Cold Reading Demos at TAM2

[Thanz]

Didn't understand the explanation one bit, did you?

I think I understand just fine. I will explain it in my own words, and you can tell me where I go wrong.

In my count of the J guesses, they were about 21% of the total guesses. The number of J's in the total population is about 13%. With a relatively small N, this may not be significant. However, if we increase N, AND the number of J guesses remains at 21%, this goes beyond what we expect. We expect, with a larger N, for the number of J guesses to be closer to the mean, or 13%. Therefore, if it does not do this, the result is more significant. Have I got it right so far?

I'll stop here to make sure that I understand before moving on.

 

[BillHoyt]

I think I understand just fine. I will explain it in my own words, and you can tell me where I go wrong.

In my count of the J guesses, they were about 21% of the total guesses. The number of J's in the total population is about 13%. With a relatively small N, this may not be significant. However, if we increase N, AND the number of J guesses remains at 21%, this goes beyond what we expect. We expect, with a larger N, for the number of J guesses to be closer to the mean, or 13%. Therefore, if it does not do this, the result is more significant. Have I got it right so far?

I'll stop here to make sure that I understand before moving on.

So far, so good. But that wasn't where you had erred.

 

[Thanz]

So far, so good. But that wasn't where you had erred.

Great.

So, this is where I see your error - and where I think my numerical example above comes into play. Since it is on a different page now, I'll quote it again:

Let's make this a concrete example:

Reading 1:
JE: I am getting a "J" connection here.
Sitter: J?
JE: Yes, a "J" - like John, or Joe
sitter: I had an uncle Joe....

My method: one J guess.
BillHoyt:3? 4? J guesses?

Reading 2
JE: I am getting a "J" connection..
Sitter: My grandfather was John

Thanz:1 J
BillHoyt:1 J

Reading 3:
JE: I am getting a "Jim" connection here...
Sitter: Nope, I don't know any Jim
JE:What is the Canada connection?
Sitter: Blah blah

Thanz: 1 J
BillHoyt: 1 J

Reading 4
JE: I am sensing an older female
Sitter: My Mother has passed
JE: was her name "Jennifer"
Sitter: no, it was Roberta

Thanz: 1 J
Bill Hoyt: 1 J

I say that your counting method artificially increases both the J count and the sample size. Using these examples, your count weighs the first reading as equivalent to the other three readings combined. In the first, JE is groping for one J name. He is doing the same in the other three. But your method overcounts the first one - which makes it more significant in the long run, even if all are overcounted in the same manner.

Your count of the same data was approximately double of my count. Your count of J guesses was still about 21% of the total guesses, so that suggests that your counting method may indeed overestimate all guesses by the same factor (as you have claimed). However, we see that as sample size increases, and the proportion of J's remains the same, that proportion becomes more significant statistically speaking.

If you have artificially increased the sample size as I have said you have, the statistical analysis is incorrect - just as if you counted 100 heads for every flip of the coin. The artificial increase makes the proportion of J's seem more significant than it really is.

Therefore, you still need to explain why we should count reading 1 as equivalent to readings 2, 3 and 4 combined - as your counting method does. Unless you can logically explain that, your count is inflated and the inflation of both the count and the sample size makes the proportion of J's appear significant, even though the proportion of J to the total remains the same.

 

[BillHoyt]

I say that your counting method artificially increases both the J count and the sample size. Using these examples, your count weighs the first reading as equivalent to the other three readings combined. In the first, JE is groping for one J name. He is doing the same in the other three. But your method overcounts the first one - which makes it more significant in the long run, even if all are overcounted in the same manner.

My count weighs nothing.Your assertion of "artificially" has yet to be demonstrated.

Your count of the same data was approximately double of my count. Your count of J guesses was still about 21% of the total guesses, so that suggests that your counting method may indeed overestimate all guesses by the same factor (as you have claimed). However, we see that as sample size increases, and the proportion of J's remains the same, that proportion becomes more significant statistically speaking.

