JE on LKL Again

[SteveGrenard]

Thursday night, 10/2/03, at 6pm PST on CNN's Larry King Live, John Edward will be interviewed again and take your calls.

On the West Coast, Larry King Live starts at 6pm and then replays again at midnight. Check local listings for different time zones.

Some cable TV systems in the UK and Europe also carry the program much later in the night or in what would be early AM
Friday, 10/3/03.

Ref: (x-posted and edited from proving listserv)

[Hand Bent Spoon]

Just say 'no!' to watching John Edward.

[SteveGrenard]

Here's a chance for the LKL transcriptors to get a fresh batch of, er, transcripts. and observations.

[CFLarsen]

Originally posted by SteveGrenard
Here's a chance for the LKL transcriptors to get a fresh batch of, er, transcripts. and observations.

Too bad neither Clancie or neofight wants to address these. For many reasons....

[dimossi]

I suggest we all try to call in and maybe one of us skeptics can actually get through to challenge him just like the skeptic that got through to Sylvia Brown last time she was on LKL. (thread) (http://www.randi.org/vbulletin/showthread.php?s=&threadid=19740)

Maybe we could use a similar stategy that he used to get through..... lying. lol

The best bet is probably to get as many female skeptics to call. I would say that females have a much greater chance of getting past the screeners than a male.

Pass the word on and let's use this thread to come up with what the best question and/or statement would be if one of us actually managed to get through.

[Ersby]

Had a quick look at the transcript. It's all very clearly cold reading, but since people have said LKL isn't conducive to mediumship, then there's little we can take from that (other than asking why does JE do this?)
I found this interesting, right near the beginnning. JE said:

whenever you go see a psychic or a medium, the most important thing is not to divulge too much information because the more information that -- the more information you provide, then what happens is, it's less that the medium can actually use because now I know your dad's passed and now I know he passed from cancer and I know he passed two years ago.

and in my thread about my cold reading examples a month or so ago, I said:

What's interesting is why a medium wouldn't want the sitter to say too much. Not because it make the reading easier if the sitter talks a lot (and thus takes away from the amzingness of the reading), but because it makes it harder. I found that I was limited in my guesses refering to size of family and names because the sitter volunteered that information himself. Suddenly two of the better areas for guesswork were removed from me, and I struggled a bit.

Cold reading and mediumship, like two peas in a pod.

[Clancie]

Posted by ersby

What's interesting is why a medium wouldn't want the sitter to say too much. Not because it make the reading easier if the sitter talks a lot (and thus takes away from the amzingness of the reading), but because it makes it harder.
Actually, I agree with you ersby, and was thinking about this during the LKL readings. I'm not saying JE is a cold reader--I don't know--but it does seem true that not knowing who someone wants to contact may make it easier to come up with the gender and generation and cause of death of someone the sitter knows.

If that's all a reading consisted of, it wouldn't be very impressive, though, unless that information was quite unique.

But then, on the other hand, you have readings like Sylvia gives where she gets the sitter to tell a lot about the deceased and then just adds things like, "She's sorry she hurt you," or "He's watching over you and your children now". And that's even -less- evidential than the name/relationship/cause of death thing.

[Clancie]

By the way, John Edward is on "The Wayne Brady" show again today.

It's shown live and unedited. Last time JE gave readings so perhaps he will again today....

[BNiles]

I agree. If the sitter removes the easy guesses, either intentionally or accidentally, it leaves only a small window for any decent hits. The reader must work harder for specificity and hence credibility.

If a reader can make 3 small hits (the easy ones) and then follow with 1 big miss, he still appears successful. The audience will say, “Well, he only missed one.” Yes, but it was the only one worth anything.

[Lurker]

For kicks and giggles 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.

Lurker

[RC]

If JE is truly communicating with spirit, then it really shouldn't matter if he gets some information in advance. Spirit should be able to get him other information as evidence.

I would have no problem if JE knew I lost my dad two years ago to cancer (using his example), because if that's all he could tell me, then it wouldn't be impressive.

I couldn't get past the first 10 minutes of the show, especially when they used the infamous "cheese" reading before commercial break. In this reading, JE asked pretty much the whole crowd who had a salami in their purse, because he was seeing cuts of meat. A woman proceeded to pull out a baggie of cut up cheese and everyone roared. Then the reading began. I've been scratching my head over that one for months.

[c0rbin]

Aren't there people in the world suffering right now?

Why does this man touched by God spend so much time on TV and charging old ladys hundreds of dollars?

[BillHoyt]

Originally posted by Lurker
For kicks and giggles 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.

Lurker

And you don't think this is beginning to tell you something about what JE is doing? Assuming your results are correct, why aren't the non-Js rising to wash out the Js? There are 25 contenders in there whose counts should rise dramatically and wash out the Js. Why do the Js keep popping to the Poisson tail? I wonder...

Cheers,

[Lurker]

Originally posted by BillHoyt


And you don't think this is beginning to tell you something about what JE is doing? Why aren't the non-Js rising to wash out the Js? Why do the Js keep popping to the Poisson tail? I wonder...

Cheers,

Your note above is written as if I am a believer. Regardless of your accusations in the past, I have been a skeptic of JE and remain one.

Good observation, Bill, and one which I agree to. I consider it curious that the "J" letters are always overrepresented even if not statistically. It certainly could indicate cold reading tactics. Or it could mean other possibilities which have not really been explored. I mentioned other possiblities in other threads, ones which JE would need to discuss but since he does not debate critics we will never know if non-cold reading theories for this J distribution are plausible.

Lurker

[Clancie]

Posted by Bill Hoyt

And you don't think this is beginning to tell you something about what JE is doing? Assuming your results are correct, why aren't the non-Js rising to wash out the Js? There are 25 contenders in there whose counts should rise dramatically and wash out the Js. Why do the Js keep popping to the Poisson tail? I wonder...

