Cold Reading Demos at TAM2

[BillHoyt]

[BWhat it depends on is the previously-known fact that John Edward tends to rattle off a list of names with the same first letter. [/B]

Uh-huh. And the data show, very clearly, that that tendency is greatly biased toward "J'. Not M. Not X. Not B or C or D. Not any of the other 25 letters in the alphabet but "J".

If you believe this is artifactual, please explain it statistically.

 

[69dodge]

Originally posted by BillHoyt
It would not be more correct to say "can switch". One must because, if the sub-chains begin to dominate the overall process, the distribution becomes normal.

I do not follow you at all. A Poisson distribution with large mean is very nearly identical to a normal distribution. If two distributions are nearly identical, it doesn't much matter which one you use.

In any case, this was not the main point of my post.

 

[69dodge]

Originally posted by BillHoyt
Uh-huh. And the data show, very clearly, that that tendency is greatly biased toward "J'. Not M. Not X. Not B or C or D. Not any of the other 25 letters in the alphabet but "J".

If you believe this is artifactual, please explain it statistically.

I'm not sure exactly what you're claiming the data show very clearly, but I'm pretty sure it's not the same thing as what your significance test shows.

The null hypothesis of your significance test is:<blockquote>The guesses are independent and, for each guess, the probability that it's a 'J' is equal to the relative frequency of 'J' names in the general population.</blockquote>You've provided evidence against this null hypothesis. You've provided no evidence against the claim that the only thing wrong with the null hypothesis is the assumption of independence. Doing that would require a different significance test.

(What is the point of this whole discussion anyway? I guess some participants or lurkers might increase their understanding of statistics, so it's not entirely pointless. I doubt a significance test with a p-value of 1/20 or so will change anyone's mind either way about the existence of souls; nor should it, really, if their prior opinion was based on more evidence than that.)

 

[BillHoyt]

I'm not sure exactly what you're claiming the data show very clearly, but I'm pretty sure it's not the same thing as what your significance test shows.

The null hypothesis of your significance test is:<blockquote>The guesses are independent and, for each guess, the probability that it's a 'J' is equal to the relative frequency of 'J' names in the general population.</blockquote>You've provided evidence against this null hypothesis. You've provided no evidence against the claim that the only thing wrong with the null hypothesis is the assumption of independence. Doing that would require a different significance test.

(What is the point of this whole discussion anyway? I guess some participants or lurkers might increase their understanding of statistics, so it's not entirely pointless. I doubt a significance test with a p-value of 1/20 or so will change anyone's mind either way about the existence of souls; nor should it, really, if their prior opinion was based on more evidence than that.)

69dodge,

I think you have lost track of what a null hypothesis is. It is a logical inversion used to rule out possibilities. The null hypothesis here is very simple: are JE's guessess distinguishable from a random sampling from the census data. How exactly they differ is not yet relevant. Why they differ is not yet relevant. That they differ is the only relevant question here. In two different tries, done by two different people, both at the .05 significance level, this test has rejected the null hypothesis.

 

[Clancie]

Bill,

Posted by BillHoyt

In two different tries, done by two different people [i.e. Lurker and BillHoyt], both at the .05 significance level, this test has rejected the null hypothesis.

False, and you know it. In fact, it is an extremely audacious claim, Bill, since in order to make it, you had to intentionally omit Lurker's actual conclusion. Here is the full content of his post (the bolded comments are the conclusion that you deceptively left out ).

Originally posted by Lurker

...I did the count on "J" guesses versus non-"J" guesses for JE. I also did it under two criteria. The first is Bill Hoyt's method while the second was the method Thanz and I employed.

Here are the results of the counts and what Poisson tells us:

Hoyt:

9 J guesses
10 non-J guesses

Poisson says reject null hypothesis

Lurker

2 J guesses
5 non-J guesses

Poisson says does not reject null-hypothesis

It appears the counting scheme again is paramount in the result.

Care to comment?

 

[BillHoyt]

Thanz, Kaffee Klatch,Interesting Ian,

if your next posts do not squarely, directly, accurately and insightfully address all the issues I just raised, I am done with you woo flies. I will be busy spending my time sharpening mallets for the next round of whack-a-woo.

If you're not PWI again, Ian, why not take a stab at it?

 

[BillHoyt]

Bill,

False, and you know it. In fact, it is an extremely audacious claim, Bill, since in order to make it, you had to intentionally omit Lurker's actual conclusion. Here is the full content of his post (the bolded comments are the conclusion that you deceptively left out ).

Care to comment?

Yes, you once again have engaged in the straw man fallacy. I didn't comment on Lurker's conclusions, kaffee klatch, I reported the results of his running my method on the other data set. How disingenuous of you. But you also failed to fulfill my request to directly address my points, and chose instead to try a straw ploy. I'm done with you, klatch.

Next? Thanz? Ian?

 

[Clancie]

Bill,

Face it, you lied. And you lied to bolster your argument by changing someone else's post--doubly deceptive. You know your "J" count is wrong and you just don't want to admit it.

It was a lie, Bill, just a few posts above this, when you said that you and Lurker had reached the same conclusion about the J count and Poisson. It was a lie when you said Lurker agreed that "the test rejected the null hypothesis."

That was -clearly- a lie, Bill (as you obviously would know). You lied to make it look as if someone else did a count and agreed with your conclusion. How dishonest can you get!!!

In fact, as you know, Lurker said exactly the opposite of what you attributed to him, as I have shown you by providing the -exact- quote (a conclusion which you dishonestly left out when you quoted him).

Care to make any further comment about this?

 

[CFLarsen]

Face it, you lied.

Talk about the pot and the f*cking kettle....

 

[BillHoyt]

Thanz, Ian,

Will either of you directly address the issues as I outlined them? Or would you prefer to participate in klatch's latest lie campaign?

 

[Interesting Ian]

Talk about the pot and the f*cking kettle....

I see no need for that sort of language. What a hypocrite you are for complaining about my language
The worst words I used last night were stupid and moron.

 

[Clancie]

Posted by Bill Hoyt

Or would you prefer to participate in klatch's latest lie campaign?

rofl, Bill.

Nothing to say in your own defense for lying about Lurker's real point, eh?

Well, yes, I suppose diversion and insults -are- the best you have to offer, in that case.

Sad.

 

[CFLarsen]

I see no need for that sort of language. What a hypocrite you are for complaining about my language
The worst words I used last night were stupid and moron.

Drunk again? Apparently, you are unable to read.

 

[Interesting Ian]

Drunk again? Apparently, you are unable to read.

What am I unable to read precisely? What I'm asking is why are you complaining about me using the word stupid while you are using the word f*ck??

 

[69dodge]

Originally posted by BillHoyt
I think you have lost track of what a null hypothesis is.

I trust you won't be too surprised to learn that I disagree.

It is a logical inversion used to rule out possibilities. The null hypothesis here is very simple: are JE's guessess distinguishable from a random sampling from the census data. How exactly they differ is not yet relevant. Why they differ is not yet relevant. That they differ is the only relevant question here. In two different tries, done by two different people, both at the .05 significance level, this test has rejected the null hypothesis.

That sounds ok. Naturally, you can test whatever null hypothesis you want to. But why do you consider this particular null hypothesis relevant?

Presumably, it's because you think JE's guesses would be indistinguishable from a random sample from the census data, if he were in fact communicating with the dead. I see no reason to think that, however, given the dependence between his guesses but not between the elements of a random sample.

 

[CFLarsen]

What am I unable to read precisely? What I'm asking is why are you complaining about me using the word stupid while you are using the word f*ck??

Because I don't use it to describe other people, Ian.

Lay off the booze. Please.

 

[Interesting Ian]

Because I don't use it to describe other people, Ian.

Lay off the booze. Please.

Why on earth should I lay off the booze?? I like going out on a Friday night with friends on a pub crawl. This is what people do in the UK.

I suggest you mind your own business. If you do not like my rudeness then do not read my posts. It's as simple as that.

 

[CFLarsen]

Whatever. I'm sure you're a big boy, Ian.

 

[BillHoyt]

Bill,

Face it, you lied. And you lied to bolster your argument by changing someone else's post--doubly deceptive. You know your "J" count is wrong and you just don't want to admit it.

It was a lie, Bill, just a few posts above this, when you said that you and Lurker had reached the same conclusion about the J count and Poisson. It was a lie when you said Lurker agreed that "the test rejected the null hypothesis."

That was -clearly- a lie, Bill (as you obviously would know). You lied to make it look as if someone else did a count and agreed with your conclusion. How dishonest can you get!!!

In fact, as you know, Lurker said exactly the opposite of what you attributed to him, as I have shown you by providing the -exact- quote (a conclusion which you dishonestly left out when you quoted him).

Care to make any further comment about this?

Retract your accusations. Now. They are groundless and based in the most stupendous ignorance. I said nothing about conclusions. I said "rejected the null hypothesis." I included that in the lurker quote just so fools such as you would not jump into this confusion between "rejecting the null hypothesis" and drawing a conclusion. I never said Lurker agreed with me. I never said he agreed with my conclusion. I have always said his analysis of different data using my technique also rejected the null hypothesis.

I await your retraction. No other post will be responded to.

 

[BillHoyt]

IPresumably, it's because you think JE's guesses would be indistinguishable from a random sample from the census data, if he were in fact communicating with the dead.

But that is exactly the question before us! That is exactly the claim of mediumship. And that is exactly why the first null hypothesis is so constructed.

I see no reason to think that, however, given the dependence between his guesses but not between the elements of a random sample.

And this would naturally be the next sets of tests. With two data analyses both rejecting the null hypothesis that JE's names gueses are indistinguishable from population names, then we must ask: why? That question would call for more data and more tests. Either that or getting him into a real experimental setting sans the clowns who've tested him in the past.