You misunderstand my claim about the statistics. You utterly missed the fact that, when you increase the sample size and the proportion of "J"s is close to the mean, it does not become more significant statistically. You are watching the tail effect and missing the significance of the pdf. The pdf is shifting shape, reflecting the law of large numbers.

You keep missing the fact that JE's choice of when to call out multiples and how many to call out is, itself, a Poisson process. That means it is random. That means the "J"s should have gravitated toward the mean. You keep tripping over this. The multiple Js, Ks, Ms, Zs, Qs, are all random. The effect of a consistent application of the method is to return the frequencies to their means.

Unless you can logically explain that, your count is inflated and the inflation of both the count and the sample size makes the proportion of J's appear significant, even though the proportion of J to the total remains the same.

This is statistics, not math. JE's choice of letter guesses is a Poisson process. My counting method reveals an underlying Poisson process. If this underlying process were random, the "J" count should have been returned to the mean, not walked out on the tail. This simply follows the law of large numbers.

 

[Thanz]

My count weighs nothing.Your assertion of "artificially" has yet to be demonstrated.

I obviously disagree. You have yet to demonstrate why the first reading is equivalent to the other three combined. Please, if you answer nothing else, answer this. Remember, we are trying to determine if he guesses J more often than he should as this could indicate that he is cold reading. We are comparing it to the general population. You need to tell me why his first reading should count as 4 people instead of one, when the others are only one.

You misunderstand my claim about the statistics. You utterly missed the fact that, when you increase the sample size and the proportion of "J"s is close to the mean, it does not become more significant statistically. You are watching the tail effect and missing the significance of the pdf. The pdf is shifting shape, reflecting the law of large numbers.

Thank you, I understand this now. What remains true, however, is that large numbers with the same proportions (J's being 21% of the total, for example) are different, statistically speaking, than small samples. It would seem then, that N does have an effect.

You keep missing the fact that JE's choice of when to call out multiples and how many to call out is, itself, a Poisson process. That means it is random. That means the "J"s should have gravitated toward the mean. You keep tripping over this. The multiple Js, Ks, Ms, Zs, Qs, are all random. The effect of a consistent application of the method is to return the frequencies to their means.

You keep forgetting what we are comparing it to. The control data is people - 1 person per name counted. There is no "What is your name?" "Jon" "Jon, or Jonathan?" "Just Jon" and then counting multiple J's in the census.

Further, you have no evidence that the multiple calling out is indeed random. It is possible, and consistent with the cold reading hypothesis, that he will only call out high probability names like John, and when he goes out on a limb for a less common letter does not go out on a leaf to suggest a specific name.

We are counting his guesses for actual people, and the underlying theory is that those guesses should be close to the distribution in the general population. Therefore, multiple guesses for the same person should not be counted the same as 3 guesses for different people. This is why your method is wrong.

 

[BillHoyt]

I obviously disagree. You have yet to demonstrate why the first reading is equivalent to the other three combined. Please, if you answer nothing else, answer this. Remember, we are trying to determine if he guesses J more often than he should as this could indicate that he is cold reading. We are comparing it to the general population. You need to tell me why his first reading should count as 4 people instead of one, when the others are only one.

I tire of answering the same questions over and over, sir. We are not counting people; we are counting guesses. "Counting people" makes an unwarranted assumption of a one-to-one relationship between guesses and people.

Thank you, I understand this now. What remains true, however, is that large numbers with the same proportions (J's being 21% of the total, for example) are different, statistically speaking, than small samples. It would seem then, that N does have an effect.

If you understand, then why do you persist in this specious reasoning?

You keep forgetting what we are comparing it to. The control data is people - 1 person per name counted. There is no "What is your name?" "Jon" "Jon, or Jonathan?" "Just Jon" and then counting multiple J's in the census.

We are not counting people with JE. We are counting name guesses. We are comparing these guesses against the census data.

Further, you have no evidence that the multiple calling out is indeed random. It is possible, and consistent with the cold reading hypothesis, that he will only call out high probability names like John, and when he goes out on a limb for a less common letter does not go out on a leaf to suggest a specific name.

We are testing the null hypothesis here, sir. The hypothesis is that his guesses are indistinguishable from the name distribution in the population at large. You are confounding this with a totally different hypothesis. One hypothesis at at time is how we work it in science.

We are counting his guesses for actual people, and the underlying theory is that those guesses should be close to the distribution in the general population. Therefore, multiple guesses for the same person should not be counted the same as 3 guesses for different people. This is why your method is wrong.

There is no underlying theory here. There is an hypothesis. It does not include the unwarranted assumptions you keep trying to impose.

 

[Thanz]

Again, you avoid the basic question. Why is reading 1 equivalent to readings 2, 3 and 4 combined?

Anyways, here we go again

We are not counting people with JE. We are counting name guesses. We are comparing these guesses against the census data.

Leaving aside the fact that I disagree your method accurately counts guesses, why aren't we counting people? Or, at least, guesses at people? Why compare "guesses" in the abstract against the census data? Why should there be any correlation if the guesses are divorced from any factor of actual people?

For that matter, I will once again raise one of my original objections to your counting methods which is that saying "Jean or Jane" is not the same thing as saying "J connection". It is certainly not the same thing as saying "J or J", counting the same letter twice in the first guess, as if it were two stand alone guesses.

We are testing the null hypothesis here, sir. The hypothesis is that his guesses are indistinguishable from the name distribution in the population at large. You are confounding this with a totally different hypothesis. One hypothesis at at time is how we work it in science.

They should only be indistinguishable if the guesses are related to actual people in some fashion. 3 specific name guesses with one sitter for one person is not the same as 3 guesses of "J connection" for 3 sitters on 3 occasions, yet your counting method equates them. And you have been completely unable to justify it to anyone at all, despite repeated requests.

There is no underlying theory here. There is an hypothesis. It does not include the unwarranted assumptions you keep trying to impose.

You must explain WHY you are using the control data that you are using, and why your counting method and hypothesis make sense on a logical level. Right now they do not.

I disagree quite strongly that there is no underlying theory. The basic idea is this: If JE is really bringing through spirits from the other side, over a sufficiently large sample we would expect the fornames of those spirits to basically match the name distribution at large. If, however, JE is cold reading, it makes logical sense that he would stick to the more common letters as it increases his chances of getting a hit. Therefore, we examine the guesses he makes to see if he really does favour the more common letters in his guesses.

There is no legitimate expectation that his guesses, however you want to define that word, would mirror the general name distribution in the population unless there is a link between the guesses and the population. Your method severs any link, and cannot be trusted.

I asked at the beginning, and I'll ask again now: What is the logical reason for counting reading 1 as equivalent to 2, 3 and 4 combined? Why can't you answer this simple question?

 

[Clancie]

Posted by Bill Hoyt

We can't assume JE is calling out names and initials to refer to a single person per single sitter.

So instead your method assumes that the initial--and subsequent name(s) that follow it--are all for different people. Does that really seem an unbiased premise to you? Does that really seem the most consistent and accurate counting method?

If this doesn't make theoretical sense to you, simply look at JE's own transcripts. They are replete with examples of a single guess referring to two different people.

Well, fyi, they are more often filled with a guess ("J, like John") where the two things clearly refer to a single person.

Here's my question: How does your method accurately reflect the count when JE says something like, "I'm getting grandpa. Was he a 'J-O" name like Joe or John" ? That is clearly one person, not the multiple people that you keep talking about.

How is your method (which inaccurately counts this as 3 guesses as if for three different people) preferable to Thanz's? How will your method yield a more accurate count than his?

I drop all mechanism assumptions and simply count his guesses.

Your method is inconsistent, Bill. It inaccurately makes it seem that JE guessing a string of "J" names for different people, even when you clearly know from the transcripts that he isn't.

I think your "J" count shows quite well how bias (conscious or unconscious, you know best)....can influence the selected method of analysis, in advance, with no paranormal skills involved at all....

 

[BillHoyt]

Originally posted by Clancie
So instead your method assumes that the initial--and subsequent name(s) that follow it--are all for different people. Does that really seem an unbiased premise to you? Does that really seem the most consistent and accurate counting method?

How many times must I answer this before that answer gets through. Neither the counting method nor the null hypothesis make any assumptions about people. JE is making name guesses, and the method counts those name guesses.

Well, fyi, they are more often filled with a guess ("J, like John") where the two things clearly refer to a single person.

So?

Here's my question: How does your method accurately reflect the count when JE says something like, "I'm getting grandpa. Was he a 'J-O" name like Joe or John" ? That is clearly one person, not the multiple people that you keep talking about.

Name guesses. That is what this method counts. Name guesses.

How is your method (which inaccurately counts this as 3 guesses as if for three different people) preferable to Thanz's? How will your method yield a more accurate count than his?

Name guesses. No assumptions.

Your method is inconsistent, Bill. It inaccurately makes it seem that JE guessing a string of "J" names for different people, even when you clearly know from the transcripts that he isn't.

All together now,

"N A M E G U E S S E S"

I think your "J" count shows quite well how bias (conscious or unconscious, you know best)....can influence the selected method of analysis, in advance, with no paranormal skills involved at all....

You're obviously not following the debate here. Thanz keeps claiming bias. Thanz keeps claiming distortion of the data. He has yet to demonstrate either those claims or an understanding of the nature of populations of random variables.

 

[BillHoyt]

Again, you avoid the basic question. Why is reading 1 equivalent to readings 2, 3 and 4 combined?

I have repeatedly answered it! All together now,

" N A M E G U E S S E S"

Leaving aside the fact that I disagree your method accurately counts guesses, why aren't we counting people? Or, at least, guesses at people? Why compare "guesses" in the abstract against the census data? Why should there be any correlation if the guesses are divorced from any factor of actual people?

These questions are idiotic, Thanz. Absolutely idiotic. First, I already answered them, over and over again; the method counts name guesses. Second, I never said they are divorced from people's names. I said we are out to test the null hypothesis about JE's guesses being indistinguishable from cold reading. So we look at his population of name guesses and ask, "is this sample indistinguishable from a random sampling of names from the population at large."

For that matter, I will once again raise one of my original objections to your counting methods which is that saying "Jean or Jane" is not the same thing as saying "J connection". It is certainly not the same thing as saying "J or J", counting the same letter twice in the first guess, as if it were two stand alone guesses.

They are the same. We have been through this before as well. This is tiresome. The frequencies of the first letters of the names guessed are compared with the frequencies of the first letters of names in the census data.

They should only be indistinguishable if the guesses are related to actual people in some fashion. 3 specific name guesses with one sitter for one person is not the same as 3 guesses of "J connection" for 3 sitters on 3 occasions, yet your counting method equates them. And you have been completely unable to justify it to anyone at all, despite repeated requests.

Nonsense.

You must explain WHY you are using the control data that you are using, and why your counting method and hypothesis make sense on a logical level. Right now they do not.

Nonsense.

I disagree quite strongly that there is no underlying theory. The basic idea is this: If JE is really bringing through spirits from the other side, over a sufficiently large sample we would expect the fornames of those spirits to basically match the name distribution at large. If, however, JE is cold reading, it makes logical sense that he would stick to the more common letters as it increases his chances of getting a hit. Therefore, we examine the guesses he makes to see if he really does favour the more common letters in his guesses.

This is simply more evidence you fail to grasp the definition of theory. And more evidence you fail to understand the epistemology of science.

 

[T'ai Chi]

JE is making name guesses and the method counts those name guesses.

All together now:

Sure, but the problem is is that if JE is referring to one person, the counting method you employ counts JE as referring to more than one person, and so of course it overcounts everything.

 

[CFLarsen]

The question is:

How many times does JE use "J" when he guesses names?

The number of people involved is irrelevant: JE throws out an initial, and the following name can mean anyone. There is also the possibility of a "J"-guess resulting in more than one "J"-name, even for the same sitter.

The number of people involved is simply a red herring.

 

[T'ai Chi]

The number of people involved is simply a red herring.

Not really, because when JE says something like 'I'm seeing an older gentleman' that tells me he is talking about ONE person before he states the names that he is sensing, or whatever, related to this person.

 

[CFLarsen]

Not really, because when JE says something like 'I'm seeing an older gentleman' that tells me he is talking about ONE person before he states the names that he is sensing, or whatever, related to this person.

Yes, really. As we know, this can fit anyone, even if the "J" is a woman.

Read some more transcripts.

Red herring.

 

[BillHoyt]

All together now:

Sure, but the problem is is that if JE is referring to one person, the counting method you employ counts JE as referring to more than one person, and so of course it overcounts everything. [/B]

Fascinating. And just what happens to the Poisson distribution with this alleged "overcounting"? Why didn't the "J" count move back to the mean? Thanz can't answer it. Clancie can't answer it. How about it, Tr'olldini?

 

[Clancie]

Bill,

This is my last try.

Hypothetical Scenario #1:

JE says, "I'm getting your grandfather. He has a 'J' name, something like 'John' or 'Joe'."

JE has given ONE person and said his name starts with a 'J'. He then gives 2 examples of the 'J' names he's thinking of for grandpa.

Thanz counts this as one guess for 'J', that JE is guessing that one specific person has a 'J' name. You know its one person with one "J' name, too, but you choose to count it as three separate guesses, therein inflating the count of how many times he guesses someone has a "J" name.

You score that exactly the same way as JE reading three sitters as follows:

Hypothetical Scenario #2:

JE to Sitter #1: "Did your father have a 'J' name?"

JE to Sitter #2: "I'm seeing an uncle named 'Joe'".

JE to Sitter #3: "Was your brother named 'John'?"

Your method scores the second scenario as 3 'J' guesses (correctly) just like you scored (incorrectly) as 3 guesses when he guessed one 'J' name for one person.

If bias isn't the reason for this counting method--if your intention in doing it this way isn't to get an inflated 'J' count, then what is the reason?

You still have not explained why your method is more accurate way to count JE's use of 'J' in his readings than Thanz's is. (And, no, "Name guesses! Name guesses!" isn't an answer).

 

[CFLarsen]

Thanz counts this as one guess for 'J', that JE is guessing that one specific person has a 'J' name. You know its one person with one "J' name, too, but you choose to count it as three separate guesses, therein inflating the count of how many times he guesses someone has a "J" name.

But there are three guesses: A "J"-name, "Joe" and "John".

Three.

And then we haven't even taken into account the hits for "Johnny", "Joseph", "Jose", "Joshua", "Jonathan"...all within the first 100 most popular male names (US Census 1990).

Not to speak of the 16 male names within the first 100 that begin with "J".

Throwing out a "J"-name for a male is a 1-6 chance of a hit.

(We could also include the "G"-names with a "Dj"-sound. "George", "Gerald". But we won't, even though we have seen JE get a phonetic initial....)

Anyway: Three. 1,2,3. "J", "John", "Joe".

 

[BillHoyt]

Bill,

This is my last try.



Thanz counts this as one guess for 'J', that JE is guessing that one specific person has a 'J' name. You know its one person with one "J' name, too, but you choose to count it as three separate guesses, therein inflating the count of how many times he guesses someone has a "J" name.

You score that exactly the same way as JE reading three sitters as follows:

Your method scores the second scenario as 3 'J' guesses (correctly) just like you scored (incorrectly) as 3 guesses when he guessed one 'J' name for one person.

If bias isn't the reason for this counting method--if your intention in doing it this way isn't to get an inflated 'J' count, then what is the reason?

You still have not explained why your method is more accurate way to count JE's use of 'J' in his readings than Thanz's is. (And, no, "Name guesses! Name guesses!" isn't an answer).

Drop the assumption about a feeble spirit trying desperately to mumble his name into JE's ears. Count the name guesses, and test the null hypothesis that they match the distribution from the census data. Name guesses. Name guesses.

Same question as given to Tr'olldini:

And just what happens to the Poisson distribution with this alleged "overcounting"? Why didn't the "J" count move back to the mean? Thanz can't answer it. Tr'olldini can't answer it. You haven't been able to so far. Care to try?

 

[Thanz]

I have repeatedly answered it! All together now,

" N A M E G U E S S E S"

That is not an answer. Your are equating name guesses with letter guesses, which is wrong in the first place, but if you accept that you can do that it makes absolutely no sense whatsoever to count the initial followed by examples as separate guesses. They are not separate guesses. If you are counting letters instead of names (which you are, no matter how many times you spell out "name guesses") it makes no sense to count multiple J's.

First, I already answered them, over and over again; the method counts name guesses.

No, it doesn't - it counts name guesses and then equates them with letter guesses, which is completely different. If you want to count all the names and then compare them to the distribution of those names, go ahead and count every name he says. But when you assume that saying John is the same as saying J, you can't count the examples that follow "J connection" as if they were separate guesses. They simply are not.

Second, I never said they are divorced from people's names. I said we are out to test the null hypothesis about JE's guesses being indistinguishable from cold reading. So we look at his population of name guesses and ask, "is this sample indistinguishable from a random sampling of names from the population at large."

And then you compare the NAMES to the LETTER distribution - you do not compare it to the name distribution. And you overcount on the letters in your method.

They are the same.

Complete and utter BS. What is the distribution of the name Jane in the general population? What is the distribution of the name Jean in the general population? What is the distribution of all names that start with the letter J? Are these three numbers all the same, or are they different?

The frequencies of the first letters of the names guessed are compared with the frequencies of the first letters of names in the census data.

Why is this a valid comparison? Why would his guesses - without any connection to actual people, as your method cuts that off - mirror the general population?

Lets say that we have two purported mediums. John always guesses specific names, and has a hit rate of 50%. James always guesses initial letters only, and has a hit rate of 50% as well. Are you saying that what these two people do is actually equivalent?

And saying "nonsense" in response to my arguments revealing why your count is wrong is not actually an argument. It just shows that you are completely unable to refute my points or come up with a simple, logical reason why reading one should be counted the same as the other three combined. you have had lots of time to come up with one. So far, nada.

 

[BillHoyt]

That is not an answer. Your are equating name guesses with letter guesses, which is wrong in the first place, but if you accept that you can do that it makes absolutely no sense whatsoever to count the initial followed by examples as separate guesses. They are not separate guesses. If you are counting letters instead of names (which you are, no matter how many times you spell out "name guesses") it makes no sense to count multiple J's.

JE's choice to ejaculate a multiple or not is a Poisson process. If such ejaculation choices were Poisson with the mean matching the census data, his observed values would have turned back toward the mean.

No, it doesn't - it counts name guesses and then equates them with letter guesses, which is completely different. If you want to count all the names and then compare them to the distribution of those names, go ahead and count every name he says. But when you assume that saying John is the same as saying J, you can't count the examples that follow "J connection" as if they were separate guesses. They simply are not.

And then you compare the NAMES to the LETTER distribution - you do not compare it to the name distribution. And you overcount on the letters in your method.


Complete and utter BS. What is the distribution of the name Jane in the general population? What is the distribution of the name Jean in the general population? What is the distribution of all names that start with the letter J? Are these three numbers all the same, or are they different?


Why is this a valid comparison? Why would his guesses - without any connection to actual people, as your method cuts that off - mirror the general population?

Lets say that we have two purported mediums. John always guesses specific names, and has a hit rate of 50%. James always guesses initial letters only, and has a hit rate of 50% as well. Are you saying that what these two people do is actually equivalent?

And saying "nonsense" in response to my arguments revealing why your count is wrong is not actually an argument. It just shows that you are completely unable to refute my points or come up with a simple, logical reason why reading one should be counted the same as the other three combined. you have had lots of time to come up with one. So far, nada.

Stop wasting my time with this specious nonsense about letters versus names. It is patent nonsense, and you know that full well.