for water, with a ten per cent chance of getting
the right answer, the dowsers claimed they would get 80 per cent
success. In practice they achieved an overall success rate of 12 per
cent, combining the gold test and the water test. But on the water test
alone they scored 22 per cent - well above chance. The results of the
two tests should not have been combined.”
Based on this you pulled the 35% number from out of nowhere. Assuming you are familiar with this test, it’s fairly clear that you are making up facts to support your case.
Again, provide the raw data!
Actually, the only person who has made a claim here is you. You have asserted there are flaws in Randi’s methods on a certain, unidentified experiment. You haven’t told us what they are. You haven’t cited to the statistics that, you argue, shows that the test subject was clearly able to dowse, because his results were so unusual that if you tested a hundred people, ONLY one would get a result of that level of significance by pure chance. And yes, you are arguing Randi should pay out a million bucks for a result that would happen by pure chance about one time in a hundred. They must love you in vegas.
Yes, obviously I have lost.
Wait, you can’t provide the raw data, you aren’t sure what the confidence interval is, you don’t know what the mean or standard deviation are, and based on this you decide that someone else was lying? I am very not impressed right now.
ETA: Also, I’ve proved that one in a hundred blind monkeys has dowsing powers. That alone made this conversation worthwhile.
Don’t forget–he has taught me, at least, to reject any mathematical arguments from anyone without formal training in math. He says so himself.
And that is **great **advice.
Quick–Get me a hundred blind monkeys! I’ll send them to Randi, and we can split the million 50-50.
When one plays with one’s self, the only loser is…
Won’t work. That son-of-a-bitch evidently expects them to score 80%, the bastard.
Only solution: more monkeys!
I hesitate to get involved in this after the anti-Randi has been invoked, but: it depends on what you mean by “passing”, which in will in turn affect the statistics. In the specified test, the participant has demonstrated that they can perform a test at significantly better than chance, and has “passed” if we consider passing to be “demonstrating significantly better than chance” the criteria we’re interested in, and assuming they met the other requirements for statistical success (number of trials, double blinding, etc.).
However, assuming you’re talking about the 22% dowsing incident from the old Arthur C. Clarke’s “World of Strange Powers” show, that’s not what happened.. The test wasn’t even for water dousing, it was for dousing for a large number of different substances, and clearly failed. Someone later took the water-detection rates alone and noted they were better than the chance rates for detecting everything, even though the water searches were too few in number for statistical significance by themselves. In other words, they took the criteria for passing a test A which requires, say, 100 trials, and noticed that 5 of those trials, if considered in a vacuum, would have passed the test if the same results had been obtained over 100 trials, and declared success.
We call that cherry picking, and it’s exactly the type of stuff that tests of the paranormal (Randi’s or any other scientist’s) are designed to prevent, but that pseudoscientists absolutely go gaga over: it’s a variation of the optional stopping and starting during ESP card tests where you note that during the test you got a sequence of, say 5 cards right, and claim you were 100% accurate over 5 cards, (much better than chance!) even though that sequence was in the middle of 100-card run, and the entire run gave you a much lower accuracy.
It’s this sort of stuff that drives scientists bonkers (well, along with the “why won’t you test every whacko claim for the hundredth time? you’re just close-minded.”) The pseudoscientists look at the 5-card run and are absolutely convinced that they’ve demonstrated something remarkable, and no amount of arguing will convince them otherwise. Double-blind, controlled clinical trials have a complicated set of rules for precisely this reason – even folks trained to avoid confirmation, selection, and survival bias will nevertheless “see” patterns that don’t exist. But all those “rules” are seen by those who don’t understand their need as just a bunch of “extraneous whining” that’s there to keep their amazing ability from being accepted by the scientific world. They know they’re right. But the certainty with which one believes something doesn’t affect it’s truth or falsehood; and that’s why we have the math.
But that same cherry-picking-after-the-fact is why it’s necessary that the person being tested agree ahead of time to the success/failure criteria and to how the test is being run. In your specified case, the tester made a much less supportable claim: not that they could do better than chance, but that they could do much better than chance (which, incidentally, requires a smaller number of tests to determine). They didn’t do what they claimed, so if we’re interpreting “did what they claimed” as “passing,” then they didn’t pass.
Define “passing” however you like in this case. An actual instance of it with regard to paranormal activity would be interesting, but I haven’t seen one yet. And if the tested person feels that such a result is unfair, the correct thing to do is to only make a claim of “statistically better than chance” ahead of time, and just be confident they can over-achieve, rather than complain afterward that they should have been judged by some other mark.
I’ll shift them from project “Typewriter”–but that means you’ll just have to wait longer for the sixth act of Hamlet.
Yes i have.
I didn’t claim that.
Let me repeat - the dowsers beat James Randi because he is utterly fucking stupid. Not because dowsing is real.