Bill,

Did you somehow miss Lurker's point? :confused:
Posted by Lurker

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

[BillHoyt]

Originally posted by Lurker
Good observation, Bill, and one which I agree to. I consider it curious that the "J" letters are always overrepresented even if not statistically. It certainly could indicate cold reading tactics. Or it could mean other possibilities which have not really been explored. I mentioned other possiblities in other threads, ones which JE would need to discuss but since he does not debate critics we will never know if non-cold reading theories for this J distribution are plausible.

Lurker

The interesting thing here is that when he goes for "J" he seems to be all over it. When he goes for "T", not so. Or substitute anything you want. We'd expect that any counting method, universally and consistently applied, would be like a rising tide, lifting all boats. But it doesn't. The data suggest JE has a strong predilection for "J". If we had more JE/LKL readings, I'd guess we'd see other popular letters begin to pop out as well.

Cheers,

[BillHoyt]

Originally posted by Clancie
Did you somehow miss Lurker's point? :confused:
Apparently, you did. Lurker has noted something important and, I think, his curiousity is piqued. At least mine is.

Cheers,

[Thanz]

Originally posted by BillHoyt


The interesting thing here is that when he goes for "J" he seems to be all over it.
I'm not sure what you mean by this. Do you mean that when he guesses "J", he throws out a lot of specific name guesses but not so when he guesses other letters? If that is the case, then your counting method is even more skewed.
The data suggest JE has a strong predilection for "J". If we had more JE/LKL readings, I'd guess we'd see other popular letters begin to pop out as well.
Well, not as strong as you think. No other counting method has rejected the null hypothesis. We can maybe say that when he guesses J, he guesses more than one name in the hopes of getting a specific hit. That makes sense. But it is not appropriate to count each name as a separate guess to compare against the census.

Some other observations:

We have another guess of Ellen/Helen. Has he guessed this in every LKL show? It seems to pop up very frequently. I wonder if it is worth comparing this guess against the census info for the specific names.

We have a couple more dogs in this reading. One of them (the last reading) JE named the dog as "Max". Within the past couple of days, an American association of some kind (I forget which - I heard it on the radio, and it was not obvious) released a list of the most common dog names in the United States. Max was number one on the list.

[Ersby]

Can anyone find the original thread? I've forgotten what it was called.

Anyway, I used the Thanz method of counting, and JE does use "J" guesses far more than you'd expect. Don't know about this "rejecting the null hypothesis" mullarky. Could someone say out of, say 80 guesses, how many J guesses would be needed to reject this null hypothesis.

[renata]

Originally posted by Thanz

We have another guess of Ellen/Helen. Has he guessed this in every LKL show? It seems to pop up very frequently. I wonder if it is worth comparing this guess against the census info for the specific names.



He guessed (and missed) on 5/6 of his appearances. I have come to the conclusion Ellen/Helen haunts the LKL studio.

To my great frustration, I was about 80% done with the count on my thread (shameless self promotion, I know) when my system crashed, and I did not save it. And, I have friends over this weekend, so I can't get to it until later...so, you will just have to wait to see how the numbers will fall this time! :) Any guesses? :p

Interestingly, there were several wild ass guesses that could have become part of great JE lore, had they not entirely missed- now they will be entirely forgotten. There was also a very emotional reading, except once again all detail and meaning was provided by the caller.

[BillHoyt]

Originally posted by Thanz

I'm not sure what you mean by this. Do you mean that when he guesses "J", he throws out a lot of specific name guesses but not so when he guesses other letters? If that is the case, then your counting method is even more skewed. [B]
How on earth do you assert "skewed" if you acknowledge that it reflects an underlying reality? Beats me.

[b]Well, not as strong as you think. No other counting method has rejected the null hypothesis. We can maybe say that when he guesses J, he guesses more than one name in the hopes of getting a specific hit. That makes sense. But it is not appropriate to count each name as a separate guess to compare against the census.
We've been through this, over and over. Kerberos, et. al., also rejected the null hypothesis. The "J" count makes no sense if you assume the guesses are not cold reading. Please justify this bald assertion.

Cheers,

[Thanz]

Originally posted by BillHoyt

How on earth do you assert "skewed" if you acknowledge that it reflects an underlying reality? Beats me.
Again, in order to meaningfully compare it against the census, we need to count the number of people he is guessing for, not the number of times he says a name. The count is skewed if you know that he states specific "J" names more often within a single guess ("J, like John or Joe") rather than other letters if he just says the letter ("I am getting a "T" connection...) In terms of getting a hit, the J example is not substantively different than the T example. It may only affect the "quality" of the hit - not hit/miss.

So, if you know he tries the "quality enhacement" more often with the J, it is skewed to count each as a separate guess along with the other letters that he doesn't try this with.

The "J" count makes no sense if you assume the guesses are not cold reading. Please justify this bald assertion.

I'll try, but I don't know what bald assertion you are claiming I made. Please clarify.

[BillHoyt]

Originally posted by Thanz

Again, in order to meaningfully compare it against the census, we need to count the number of people he is guessing for, not the number of times he says a name.
Why? You foist an assumption on the name. Count the names called out. The second question raised by the data from my previous count (and, apparently, Lurker's count) is: why do the Js jump out?
The count is skewed if you know that he states specific "J" names more often within a single guess ("J, like John or Joe") rather than other letters if he just says the letter ("I am getting a "T" connection...) In terms of getting a hit, the J example is not substantively different than the T example. It may only affect the "quality" of the hit - not hit/miss.
What happened to the "T"s? Or "Q"s. Or any of the other 25 letters? Why did they not also mulitply to wash out the "J" results?
So, if you know he tries the "quality enhacement" more often with the J, it is skewed to count each as a separate guess along with the other letters that he doesn't try this with.
Why "J"? Why? Why? Why?

I'll try, but I don't know what bald assertion you are claiming I made. Please clarify.
Why is it not appropriate to drop the assumed mapping between a single reading and a person? Why is it not appropriate to simply count how many times JE calls out names? You cannot assume a particular process before you test. You must first eradicate as many unwarranted assumptions as possible.

[CFLarsen]

Clancie,

Since you now feel strong enough to discuss JE again, there are some questions waiting in the Underdown thread (http://www.randi.org/vbulletin/showthread.php?s=&threadid=26642&highlight=Underdown). You said we could remind you Friday, which is today. Unless you want to postpone it until we reach a "Friday" of your choosing, of course.

[Thanz]

Originally posted by BillHoyt

Why? You foist an assumption on the name. Count the names called out. The second question raised by the data from my previous count (and, apparently, Lurker's count) is: why do the Js jump out?
I don't know why the Js jump out. My guess is that JE feels rather comfortable that he can get some sort of J connection with a lot of people, and therefore he tweaks it a bit trying for a more "special" hit - the actual name rather than just the initial. But that is just a guess. And in any event, trying to figure out why he guesses more specific names with J than other letters is not what the analysis is focussed on. That analysis is just focussed on comparing guesses to the population. And if you are comparing guesses to the population, you need to make some sort of mapping of guesses to people.

What happened to the "T"s? Or "Q"s. Or any of the other 25 letters? Why did they not also mulitply to wash out the "J" results?

Why "J"? Why? Why? Why?
See above.

Why is it not appropriate to drop the assumed mapping between a single reading and a person? Why is it not appropriate to simply count how many times JE calls out names?
Again, because of the control data you are using. The control data is a collection of people. Not random names. I don't even think that saying "Who is John?" is an appropriate thing to lump in with "I get a J connection" as I see it as a different guess. But, you ridiculed me earlier for saying that. I have been willing to accept that "John" be equivalent to "J" for our purposes, if we assume that he would accept any J as validation. I am not willing to further bastardize the data by assuming that one guess of "a J, like John or Joe" is somehow equivalent to three separate guesses of "John?" "Jennifer?" and "J connection" because it simply is not equivalent.

You cannot assume a particular process before you test. You must first eradicate as many unwarranted assumptions as possible.
You are making unwarranted asusmptions by assuming that each name, even in "J, like John or Joe" is a separate guess for a separate person, when clearly that is one J guess. You claim to be eradicating unwarranted assumptions - instead, you are just making different assumptions with no rational basis.

You still have not answered why the one reading would be equivalent to the other three combined.

[alfaniner]

EDWARD: She also making me feel like I need to talk you about -- either the name of the dog is Max or I need to talk to you about a gray dog like a Schnauzer. What's Max?

From my Datacast news yesterday: THE LIGHTER SIDE -- New York Health Department released a list of the 10 most popular breeds and dog names. For the third year in a row, Max was the most popular dog name listed. (No info on the most popular breed).

Ooops, just saw that Thanz noted the same above.

[BillHoyt]

Originally posted by Thanz
My guess is that JE feels rather comfortable that he can get some sort of J connection with a lot of people, and therefore he tweaks it a bit trying for a more "special" hit - the actual name rather than just the initial.
Hello?
And in any event, trying to figure out why he guesses more specific names with J than other letters is not what the analysis is focussed on.
That is precisely what the analysis is focussed on! We seek to determine whether or not JE's performance is simple cold reading. The data show us that he picks on the most frequent first initial more often than chance would dictate.
That analysis is just focussed on comparing guesses to the population. And if you are comparing guesses to the population, you need to make some sort of mapping of guesses to people.
Not at all. We compare his names guesses to the population of names. The census data is simply name frequency. The JE data is simply name frequency. Basis statistical question: is the JE name population drawn from the same name population? Are the means equal?

You still have not answered why the one reading would be equivalent to the other three combined.
This has been answered so many times, Thanz, one can only speculate why you manage to miss it. I am not measuring "readings" here. Neither am I measuring "people" here. I am simply looking at JE's name/initial guesses and comparing them to the frequencies of names/initials in the census data. I am dropping all unwarranted assumptions about an unknown process and am attempting to see if this process is distinguishable from cold reading.

Cheers,

[T'ai Chi]

Originally posted by Lurker

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.

Lurker

Why not just take an average of the results of the two methods. :)

11/2 = 5.5 J guesses
15/2 = 7.5 non-J guesses.
Poisson says: (don't have calculator in front of me, someone else do it) :)

[BillHoyt]

Originally posted by T'ai Chi


Why not just take an average of the results of the two methods. :)

11/2 = 5.5 J guesses
15/2 = 7.5 non-J guesses.
Poisson says: (don't have calculator in front of me, someone else do it) :)

Whodini,

Beat your profs with a large stick this time. Line a bird cage with the degree, dude.

Somebody buy this kid a basic stat textbook. Holy bovine.

[T'ai Chi]

Originally posted by BillHoyt

That is precisely what the analysis is focussed on! We seek to determine whether or not JE's performance is simple cold reading. The data show us that he picks on the most frequent first initial more often than chance would dictate.


Of course, there are more high-frequency letters than just J. For example, if he uses J more but uses the other high-frequency letters less, can we really say that there is evidence of JE cold reading because of high J counts?

I think by only looking at the J's and not the other high frequency letters, we can only say that there is evidence of JE using more J's or not, and cannot really make any conclusions on if JE is cold reading or not.

[Thanz]

Originally posted by BillHoyt

Hello?
Hi!

That is precisely what the analysis is focussed on! We seek to determine whether or not JE's performance is simple cold reading. The data show us that he picks on the most frequent first initial more often than chance would dictate.

Not at all. We compare his names guesses to the population of names. The census data is simply name frequency. The JE data is simply name frequency. Basis statistical question: is the JE name population drawn from the same name population? Are the means equal?
You completely misuderstand. Let's go back to the basics.

What are we trying to determine here? We are trying to determine if JE's guesses are indistinguishable from cold reading. How do we do that? We look at the guesses he makes, and see how they compare to the general population.

Why would we do this? The theory is that if JE is a genuine medium, then the identities of the people he "brings through" should generally reflect the distribution in the population. If he is cold reading, however, he will likely skew his guesses toward the more common letters as it increases his chances of getting a hit. So, if he is cold reading, we expect him to make more J guesses than the general population would account for.

Are we together so far? Disagree with anything I have said here?

Moving on, then, rationally we need to look at his guesses. The theory is that he will cast a wider net for an increased chance of a hit. Saying "J" is more likely to get a hit than saying "X".

Now, you have also made the assumption that saying "John" is the equivalent of saying "J". I disagree. For cold reading purposes, saying "John" is a much narrower net than saying "J".

If however, we accept that assumption for counting purposes, certain things must follow in order to be consistent. If he says "J, like John or Joe", then under your "John=J" assumption, he is really saying "J, like J or J". For cold reading purposes, that is not any broader than simply saying "J". We should only count it once - as he is only going for one hit for one letter.

If, however, he says "Ellen or Helen", or "Bill, Patrick or Kevin" we need to count each (E or H) or (B, P or K).

This has been answered so many times, Thanz, one can only speculate why you manage to miss it. I am not measuring "readings" here. Neither am I measuring "people" here. I am simply looking at JE's name/initial guesses and comparing them to the frequencies of names/initials in the census data. I am dropping all unwarranted assumptions about an unknown process and am attempting to see if this process is distinguishable from cold reading.
No, it hasn't been answered, it has been artfully avoided using phrases like "dropping unwarranted assumptions".

Like it or not, you are measuring readings here - what else could you be measuring? What rational basis is there for equating the one reading with the other three, whether the process is cold reading, mediumship or something else? Your count of initials and names is meaningless unless connected to guesses for actual people.

[Lurker]

Imagine Je did readings over the Internet. He then said,

"I am getting a Johnor Jon, I am not sure how to spell it,"

One guess or two? Thanz and I say one. Bill's method, two.

It is unfortunate that the counting method has such a large impact on accepting or rejecting the hypothesis. From my point of view, I don't see how JE saying "I am getting an L name like Lynn or Lindsey or Lindy" would count as three guesses for our analysis unless one wants simplicity over accuracy.

Lurker

[Barkhorn1x]

...on LKL.

Saw about 45 minutes of the show. Some observations;

1. So many questions;

a. Asking the Irish woman if she had anybody die of a drinking ailment.

b. Asking what's broken in your kitchen - I looked around mine I counted 3 things.


2. Tried to call in - too many fools clogging the lines.

3. LK - that credulous bag of wind - joking at one point - "and they say you're not specific". This comment comes after some VERY vague guessing from JE. Time to go LK the worms have eaten more than half of your brain.

4. JE right before he - asks - an oh so obvious question about a grandma and her knitting needles says, "now don't write me you skeptics, because this is what I'm seeing". This comment probably means even he is embarassed at some level.

Overall = Weak Mr. E, very weak.

Tag line running continually on bottom of the screen = "News You Can Trust" - anyone else see the irony here??

Barkhorn.

[Walter Wayne]

Originally posted by Lurker
It is unfortunate that the counting method has such a large impact on accepting or rejecting the hypothesis. Well, if we combine the different readings counting by 1 (as I also prefer) the stats are getting closer to the alpha value. I think we can combine here, as it is reasonable to assume the method is the pretty close to the same accross all the LKL's (one could argue because we are not sure that of similar methodology we can't combine though).

However, I still have a problem with using census figures for J probability. I am not entirely convinced that JEs audience is a cross-section of america. But as long as one admits it is "back-of-the-envolope" statistics, we can say we are ballpark on frequency of J.

Walt

[T'ai Chi]

Originally posted by BillHoyt

Beat your profs with a large stick this time. Line a bird cage with the degree, dude.


They gave me two degrees when I beat them. Shall I do it again and get a third one? Any suggestions are greatly appreciated.

Bird cage... it would be more rational to use newspapers, so I will have to reject your idea of using a degree for that purpose.


Somebody buy this kid a basic stat textbook. Holy bovine.

This kid has many, as well as advanced ones. If you'd like some recommendations of books that are best for certain areas of statistics, please let me know.

Please note the smiley I used in my post that you responded to. I wasn't considering seriously taking an average and doing the calculations based on that. It was a joke to hint that the decision made is seriously affected by the counting method.

Especially when people are counting Jeff, John, Johansan, Jered, Jimmy, Johosaphat, Jackoff, Jamboree, Jammy, as 9 instead of 1. It seems that there need to be some objective counting methods that are agreed upon by all parties who wish to analyze the statistics in a meaningful manner. Agree?

[BillHoyt]

Originally posted by Lurker
Imagine Je did readings over the Internet. He then said,

"I am getting a Johnor Jon, I am not sure how to spell it,"

One guess or two? Thanz and I say one. Bill's method, two.

It is unfortunate that the counting method has such a large impact on accepting or rejecting the hypothesis. From my point of view, I don't see how JE saying "I am getting an L name like Lynn or Lindsey or Lindy" would count as three guesses for our analysis unless one wants simplicity over accuracy.

Lurker

If you wish to argue a point, then get the facts straight. Read the method again.

[BillHoyt]

Originally posted by T'ai Chi
Especially when people are counting Jeff, John, Johansan, Jered, Jimmy, Johosaphat, Jackoff, Jamboree, Jammy, as 9 instead of 1. It seems that there need to be some objective counting methods that are agreed upon by all parties who wish to analyze the statistics in a meaningful manner. Agree?
[/B]
We count Mary and Manfred and Meathead the same way. Also, Peter, Poindexter and Pinhead. As well as Ned, Natalie and Nitwit. Twenty-five other initials all counted by the same rules and J sails way out on the tail.

Why?

[BillHoyt]

Originally posted by Thanz
My guess is that JE feels rather comfortable that he can get some sort of J connection with a lot of people, and therefore he tweaks it a bit trying for a more "special" hit - the actual name rather than just the initial.

[Clancie]

Posted by T'ai Chi

It seems that there need to be some objective counting methods that are agreed upon by all parties who wish to analyze the statistics in a meaningful manner. Agree?
T'ai Chi,

I think everyone does agree with the "one guess at a time means one initial, even if JE gives many examples of names starting with that initial."

Well, everyone agrees except Bill.

And he will never agree. In fact, I have yet to read any discussion with him on any topic in which he has modified his views or actually seemed responsive to others' points of view. I can't think of one.

But at least, at this point, he is 100% predictable.....

[CFLarsen]

Clancie,

If JE was not using psychic mediumship, but "merely" his psychic abilities on LKL (because he didn't get any "spirit communication" whatsoever), what is the difference between a psychic and a cold reader?

[T'ai Chi]

Originally posted by BillHoyt

We count Mary and Manfred and Meathead the same way. Also, Peter, Poindexter and Pinhead. As well as Ned, Natalie and Nitwit. Twenty-five other initials all counted by the same rules and J sails way out on the tail.

Why?

So JE uses some high-frequency letters more, and some high-frequency letters less than expected.

Why?

[BillHoyt]

Originally posted by Troll Chi


So JE uses some high-frequency letters more, and some high-frequency letters less than expected.

Why?

Not "more than expected", Troll Chi. The "J"s are used significantly more than expected.

[Ersby]

Originally posted by T'ai Chi


So JE uses some high-frequency letters more, and some high-frequency letters less than expected.

Why?

J and C are used more than you'd expect.

M and R are used as expected.

D names are used less because JE seems to keep them for younger energies.

B names, which aren't part of the six most common initials, are used more because they cover the Bobby/Billy/Betty names.

He uses L more than expected, but I don't know why.

[thaiboxerken]

I think everyone does agree with the "one guess at a time means one initial, even if JE gives many examples of names starting with that initial."

An initial isn't anything more than a fishing expedition. This shouldn't even count as a guess.

[Kevin_Lowe]

Originally posted by thaiboxerken
I think everyone does agree with the "one guess at a time means one initial, even if JE gives many examples of names starting with that initial."

An initial isn't anything more than a fishing expedition. This shouldn't even count as a guess.

Sure.

But if the percentage of times JE goes fishing for "J" exceeds the percentage of the population with "J" names, then that alone is evidence he is cold reading. Well, evidence that he is cold reading a lot of the time.

Creduloids confronted with this will retreat to the good old "Yes they cheat a lot of the time, but sometimes they get real psychic hits!" position, and the dance goes on.

[Lurker]

Originally posted by Kevin_Lowe


Sure.

But if the percentage of times JE goes fishing for "J" exceeds the percentage of the population with "J" names, then that alone is evidence he is cold reading. Well, evidence that he is cold reading a lot of the time.

Creduloids confronted with this will retreat to the good old "Yes they cheat a lot of the time, but sometimes they get real psychic hits!" position, and the dance goes on.

As a skeptic, I think you are wrong. Both you and Bill have said this (the first paragraph) but the "J" analysis hypothesis makes no mention of cold reading. Yes, we know JE uses J more than chance but our inference based on that statistical anomaly is just that, an inference which may or may not be wrong. It is evidence, much like anecdotal stories are evidence. The value of each is subjective.

Lurker

[BillHoyt]

Originally posted by Lurker


As a skeptic, I think you are wrong. Both you and Bill have said this (the first paragraph) but the "J" analysis hypothesis makes no mention of cold reading. Yes, we know JE uses J more than chance but our inference based on that statistical anomaly is just that, an inference which may or may not be wrong. It is evidence, much like anecdotal stories are evidence. The value of each is subjective.

Lurker

Guano. I tire of this blur campaign. "I am a skeptic" / "Anecdotal stories are evidence" / "[evidence] is subjective"

This isn't halloween, yet, "skeptic". Next time you don the mask, better try to at least imitate what a skeptic would say.

Guano.

[Lurker]

Originally posted by BillHoyt


Guano. I tire of this blur campaign. "I am a skeptic" / "Anecdotal stories are evidence" / "[evidence] is subjective"

This isn't halloween, yet, "skeptic". Next time you don the mask, better try to at least imitate what a skeptic would say.

Guano.

Bill, are you saying evidence is black or white? That it is either 100% right or 100% wrong? You need to understand that for some people, anecdotal evidence is sufficient. for some, it is not. For me, it is not but I understand that other people have different ideas on what constitutes evidence. That is what I was getting across. I might be more inclined to lend credence to the anecdotal evidence if I know the person and respect them. In reality, everyone does this including you, Bill.

Think about it, Bill. If something happened to you, you might place more significance on the veracity of that experience than I would.

Beyond anecdotes, we have eyewitness testimony in courts yet this is fallible. Why is it still called evidence?

Even for a scientific study the result can be evidence for a conclusion yet be disproven later by a new study showing the previous lacked in some way.

Bill, evidence IS subjective. We each pick and choose how we value each piece of evidence. Personally, I think you just get snitty when people dare question you. Recall, your "J" analysis hypothesis makes no mention of cold reading. That is merely a conclusion one can derive from the results but it is an extrapolation of the evidence. It does not mean it is conclusively proven. Each will weigh the evidence by their own subjective sense.

Lurker

[Lurker]

Actually, let me add that the analysis IS evidence that JE is cold reading. It is just not conclusive evidence. It merely points in that direction.

My bad,

Lurker

[BillHoyt]

Originally posted by Lurker
Bill, are you saying evidence is black or white? That it is either 100% right or 100% wrong? You need to understand that for some people, anecdotal evidence is sufficient. for some, it is not.
Where did I say anything so bifurcated?

You need to understand that anecdotal evidence is not sufficient for rational people. It is insufficient for scientists and insufficient for skeptics.
For me, it is not but I understand that other people have different ideas on what constitutes evidence. That is what I was getting across. I might be more inclined to lend credence to the anecdotal evidence if I know the person and respect them. In reality, everyone does this including you, Bill.
So what if other people have other ideas? That doesn't make those ideas right, sir. You are now stepping into argumentum ad populem and argument from authority, fallacious territory.

Beyond anecdotes, we have eyewitness testimony in courts yet this is fallible. Why is it still called evidence?
Fallacy of equivocation, sir.

Even for a scientific study the result can be evidence for a conclusion yet be disproven later by a new study showing the previous lacked in some way.

Bill, evidence IS subjective. We each pick and choose how we value each piece of evidence. Personally, I think you just get snitty when people dare question you. Recall, your "J" analysis hypothesis makes no mention of cold reading. That is merely a conclusion one can derive from the results but it is an extrapolation of the evidence. It does not mean it is conclusively proven. Each will weigh the evidence by their own subjective sense.

Lurker
Evidence is not subjective. It is objective, sir. That people often use subjective criteria does not argue in favor of continuing such an error. Teaching people not to make such errors is part of the JREF mission.

Cheers,

[Clancie]

There are several things I'm unclear of about this "J" analysis. I hope someone can help. :confused:

1. Did everyone's analysis show JE using 'J' significantly more than expected? Or was it only Bill's method that got that result?

2. If you all agree that the transcript information does show 'J' used more than expected, was there also ever agreement on whether or not the sample size was large enough to be sure the conclusion for 'J' is valid?

3. If "'J' used more than expected is "evidence" of cold reading" (as Bill says)....would 'J' used less than expected have been considered evidence JE doesn't cold read? If not, why not?

4. Why does it only matter if JE guesses lots of 'J's' and doesn't matter at all if he's correct about these guesses? (For example, just supposing ADC is real, maybe it would be easier for him to get 'j' names...maybe because his own name starts with 'J'...or for some other mediumistic reason).

I know it was decided in the beginning of the count that it didn't matter if his guess turned out to be right or turned out to be wrong. But would it still be "evidence" of cold reading if he says 'J' much more than expected--and each and every 'J' name he gets is a hit?

5. If we only get a result for one letter, how do we know it's not just a fluke? What if all the other high frequency letters are used less than expected?

How come 'J' results can be used to draw such a big conclusion without knowing the pattern of how all the other 25 letters show up in his readings?

6. Why isn't his pattern of using all letters equally signficant as the results for 'J'? :confused:

[Thanz]

Originally posted by Clancie
There are several things I'm unclear of about this "J" analysis. I hope someone can help. :confused:
I'll see if I can do anything to clarify things for you.

1. Did everyone's analysis show JE using 'J' significantly more than expected? Or was it only Bill's method that got that result?
If you mean "significant" in a statistical sense, using the analytical method advocated by Bill (Poisson), the only COUNTING method that yielded a significant result was Bill's. Other counts did not yield a statistically significant result (using Bill's stat tools and parameters)

2. If you all agree that the transcript information does show 'J' used more than expected, was there also ever agreement on whether or not the sample size was large enough to be sure the conclusion for 'J' is valid?
Well, this is really about whether Bill's stat tool is appropriate. Apparently, there is enough info to run Bill's stat tool. There is not enough info to run some other stat tools that were proposed.

Personally, I think that there is a 'common sense' element to this as well, and that tells me that no there is not enough of a sample size.

3. If "'J' used more than expected is "evidence" of cold reading" (as Bill says)....would 'J' used less than expected have been considered evidence JE doesn't cold read? If not, why not?
You would need to explain why less J's is evidence of non-cold reading. The idea of more Js is that JE knows this and uses it to get more hits. Less Js could just mean that he is unaware of the distribution or uses other cold reading techniques. He does seem overly fond of Helen/Ellen... :)

4. Why does it only matter if JE guesses lots of 'J's' and doesn't matter at all if he's correct about these guesses? (For example, just supposing ADC is real, maybe it would be easier for him to get 'j' names...maybe because his own name starts with 'J'...or for some other mediumistic reason).

I know it was decided in the beginning of the count that it didn't matter if his guess turned out to be right or turned out to be wrong. But would it still be "evidence" of cold reading if he says 'J' much more than expected--and each and every 'J' name he gets is a hit?
Personally, I think that if each J was a hit it would weaken the case for cold reading. One of the assumptions made was that JE's audience would basically mirror the census. We don't know this and it could be that J names are simply overrepresented (as compared to the census) in JE's audience, and if each were a hit we can't accuse him (on that basis alone) of cold reading as mediumship would have produced the same results.

5. If we only get a result for one letter, how do we know it's not just a fluke? What if all the other high frequency letters are used less than expected?

How come 'J' results can be used to draw such a big conclusion without knowing the pattern of how all the other 25 letters show up in his readings?
The J results would only be suggestive. We don't have enough data to run multiple letter tests, it would seem. I think that it is obvious that a multi-letter analysis would give us much more interesting results than simply looking at J.

6. Why isn't his pattern of using all letters equally signficant as the results for 'J'? :confused:
With the data we have, we can only really look at one letter (from a statistical point of view). Doing individual tests of each letter is quite different from doing one test of multiple letters, and we don't have enough data to do the multi-letter test.

That is how I see it. I hope that I have been helpful.

[BillHoyt]

Originally posted by Thanz
If you mean "significant" in a statistical sense, using the analytical method advocated by Bill (Poisson), the only COUNTING method that yielded a significant result was Bill's. Other counts did not yield a statistically significant result (using Bill's stat tools and parameters)
I thought Lurker claimed similar results on new data. Is that not so?

You would need to explain why less J's is evidence of non-cold reading.
It isn't, sir. :eek:

Cheers,

[Thanz]

Originally posted by BillHoyt

I thought Lurker claimed similar results on new data. Is that not so?
Lurker did two counts of the new data, one using your method and one using mine. Your counting method said we reject the null hypothesis, mine did not.

If nothing else, the results are consistent with the old data, which found the same thing.
It isn't, sir. :eek:
I apologize if I was unclear. What I meant was that someone would need to present some sort of theory or hypothesis as to why less Js would mean that JE was not cold reading, and then test that. It is not enough to fail to conclude that he IS cold reading to say that he is not.

[BillHoyt]

Originally posted by Thanz
If nothing else, the results are consistent with the old data, which found the same thing.
So we have two different sets of data, two different people using the same technique, and both show the same strange result. The "J"s sail out to the Poisson tail.

Why?

[BillHoyt]

Originally posted by Thanz
I apologize if I was unclear. What I meant was that someone would need to present some sort of theory or hypothesis as to why less Js would mean that JE was not cold reading, and then test that. It is not enough to fail to conclude that he IS cold reading to say that he is not.

You weren't unclear, Thanz. You were wrong. That is not how hypothesis testing works. Less "J"s do not mean "not cold reading". Please do not argue this point with me. We have been around and around on the basic scientific method. Please do some homework, look it up and then correct your posts, sir.

[Lurker]

Originally posted by BillHoyt


You need to understand that anecdotal evidence is not sufficient for rational people. It is insufficient for scientists and insufficient for skeptics.


Well, it really depends on what we are discussing. I will advise that we ALL depend on anecdotal evidence. You do read newspapers, right? We read textbooks in college, right? Every single experiment is anecdotal, is it not? It really depends to what extreme you are willing to define anecdotal. Further experiments that come to the same conclusion proves it? Hmm, seems like an argument by popularity. Now don't get your panties in an uproar. I stand fully behind experiments and replication.

Yes, I find anecdotes to be insufficient evidence as it relates to JE. But I do find anecdotal evidence to convince me in other aspects of my life. I do hope that you don't demand rigourous experiments from your teacher every time he makes a claim.



So what if other people have other ideas? That doesn't make those ideas right, sir. You are now stepping into argumentum ad populem and argument from authority, fallacious territory.

Fallacy of equivocation, sir.


Um, I was not making an argument at all except that anecdotes are evidence. Poor evidence in my opinion but evidence nonetheless. Cold fusion had anecdotal evidence in the experiment that supported it. Later work established that that anecdotal evidence was erroneous.


Evidence is not subjective. It is objective, sir. That people often use subjective criteria does not argue in favor of continuing such an error. Teaching people not to make such errors is part of the JREF mission.



Ideally it would be objective, yes. But the interpretation of evidence is subjective. That is what I am getting at. The J analysis is a good example. We each are weighing in our mind what value we are placing on that evidence, correct? Subjectivity.

If Randi said in his next column that he had found someone with ESP what might you think? Sure, you might still think it was false but you would probably put more weight on Randi saying this than Gregnard saying it, right? By doing so you are applying a quite subjective value on anecdotal evidence.

So, in a way, you are correct in saying evidence is not subjective. But the subsequent interpretation of that evidence is quite subjective.

Lurker]

[Clancie]

Thanz,

Thank you. That was very much help.

Bill,

What -would- less J's than expected have meant to you, if you -had- gotten that result?

And, could you get -any- counting result that would support the idea JE was not cold reading? Hypothetically, what result would that be?

[Thanz]

Originally posted by BillHoyt

So we have two different sets of data, two different people using the same technique, and both show the same strange result. The "J"s sail out to the Poisson tail.

Why?
Mr. Hoyt -

We have been round and round the mulberry bush on this before, and I have no intention of doing it again.

Suffice it to say that I do not think that your counting method is appropriate for the analysis, and therefore any results that one gets from your counting method cannot be relied upon.

[Thanz]

Originally posted by BillHoyt


You weren't unclear, Thanz. You were wrong. That is not how hypothesis testing works. Less "J"s do not mean "not cold reading". Please do not argue this point with me. We have been around and around on the basic scientific method. Please do some homework, look it up and then correct your posts, sir.
Mr. Hoyt -

I am making a resolution to myself to try and not get sucked into these arguments with you. I will clarify what I said one more time.

I did not say that less Js meant that JE was not cold reading. All that I meant was that if somebody wanted to try and assert that less Js meant JE was not cold reading, the approach would have to be an independant test of that.

That approach would involve the formulation of a logical, testable hypothesis that would somehow explain or theorize that less Js meant JE was not cold reading. Then, you need to come up with a test of that hypothesis. I did post an opinion that less Js did mean that JE is not cold reading. I just pointed out that such a conclusion would require its own test.

Feel free to go blue in the face pointing out whatever errors you think I have made, complete with rolleyes and the like. I trust that other posters understand what I mean.

[BillHoyt]

Originally posted by Clancie
What -would- less J's than expected have meant to you, if you -had- gotten that result?

The result obtained was not simply more than expected. It was significantly more than expected. Getting more than expected "J"s would have meant nothing. Getting less than expected "J"s would have meant nothing. Getting significantly less than expected "J"s is not possible with a one-tailed test.

[BillHoyt]

Originally posted by Thanz

Mr. Hoyt -

We have been round and round the mulberry bush on this before, and I have no intention of doing it again.

Suffice it to say that I do not think that your counting method is appropriate for the analysis, and therefore any results that one gets from your counting method cannot be relied upon.

You can't simply declare the method null and void. You found fault with it. Now Lurker reports that the same method with different data yields a similar result. You must address this by identifying a systematic error with the method or by refuting Lurker's results.

So, do you refute Lurker's results?

[BillHoyt]

Originally posted by Thanz

Mr. Hoyt -

I am making a resolution to myself to try and not get sucked into these arguments with you. I will clarify what I said one more time.

I did not say that less Js meant that JE was not cold reading. All that I meant was that if somebody wanted to try and assert that less Js meant JE was not cold reading, the approach would have to be an independant test of that.

