VFF to Appear on Rational Alchemy

[Farencue]

Why would anybody want to interview her at this point anyway?

More importantly, why does it need to be recorded, cleaned up, edited and posted on her website? Is there a demand for this that I've missed?

Oh yeah, the believers will want to hear all about how the skeptics can't prove there is no ability.

It needs to be cleaned and edited in case anyone asks her about Alenara the Breatharian. The Rational Alchemy podcast edited that bit out at her request.

 

[Akhenaten]

Oh yeah, the believers will want to hear all about how the skeptics can't prove there is no ability.

It needs to be cleaned and edited in case anyone asks her about Alenara the Breatharian. The Rational Alchemy podcast edited that bit out at her request.

Well, we wouldn't want anyone thinking that VisionFromFeeling was anything other than the latest and brightest rising star of the skeptical community, now would we?

 

[Audible Click]

Well, we wouldn't want anyone thinking that VisionFromFeeling was anything other than the latest and brightest rising star of the skeptical community, now would we?

Must be that whole Arcturus schitck.

 

[Farencue]

well, we wouldn't want anyone thinking that visionfromfeeling was anything other than the latest and brightest rising star of the skeptical community, now would we?

Yep fig jam, I know

 

[KoihimeNakamura]

 

[Ashles]

Whatever Anita's methodology was, when all was said and done she achieved a result that was very nearly statistically significant. Not a spectacular result, but arguably worthy of a re-test.

Wow this is amazing!

"Very nearly statistically significant"

You know what is another way of saying "very nearly statistically significant"?

NOT statistically significant.

The standard recognised probability for significance is there for a reason. It sets the absolute minimum accepted point something should be agreed as "interesting". The minimum, not "well anywhere roughly around here will do statistically"...

By DEFINITION anything below this is not statistically significant or interesting.
Not near it, or close to it... ANYTHING below it.
That's kind of the whole point of it being there!
It is already the minimum point before we can be confident chance is the overriding factor.

Otherwise you get into ridiculous situations like, er, this, with people desperately scrabbling to ascribe significance where there is none.

Hell let's just have no-one fail anything. Because surely ANY results are almost, nearly, close to, not that far from the minimum accepted point of statistical significance.

I just wish some people would be a little more honest and say "Look I just want to believe okay?"

No wonder paranormal claimants have such an easy ride of it everywhere where even on skeptical sites people do not understand statistical significance or the concept of encouraging them to repeatedly try any paranormal claim over and over and over again until they get a hit.

Tell you what - I can predict coin tosses.

I'm going to do it right now ten times.

I predict H, H, H, H, H, H, H, H, H, H.

Actual coin tosses: T, H, H, H, T, T, T, H, T, T.

Look at that! In the first 4 I got 3 out of four correct in a row! What are the odds of that?
And I knew I was going to get those correct - I felt really confident.

Unfortunately I got fatigues after that.

I assume certain people here would find my results 'interesting' and 'nearly statistically significant' so I should be encouraged to continue testing myself.

Yes?

 

[Rodney]

The standard recognised probability for significance is there for a reason. It sets the absolute minimum accepted point something should be agreed as "interesting" . . . Yes?

No.

 

[Ashles]

No.

Otherwise you get into ridiculous situations like, er, this, with people desperately scrabbling to ascribe significance where there is none.

I just wish some people would be a little more honest and say "Look I just want to believe okay?"

QED.

 

[Uncayimmy]

Wow this is amazing!

"Very nearly statistically significant"

You know what is another way of saying "very nearly statistically significant"?

NOT statistically significant.

The standard recognised probability for significance is there for a reason. It sets the absolute minimum accepted point something should be agreed as "interesting". The minimum, not "well anywhere roughly around here will do statistically"...

By DEFINITION anything below this is not statistically significant or interesting.

To be fair the "standards" exist primarily from the days of yore when you had to look up stuff in printed tables, and they had limited values they would print. The common values are 0.05, 0.01 and 0.001. It's simply a matter of opinion that 0.05 is significant and 0.051 is not, but you have to put the cut-off somewhere.

Rodney likes to forget that you set the cut-off in advance, not after the fact. In this case the cut-off was 0.00057 (1 in 1,728). This is reasonable for looking to see if a claim of near perfect accuracy accomplished by means unknown to science has merit. Extraordinary claims require extraordinary evidence.

The lowest standard, 0.05 (1 in 20), can be useful in medical studies. We know the body is a complex system and not everyone responds the same way. So, if we're researching a medicine that has a plausible mechanism (at the very least a chemical is physically entering the system) we probably don't want to toss out potential useful results in our quest to eliminate false positives. It's a fair cut-off, but even then we still have to evaluate the potential benefits and risks and our confidence in those effects. Part of our confidence comes from understanding the mechanism involved. If the medicine is not harming people and might provide a powerful benefit, that's all the more reason to proceed despite the fact that it could just be chance.

When the claim is that a person can see inside the human body down to the atomic level, we know that's just ridiculous. There is absolutely nothing known to science that says this is even remotely possible. Our risk, then, is of false positives rather than false negatives, so we set the bar much higher than we might for (say) a study on aspirin and heart attacks.

Also, when you set up a study, you have to consider the possible outcomes in relation to your cut-off. In this challenge (it wasn't a study), the three possible outcomes were 0.22 (1 in ~4), 0.019 (1 in ~50), and 0.00057 (1 in 1,728). No other values were possible. They could have used the 1 in 20 bar, but for reasons given, they didn't. You could argue that they actually chose the 1 in 1,000 bar - her only way of beating it was to get all three correct.

If they wanted to use the 1 in 100 bar, they could have done six trials and required three correct. That would have been 0.009, and they could have said that's close enough to 0.01. The would have said that in advance, of course.

Rodney's numbers are crap. Anita's results had a p-value of 0.22, which is miles away from the lowest standard, 0.05. He got his number by looking at things that were not tested and ignoring relevant data that negatively impact his number. It's bad enough finding signficance in things that weren't tested, but it's even worse when you cherry pick the data used to form your conclusion.

 

[Moochie]

I can't help thinking that the Rodneys of this world don't consider claims like VfF's extraordinary, hence their inability to get their craniums around the concept.


M.

 

[ynot]

This debate wouldn’t exist if the IIG test had been stopped as soon as VFF failed.

 

[Uncayimmy]

This debate wouldn’t exist if the IIG test had been stopped as soon as VFF failed.

LOL! There's another thread on that. Suffice it to say that there still would have been spin. This is not the first "test" where she has attempted to snatch victory from the jaws of defeat.

 

[tsig]

LOL! There's another thread on that. Suffice it to say that there still would have been spin. This is not the first "test" where she has attempted to snatch victory from the jaws of defeat.

Nor the first time she got an F and tried to convert it to an A.

 

[Rodney]

Rodney's numbers are crap. Anita's results had a p-value of 0.22, which is miles away from the lowest standard, 0.05. He got his number by looking at things that were not tested and ignoring relevant data that negatively impact his number.

My numbers are accurate based upon Anita's claim. Specifically, she claimed "to be able to detect which Subject in a group of six Subjects is missing a kidney, to further identify which kidney (left or right) is missing in her selected Subject, and to be able to do this with 100% accuracy in three consecutive trials." See http://iigwest.org/anitaikonenprotocol.html

So, while Anita failed to do what she claimed, she specified a two-step process: (1) detect the Subject who is missing a kidney, and (2) identify which of the Subject's kidneys is missing. She did not specify a one-step process; i.e., identify the missing kidney among 12 kidneys. It therefore is illogical to combine the two steps into one when evaluating the probability. Show a statistician the IIG protocol and my analysis in Post #425, and see whether s/he agrees with you or me.