Randi lied and claimed to have beaten them. He did so because certain sceptics believe hi9m and give him money for it.
because his results were so unusual that if you tested a hundred people, ONLY one would get a result of that level of significance by pure chance. And yes, you are arguing Randi should pay out a million bucks for a result that would happen by pure chance about one time in a hundred. They must love you in vegas.
Very interesting, and this is much different than the story Peter laid out earlier. I think we can call this case closed.
ETA: If Peter was aware of the facts in this case, then I think it’s clear that we can say his behavior in this discussion was deliberately deceptive and in bad faith.
I’m glad you agree. I see the beginning of wisdom.
Now, please apply the same to what Randi says.
no, I didn’t think you would. 
One last piece of information I dug up on this test (although thanks to Peter’s coyness, I’m not sure to which test exactly he’s referring):
“the Australian dowsers … attained less than their own
estimated success rate, so they did not win the prize - which at that
point was only US$10,000. You should know that the two Aussies who got
slightly over the expected success rate - not statistically significant,
but over expectation - were re-tried in Perth, Australia, a few weeks
later after I had left the country, and at that time they attained
results significantly below expectation. … even Arthur C.
Clarke now understands this; he had been misinformed.”
Where’s the difference?
With me the winner, as usual.
Since when is telling the truth deceptive?
I think it’s obvious that the Randi fan knows he’s wrong but is too much of a coward to admit it.
I wanted to understand the numbers in the Randi portion of this thread. The contract for paying out the cash was clearly upheld correctly. As to whether dowsing was established (a separate question), here’s what I can piece together:
Water, gold, and brass were all looked for, and the dowsers chose the material they wanted to go after. So, combining doesn’t seem too dirty at first blush. But, okay, let’s not combine. One of the three materials had a success rate that would happen about 1% of the time by chance. But there were three materials! .So, the chance of at least one of the sets of trials showing this level of significance is about 3%. So, even keeping these as separate trials leaves things at a 1-in-33 shot. Meh.
Also, the number of trials were: brass=26, gold=35, water=50. Why the weird numbers? I couldn’t find anything saying if these numbers were fixed in advance. I wonder if something like this didn’t happen – the water guy got a hot streak, so they kept going (until the streak seemed to have died), and the brass guy (who ended up with zero successes) was running cold, so they stopped early. Both of these sorts of actions would represent huge biases.
Not that one needs these biases to get any sleep. 1-in-33 ain’t all that crazy. (If I claim I can fly, would sufficient demonstration be something about as likely as rolling snake eyes in Monopoly?)
If you can’t tell the difference then you need to immediately stop commenting on these sort of tests
Telling the truth wouldn’t be deceptive. Your blunder was in not telling the truth.
Let me make a list of your deceptive behaviors in this thread:
[ol]
[li]You described the test in a way that had no relation whatsoever to the actual facts. Compare your description of the test to TimeWinder’s description of the test. Strangely, your description of the test was much more favorable to your case than what actually happened: probably just a coincidence, that.[/li][li]You claim that Arthur C. Clarke said the results were statistically significant, but in fact Clarke was working from bad data and when he discovered that the data was bad he withdrew his claim–something that would be important to mention, don’t you think?[/li][li]The tests were re-run immediately after the initial tests and the testees failed miserably. Also kind of important to mention, don’t you think?[/li][li]When questioned, you refuse to provide any detail about the tests. Considering the differences between what you said happened and what actually happened, I’m not surprised that you wouldn’t provide info.[/li][/ol]
I don’t give a shit about Randi, but I don’t tolerate being lied to. Calling you deliberately deceptive was the edited, polite version of what I’m actually thinking.
[quote=“Evil_Economist, post:99, topic:525955”]
If you can’t tell the difference then you need to immediately stop commenting on these sort of tests
Telling the truth wouldn’t be deceptive. Your blunder was in not telling the truth.
Let me make a list of your deceptive behaviors in this thread:
[LIST=1]
[li]You described the test in a way that had no relation whatsoever to the actual facts. Compare your description of the test to TimeWinder’s description of the test. [/li][/quote]
What description of the test is that? When did I misdescribe the test?
Actually, when did I describe the test at all. I gave a description of a HYPOTHETICAL test. Which is explicitly not the same.
[li]You claim that Arthur C. Clarke said the results were statistically significant, but in fact Clarke was working from bad data and when he discovered that the data was bad he withdrew his claim–something that would be important to mention, don’t you think? [/li][/quote]
When did he withdraw his claim? Give an exact reference.
Randi is a liar. He may claim that Clarke withdrew his claim in order to cover hs own ass. That does not make it true.
Find any cite where Arthur C. Clarke withdrew his claim. Until then, you have nothing.
And it’s not as though Clarke was the only one.
There is no difference, and I gave info.
yeah, I’ve encountered your type before. Giving accurate information is “lying” while lying on Randi’s behalf is honest. :rolleyes: