A Test for Sandsted

[Carl-NC]

I showed a simulation of 100 "guesses" to have the following results:

Hits: 0 1 2 3 4 5
----------------------------------
+/- 0: 74 24 2 0 0 0
+/- 1: 43 37 19 1 0 0
+/- 2: 25 40 27 6 2 0
+/- 3: 16 36 30 12 5 1

These numbers do not exactly match my calculated odds, which were estimations in the first place because of the overwhelming difficulty of doing exact calculations. So what would happen if we perform a second run of 100 guesses? Well, let's see... here's a second set of 100+1 simulations:

Hits: 0 1 2 3 4 5 6
---------------------------------------
+/- 0: 76 23 1 0 0 0 0
+/- 1: 45 39 10 5 1 0 0
+/- 2: 33 38 15 12 0 2 0
+/- 3: 21 37 22 13 4 2 1

There are some things of interest... first, there is a success of 6 guesses at the +/-3 error level, which is the threshold for Sandsted. Second, while some of the numbers are pretty close to the first run (76-23-1 vs 74-24-2 for exact guessing) some are radically different. Notice in the first run that for +/-2 there were 2 correct guesses 27 times, but in the second run this only happened 15 times.

Why is this so, and which number is "correct"? Statistically, they are both correct. This is what happens when you look at a relatively small sample of data... there can be anomalies that skew our perception of what's going on. Ferinstance, in the first set there were no successes of 6 correct. Should I therefore assume that it would be impossible to guess correctly 6 times? Of course not! This is why a simple "pre-test" in which a dowser purely guesses is a poor benchmark... it does not provide nearly enough information as to how likely particular outcomes are. And this is why we use statistical calculations.

If we combine the data from these 2 runs, we will then have 200 simulations which will provide a more accurate picture of the distributions. But let's go even further... let's combine 10 runs, or 1000 simulations:

Hits: 0 1 2 3 4 5 6 7
----------------------------------------------------------------------
+/- 0: 76.00 21.30 2.50 0.10 0.10 0.00 0.00 0.00
+/- 1: 46.60 36.30 13.50 3.00 0.60 0.00 0.00 0.00
+/- 2: 30.00 37.60 21.80 8.10 2.20 0.30 0.00 0.00
+/- 3: 19.20 32.60 28.80 13.1 5.50 0.60 0.20 0.00

Numbers are percentage levels. Now you can see that in 1000 sims, there were 2 sims in which 6 hits occured at the +/-3 level. So you can see that even though one 6-hit occured in the first 200 sims, only one more occured in the next 800 sims.

Back to my calculations:

Hits: 0 1 2 3 4 5 6 7
----------------------------------------------------------------------
+/- 0: 77.63 19.91 2.30 0.16 0.01 0.00 0.00 0.00
+/- 1: 45.86 37.18 13.57 2.93 0.42 0.04 0.00 0.00
+/- 2: 26.31 37.58 24.16 9.20 2.30 0.39 0.05 0.00
+/- 3: 14.61 30.98 29.57 16.73 6.21 1.58 0.28 0.03

These numbers correlate pretty well with the cumulative 1000 sims, though some of the numbers appear to be diverging a bit at the +/-3 level. This doesn't surprise me in the least, as the calculation become progressively more quirky as you allow more and more variance in the guesses. In any case, the calculations are more than sufficient as a statistical baseline to which dowsing can be compared.

Finally, here are the histograms for the 1000-run total:



Seeing the distributions as plots provides an easier way look at the trends, and helps show how means and spreads are shifting as we loosen up the test requirements.

- Carl

 

[Carl-NC]

Can you say yet, at what stage (or status) the real test is in?

Nope... haven't heard.

 

[Oroblanco]

SWR wrote:

Oro…wouldn’t it be much simpler to use the proper quote/reply protocol. The purpose of proper protocol is that is gives the reader all the information as to what has been quoted, and is forum etiquette as well

Is it just too much for you dowser types to try an play nice and use some etiquette for a change? The repeated abuse of quote/reply is what got your Amigo on ignore, and you are heading straight down that same highway

As a dog that returneth to his vomit, so is a fool that repeateth his folly ~ Ancient Hebrew Proverb

Is that easier for you to read SWR?

Too much to play nice and use some etiquette? No problem, I hope your fellow skeptics can do so though. You are sure free and welcome to "ignore" my posts.

Oroblanco

The way of a fool is right in his own eyes, but a wise man listens to advice. ~proverb

Do not reprove a scoffer, or he will hate you; reprove a wise man, and he will love you. ~proverb

 

[aarthrj3811]

Okay, be sure to read this carefully.... Getting 7 out of 10 targets by trying to guess them is nearly impossible. BUT....Sandy is not guessing, he's dowsing. This is why the coin test does not translate to guessing. Why is this so hard to understand?
Here's the problem with your logic, Art. If the number once could expect to get by guessing these dates and the number one should expect by dowsing the dates are the same, then how is dowsing better than guessing?
Gee Jean...Seems your numbers have a little problem with others...It's nice not to be alone...Art

 

[aarthrj3811]

..

Getting 7 out of 10 targets by trying to guess them is nearly impossible.

I GUESS that your program is WRONG and this statement comes from our favorite skeptic...Art

 

[aarthrj3811]

I've done this exercise a few times...it's not the easiest thing to do, but it is the only thing I believe I can do without having to leave my home. Anyway....best score (if you count within 3 years) out of 10 or 11 coins was probably 5 or 6. Worst score probably 2. I don't set this up for 10 coins a lot. Every once and a while since we've conversed about this test I just happen to see a coin laying there so I grab it and see if I can date it. Which...my accuracy I would believe to be 40%...can't really recall on those.
Well dowsing for coins...I don't have a whole lot of confidence in. I've gone it at probably a 40-60% success rate (when alone). I've dowsed three coins successfully to the year and some several in a row all fairly close in date (within 2 or 3 years). I can tell you it is more then coincidence. But as I've shown before, at least for myself, dowsing can't work under the conditions of a test such as this.

My reasoning for submitting to it is that I don't have to pay for anything, go anyway, or waste very much time.
Apprehension in my statements? No, though I will tell you coin dating is not the easiest thing in the world. A test such as this will validate nothing, neither disprove anything. It's Carl's decision if he wants to send the coins, I disagree with a test...but I loose nothing by submitting to this particular one...so...I submit.

It looks like you forgot to put Sandsteds FACTS in your generators. In case anyone is interested this is what Sandsted said he could do....Art

 

[ClonedSIM]

Okay, be sure to read this carefully.... Getting 7 out of 10 targets by trying to guess them is nearly impossible. BUT....Sandy is not guessing, he's dowsing. This is why the coin test does not translate to guessing. Why is this so hard to understand?
Here's the problem with your logic, Art. If the number once could expect to get by guessing these dates and the number one should expect by dowsing the dates are the same, then how is dowsing better than guessing?

This was my statement, Art. What do you find to be wrong with it?

 

[ClonedSIM]

..

Getting 7 out of 10 targets by trying to guess them is nearly impossible.

I GUESS that your program is WRONG and this statement comes from our favorite skeptic...Art

Same with this one, Art. What's the problem with it?

 

[aarthrj3811]

This is why the coin test does not translate to guessing

If the number once could expect to get by guessing these dates and the number one should expect by dowsing the dates are the same, then how is dowsing better than guessing?

If the coin test does not translate to guessing How can you say they are the same in the next sentence?

Getting 7 out of 10 targets by trying to guess them is nearly impossible.

I agree with you on that statement......Art

 

[ClonedSIM]

This is why the coin test does not translate to guessing

If the number once could expect to get by guessing these dates and the number one should expect by dowsing the dates are the same, then how is dowsing better than guessing?

If the coin test does not translate to guessing How can you say they are the same in the next sentence?

And now you see the problem. There's been much argument about how Sandy's coin test will only test his guessing ability. That looks fine on paper, but in fact he won't be guessing the targets. This is how the coin test does not equate to guessing.
In the same vein, when you told me that Sandy needs to only get three dates right in order to claim victory, then I asked you how dowsing could possibly be better than guessing when all you expect to get are results equal to guessing.

Getting 7 out of 10 targets by trying to guess them is nearly impossible.

I agree with you on that statement......Art

Great, so if Sandy gets 7 correct, we can both agree it was due to his dowsing ability?

 

[aarthrj3811]

Hey af1733....I have read Sandsteds post and base the results I expect from his words. You are basing your results on a math formula.

In the same vein, when you told me that Sandy needs to only get three dates right in order to claim victory, then I asked you how dowsing could possibly be better than guessing when all you expect to get are results equal to guessing.

The word better has a lot of meanings...When my rods cross on a target I know that something is there. If I were guessing thats what I would be doing and I would not be sure something was there. In this example Dowsing is BETTER than guessing. We are given statistics based on math formulas all the time. How many are right when real statistics are totaled up? All we have been given is the odds of what someones math formula says they should be. In other words ...It is a guess. I think Carl has it right...IT IS IMPOSSIBLE TO DISPROVE DOWSING.....Art

 

[ClonedSIM]

Hey af1733....I have read Sandsteds post and base the results I expect from his words. You are basing your results on a math formula.

In the same vein, when you told me that Sandy needs to only get three dates right in order to claim victory, then I asked you how dowsing could possibly be better than guessing when all you expect to get are results equal to guessing.

The word better has a lot of meanings...When my rods cross on a target I know that something is there. If I were guessing thats what I would be doing and I would not be sure something was there. In this example Dowsing is BETTER than guessing. We are given statistics based on math formulas all the time. How many are right when real statistics are totaled up? All we have been given is the odds of what someones math formula says they should be. In other words ...It is a guess. I think Carl has it right...IT IS IMPOSSIBLE TO DISPROVE DOWSING.....Art

But it is very possible to disprove individual claims...

When Sandy claimed to be able to dowse the individual dates on coins, he never once mentioned guessing, yet this thread was filled with dowsers who were outraged because they somehow got it into their heads that this coin test was only going to judge Sandy's guessing ability. Even you said this more than once, Art. Are you contradicting yourself now?

And as far as "real statistics" (whatever those are) go, maybe you'll understand when the coin dates are revealed and you can compare these to the guesses several of us made of the coin dates on whatever thread it was.

 

[aarthrj3811]

[size=10pt][size=10pt]Well dowsing for coins...I don't have a whole lot of confidence in. I've gone it at probably a 40-60% success rate (when alone). I've dowsed three coins successfully to the year and some several in a row all fairly close in date (within 2 or 3 years). I can tell you it is more then coincidence. But as I've shown before, at least for myself, dowsing can't work under the conditions of a test such as this.
My reasoning for submitting to it is that I don't have to pay for anything, go anyway, or waste very much time.
Apprehension in my statements? No, though I will tell you coin dating is not the easiest thing in the world. A test such as this will validate nothing, neither disprove anything. It's Carl's decision if he wants to send the coins, I disagree with a test...but I loose nothing by submitting to this particular one...so...I submit.[/size][/size]

Looks to me like Sandsted told us what to expect from the test. ....Art

 

[ClonedSIM]

[size=10pt][size=10pt]Well dowsing for coins...I don't have a whole lot of confidence in. I've gone it at probably a 40-60% success rate (when alone). I've dowsed three coins successfully to the year and some several in a row all fairly close in date (within 2 or 3 years).

And based on this Sandy agreed that dowsing ten coins with a target rate of +/- 3 years was acceptable.

I can tell you it is more then coincidence.

Telling us that he's not just guessing.

But as I've shown before, at least for myself, dowsing can't work under the conditions of a test such as this.

He has explained that dowsing tests that he's taken with skeptics have failed because of his nervousness when testing in front of other people. But the test at his home, in his own time, was one he felt he could do well with.

My reasoning for submitting to it is that I don't have to pay for anything, go anyway, or waste very much time.
Apprehension in my statements? No, though I will tell you coin dating is not the easiest thing in the world.

No apprehension stating he can expect to achieve a 40-60% success rate, even though it's not the easiest thing in the world.

A test such as this will validate nothing, neither disprove anything. It's Carl's decision if he wants to send the coins, I disagree with a test...but I loose nothing by submitting to this particular one...so...I submit.

[/size][/size]

By saying the test will validate nothing he is simply doing a nice little CYA job, but even if he disagreed with the test he still agreed to take it, so the results do have some validity, at the very least confirming his own self-administered test results of 40-60%.

 

[aarthrj3811]

Hey af1733....As a long time poker player I know how important the odds are. I also know that odds are just part of the game. The odds of poker are a real stastistic that has been proven over and over. Your odds are just a bunch of numbers produced by a number generator. No one knows if they are correct or not. Why would any one want to bet on odds that are not backed up be any kind of real facts....Art

 

[ClonedSIM]

Hey af1733....As a long time poker player I know how important the odds are. I also know that odds are just part of the game. The odds of poker are a real stastistic that has been proven over and over. Your odds are just a bunch of numbers produced by a number generator. No one knows if they are correct or not. Why would any one want to bet on odds that are not backed up be any kind of real facts....Art

Hey Art, what you've obviously failed to notice is that Sandy himself said he expected a 40-60% success rate when doing the test alone. This would handily beat the odds of guessing as defined by the charts that Carl generated using his "unreliable" computer program. So, if Sandy does get his 40-60%, then he's beaten the odds, regardless of what you define as "real facts."

If he gets less than 40%, then he did not live up to his own expectations and he did not beat the odds of random guessing. Coincidence?

 

[aarthrj3811]

I've done this exercise a few times...it's not the easiest thing to do, but it is the only thing I believe I can do without having to leave my home. Anyway....best score (if you count within 3 years) out of 10 or 11 coins was probably 5 or 6. Worst score probably 2. I don't set this up for 10 coins a lot. Every once and a while since we've conversed about this test I just happen to see a coin laying there so I grab it and see if I can date it. Which...my accuracy I would believe to be 40%...can't really recall on those.