 

[Uncayimmy]

My numbers are accurate based upon Anita's claim. Specifically, she claimed "to be able to detect which Subject in a group of six Subjects is missing a kidney, to further identify which kidney (left or right) is missing in her selected Subject, and to be able to do this with 100% accuracy in three consecutive trials." See http://iigwest.org/anitaikonenprotocol.html

So, while Anita failed to do what she claimed, she specified a two-step process: (1) detect the Subject who is missing a kidney, and (2) identify which of the Subject's kidneys is missing. She did not specify a one-step process; i.e., identify the missing kidney among 12 kidneys. It therefore is illogical to combine the two steps into one when evaluating the probability. Show a statistician the IIG protocol and my analysis in Post #425, and see whether s/he agrees with you or me.

Rodney, this was discussed to death in another thread. You were as wrong then as you are now, and I say that with a very high confidence level. If you bring it up again in the correct thread, don't expect much more than seeing me refer to posts where your arguments were sliced, diced, dried, ground up, and sprinkled over a big ole bowl of Kellogg's Woo Flakes.

 

[bookitty]

Of the three tests I've seen (C. Sonne, R. Traynor, and A. Ikonen) C. Sonne got zero correct, R. Traynor got zero correct and A. Ikonen got one correct.

There have been many challenges and I am not familiar with all of them. How many others have gotten one or two right? Who got closest to statistical improbability?

 

[Uncayimmy]

Of the three tests I've seen (C. Sonne, R. Traynor, and A. Ikonen) C. Sonne got zero correct, R. Traynor got zero correct and A. Ikonen got one correct.

There have been many challenges and I am not familiar with all of them. How many others have gotten one or two right? Who got closest to statistical improbability?

Gettting "one or two right" is meaningless unless you know the protocol. Connie Sonne had about a 34% chance of getting one correct, and her test was properly blinded. Anita had a 25% chance of getting one right, but her test allowed for plenty of leakage and that number is based on proper blinding. A test with Zener cards (five different cards) will result in the person getting ~20% correct by chance alone. Your question is meaningless as asked.

And guess what? If we run a series of tests where there was a 1 in 1,000 chance of passing, once we crossed the 700 tests mark, the odds are in favor of somebody beating those "improbable" odds. If we use the absurdly low threshold of 0.05 (Rodney's favorite), then once we've run 14 tests, the odds are in favor of someone beating the mark. At the 1 in 100 threshold, it only takes 69 tests before the odds are in favor of the claimant.

We have to look at the claims. Connie Sonne would have been 100% accurate if the cards were laid down face up - anybody with at least one good eye would. The sonogram lady was 100% accurate using her gear to detect missing kidneys, and if VFF could do what she claimed (and demonstrated before the formal test), she'd be 100% accurate. Regen Taylor would have been 100% accurate if he were allowed to use a cell-phone to tell the receiver what he saw.

This is why repeated testing of these claims is so ridiculous. If VFF runs enough "tests," she'll beat 1 in 20 odds pretty quickly on one of those tests, and we'll never hear the end of it. The fact is she has run a bunch of "tests" already and failed to demonstrate anything even remotely interesting in any one of them.

The only way to gauge her "success" in more kidney tests is count all of the trials as one test. The proper way to do it is set the number of trials in advance, otherwise you run the risk of early stopping. Suppose she did 12 trials total. She'd need to get a total of 6 correct to beat 1 in 1,000 odds. That means she'd need to beat 1 in 2,638 odds on the next 9 trials. During that process she will have a 54% chance of getting another one correct, and we know how that would turn out.

 

[bookitty]

Gettting "one or two right" is meaningless unless you know the protocol. Connie Sonne had about a 34% chance of getting one correct, and her test was properly blinded. Anita had a 25% chance of getting one right, but her test allowed for plenty of leakage and that number is based on proper blinding. A test with Zener cards (five different cards) will result in the person getting ~20% correct by chance alone. Your question is meaningless as asked.

And guess what? If we run a series of tests where there was a 1 in 1,000 chance of passing, once we crossed the 700 tests mark, the odds are in favor of somebody beating those "improbable" odds. If we use the absurdly low threshold of 0.05 (Rodney's favorite), then once we've run 14 tests, the odds are in favor of someone beating the mark. At the 1 in 100 threshold, it only takes 69 tests before the odds are in favor of the claimant.

We have to look at the claims. Connie Sonne would have been 100% accurate if the cards were laid down face up - anybody with at least one good eye would. The sonogram lady was 100% accurate using her gear to detect missing kidneys, and if VFF could do what she claimed (and demonstrated before the formal test), she'd be 100% accurate. Regen Taylor would have been 100% accurate if he were allowed to use a cell-phone to tell the receiver what he saw.

This is why repeated testing of these claims is so ridiculous. If VFF runs enough "tests," she'll beat 1 in 20 odds pretty quickly on one of those tests, and we'll never hear the end of it. The fact is she has run a bunch of "tests" already and failed to demonstrate anything even remotely interesting in any one of them.

The only way to gauge her "success" in more kidney tests is count all of the trials as one test. The proper way to do it is set the number of trials in advance, otherwise you run the risk of early stopping. Suppose she did 12 trials total. She'd need to get a total of 6 correct to beat 1 in 1,000 odds. That means she'd need to beat 1 in 2,638 odds on the next 9 trials. During that process she will have a 54% chance of getting another one correct, and we know how that would turn out.

So what about all the other tests over the years? Is there a good place to look up the "scores" for other claimants?

 

[Uncayimmy]

So what about all the other tests over the years? Is there a good place to look up the "scores" for other claimants?

The MDC Forum is probably the best place to look, but I don't think there's a table listing them all. Besides, not all tests rely on stats. For example, there was a lady who believed she could make Jeff Wagg pee within a set period of time. I hear she's gonna be on Montel's podcast soon.

The threshold for a preliminary test usually requires beating at least 1 in 1,000 odds. Nobody has done that. As I explained earlier, in some cases it's not really possible to even "come close." For example, in Anita's test as designed the only three outcomes for getting something right were 1 in ~4, 1 in ~50 and 1 in 1,728. For Connie it was 1 in ~4, 1 in ~35, and 1 in 1,000.

Remember, these are challenges to put up or shut up spin the results. They are not experiments designed to find small effect sizes. Take migraines for example. The cause is not fully understood. The frequency of migraines for a single individual varies greatly. Migraines also go away on their own - people don't get a migraine that lasts until they die.

In order to determine if a prophylactic (prevent migraines) works, we need a lot of people over a long period of time because of all the variances. If we can find a medicine that reduces migraines by (say) 10% for 10% of sufferers, that could be useful. For me personally it would take years to say it's working.

The same thing goes for treatment when a migraine hits. If I am unable to do anything myself (say I'm hiking in the middle of nowhere), it will still go away on its own. If I can lay down in a dark place on my Chillow (highly recommended), it goes away faster. If I can pop some Excedrin, that helps as well. However, the duration and severity varies all on its own. It has taken me years to figure out what seems to work, and I am not 100% confident that it does. At least my self-treatments have a basis in reality.

Again, the point here is that you need a bunch of people keeping accurate reacords over a period of time to figure out if those treatments alter the duration and/or intensity by some percentage for some percentage of the people. The 0.05 threshold Rodney loves so much helps find possible effects and weed out potential bad side-effects.

Compare/contrast this to a poison to kill bugs. We spray it on a container of bugs and watch them die. We leave another container next to it and watch the bugs live. The effect is huge and immediate. We're not worried about "chance" here.

If you want to discuss test results in a meanginful manner, you need to understand the stats. Again, I recommend the Cartoon Guide to Statistics as required reading.

 

[Rodney]

The fact is she has run a bunch of "tests" already and failed to demonstrate anything even remotely interesting in any one of them.

Excepting the IIG test, what tests have been run regarding Anita's alleged paranormal abilities?