Hey af1733...I see all kinds of charts and guesses as to what they mean. It seems like simple Math and that the numbers to work with are 2-4-5-and 6.,...with 4 being what he says is his average. This is a test of Sandsted not of random odds...It is not a test to see if Dowsing is better than some computer generated odds which I don't think are correct. You are all guessing at the out come and Sandsted is just doing what he does. Sandsted will find out if he is accurate or not. Carl may learn something but that will depend on how he views this test.....Art

 

[ClonedSIM]

It is not a test to see if Dowsing is better than some computer generated odds which I don't think are correct.

But, Art, you've never been able to say why the numbers are incorrect. Just feeling like something is wrong does not make it so.

You are all guessing at the out come and Sandsted is just doing what he does.

We're not guessing the outcome, Art. We're not doing anything at all to do with the outcome. What we are establishing is the number of correct responses Sandy will have to get correct in order to beat the odds of random chance.

For example: If Sandy gets just once date correct, well, that just wouldn't be very good. But, an average person guessing, with no dowsing experience or previous knowledge of the dates could be expected to get at least one right, as well.

If this happens, then Sandy's dowsing is just as good as an average person's guessing.

But if he gets 4 or 5 correct, then his dowsing has shown itself to be 4-5 times better than the average person's guessing. This is what all these numbers meant, Art. No one was trying to discount his results before they even came in. We were just trying to establish a baseline with which to view his final results.

Another way of saying it is this: Say Bob claims to be able to jump twice as high as anyone else on Earth. Okay, easy to check. Get a few average folks together and measure the height of their highest possible jump. Let's say this number is 20 inches. So Bob then jumps and the highest jump he can manage is 24 inches. True, he's jumping higher than the other people tested, but not twice as high, as he claims. So then we'll have to look at Bob and his physical prowess. He's probably an athletic guy, especially to make this kind of claim, so we take someone similar to his build and weight and have that person jump as high as they can. If they manage 23-24 inches, then Bob's claim is effectively disproven.

 

[aarthrj3811]

Another way of saying it is this: Say Bob claims to be able to jump twice as high as anyone else on Earth. Okay, easy to check. Get a few average folks together and measure the height of their highest possible jump. Let's say this number is 20 inches. So Bob then jumps and the highest jump he can manage is 24 inches. True, he's jumping higher than the other people tested, but not twice as high, as he claims. So then we'll have to look at Bob and his physical prowess. He's probably an athletic guy, especially to make this kind of claim, so we take someone similar to his build and weight and have that person jump as high as they can. If they manage 23-24 inches, then Bob's claim is effectively disproven.

Now you understand what I am saying....Use real facts...Not some formula that has no basic facts ....Art

 

[ClonedSIM]

Another way of saying it is this: Say Bob claims to be able to jump twice as high as anyone else on Earth. Okay, easy to check. Get a few average folks together and measure the height of their highest possible jump. Let's say this number is 20 inches. So Bob then jumps and the highest jump he can manage is 24 inches. True, he's jumping higher than the other people tested, but not twice as high, as he claims. So then we'll have to look at Bob and his physical prowess. He's probably an athletic guy, especially to make this kind of claim, so we take someone similar to his build and weight and have that person jump as high as they can. If they manage 23-24 inches, then Bob's claim is effectively disproven.

Now you understand what I am saying....Use real facts...Not some formula that has no basic facts ....Art

Oh, Art. If I remember correctly you were clamoring on and on about all these tests not being fair because there wasn't a "control" group. The numbers provided by Carl were, in essence, the control to Sandy's coin test. Since no one else claims to be able to dowse coin dates, we couldn't very well compare his claim to other dowsers. This left comparing his results to ordinary guessing. When guessing at anything, there is a certain percentage of guesses that will be correct based on the number of possible outcomes.

You can simplify this by looking at it like this. Rather than 40 possible years and getting the correct answer with +/- 3 year accuracy, let's say for a moment there are only 10 coins with 10 different dates, these dates being 1990-2000. You only have to pick one of those coins, and guess a date for it. The math works like this:

10 possible coins X 10 possible dates. 10 X 10 = 100 possible outcomes for only one guess. The odds that you will pick one coin from 10 and then guess the correct date for that one coin from 10 possible choices is 1 in 100.

Now, Sandy's test is nowhere this simple, unfortunately. He has 40 coins that he has to choose 10 from but, once he makes his choices, he only has to get the date right within a 7 year timespan. Essentially, if the coin is a 1975 cent, he can either get it perfectly correct, or dowse the date as 1972, 1973, 1974, 1976, 1977 or 1978 and still be considered correct. There's a whole lot more math involved with this one so I won't get into it, but can you see what I'm getting at?

We know the odds are 1 in 100 to get the date right in my first example. It's not a guess, and it is based on "real facts." This is the same math Carl used to extrapolate the chart you think is incorrect.