That approach would involve the formulation of a logical, testable hypothesis that would somehow explain or theorize that less Js meant JE was not cold reading. Then, you need to come up with a test of that hypothesis. I did post an opinion that less Js did mean that JE is not cold reading. I just pointed out that such a conclusion would require its own test.

Feel free to go blue in the face pointing out whatever errors you think I have made, complete with rolleyes and the like. I trust that other posters understand what I mean.

It is your mulberry bush, sir, not mine. I did not say much beyond "the test refutes the null hypothesis." Now one does not need to do the test you propose. The question is not about the "J"s but about the cold reading.

[Thanz]

Originally posted by BillHoyt

You can't simply declare the method null and void. You found fault with it. Now Lurker reports that the same method with different data yields a similar result. You must address this by identifying a systematic error with the method or by refuting Lurker's results.

So, do you refute Lurker's results?
I will respond one more time, against my better judgment.

I have already pointed out, several times, what I see as a systemic error with your counting method. If you still do not understand what that systemic error is, I refer you to my previous posts on the subject. If you read those again and still don't understand the systemic error, I do not think that I can help you. I think that other posters understand my objections to your method. Perhaps you can ask someone else to help you.

[BillHoyt]

Originally posted by Thanz

I will respond one more time, against my better judgment.

I have already pointed out, several times, what I see as a systemic error with your counting method. If you still do not understand what that systemic error is, I refer you to my previous posts on the subject. If you read those again and still don't understand the systemic error, I do not think that I can help you. I think that other posters understand my objections to your method. Perhaps you can ask someone else to help you.

Perhaps I am forgetting, but I do not recall anything you have said in your objections that would either predict or explain why the "J"s become significant by my method. Where did you explain why the "J"s become significant? Please explain exactly what makes their numbers rise relative to all other initials? If you think that rise was a fluke, then please explain whether you think Lurker's recent test is correct. If you do not refute Lurker, then explain why he reports his test yielded a similar rise in "J"s.

[Thanz]

Originally posted by BillHoyt


Perhaps I am forgetting, but I do not recall anything you have said in your objections that would either predict or explain why the "J"s become significant by my method. Where did you explain why the "J"s become significant? Please explain exactly what makes their numbers rise relative to all other initials? If you think that rise was a fluke, then please explain whether you think Lurker's recent test is correct. If you do not refute Lurker, then explain why he reports his test yielded a similar rise in "J"s.
I do not need to explain your results. I need only explain what the error is in your method which shows that it is not rational for the test we are conducting, which I have done. Your results, based on your non-rational method, are irrelevant.

[BillHoyt]

Originally posted by Thanz

I do not need to explain your results. I need only explain what the error is in your method which shows that it is not rational for the test we are conducting, which I have done. Your results, based on your non-rational method, are irrelevant.
Let's see. I propose a counting method. I give results that show significantly more "J"s than expected. You object, giving no explanation for why "J"s become significant. Now Lurker reports a similar phenomenon when using my method with a new set of data. And your response is that my "results are irrelevant?" I assume you apply this comment equally to Lurker's reported results? Yes or no?

Is this the way you deal with evidence? Do you know why I chose the telescope as my avatar? Do you know the historical significance of it?

Cheers,

[Thanz]

Originally posted by BillHoyt

Let's see. I propose a counting method. I give results that show significantly more "J"s than expected. You object, giving no explanation for why "J"s become significant. Now Lurker reports a similar phenomenon when using my method with a new set of data. And your response is that my "results are irrelevant?" I assume you apply this comment equally to Lurker's reported results? Yes or no?
I feel the need to remind you that I agree that JE is cold reading.

Having said that, if your data collection method is flawed, you cannot rely on the results you obtain, no matter what those results say. Your counting method results in incorrect numbers for both the number of J guesses and the total number of guesses. Given the method of analysis (poisson) that you propose, an extra J guess will have far more impact than an extra D guess.

If I recall correctly, your count had 18 J's in 85 total guesses. If we have one less J and everything else remains equal, we do not reject the null hypothesis.

[Paul C. Anagnostopoulos]

Why don't we suggest to JE that during 2004 he throw out the initial of the last name rather than the first. If he's talking to dead people, he can ask them for their last names just as easily, no?

Heck, if dead people forget their last names, how about the final letter of the first name? Also, he should take a trip to India or Japan:

I'm getting that glyph that's 26 rows down and seven columns in on the chart . . .

~~ Paul

[CFLarsen]

Originally posted by Paul C. Anagnostopoulos
Heck, if dead people forget their last names, how about the final letter of the first name? Also, he should take a trip to India or Japan:

I'm getting that glyph that's 26 rows down and seven columns in on the chart . . .

JE has actually once gotten something in Chinese. Clancie, claiming to know Chinese (although apparently being unaware that Mandarin and Cantonese are two very different spoken languages), had no problems with JE getting it clairaudienty, and thought not to ask whether it was Mandarin or Cantonese. Or any other of the many dialects we find in China.

So, JE does get other languages. I would love to see if JE ever got a dead Danish woman called "Yvonne". The "Y" sound is impossible for Americans to pronounce! :D

[Lurker]

Originally posted by CFLarsen


So, JE does get other languages. I would love to see if JE ever got a dead Danish woman called "Yvonne". The "Y" sound is impossible for Americans to pronounce! :D

Sorry Claus, we need evidence here. Prove it is impossible for Americans to pronounce this. :)

Lurker

[CFLarsen]

Originally posted by Lurker
Sorry Claus, we need evidence here. Prove it is impossible for Americans to pronounce this. :)

Lurker

I'd have to say it. You'll have to wait until TAM2 :)

[BillHoyt]

Originally posted by Thanz

I feel the need to remind you that I agree that JE is cold reading.

Having said that, if your data collection method is flawed, you cannot rely on the results you obtain, no matter what those results say. Your counting method results in incorrect numbers for both the number of J guesses and the total number of guesses. Given the method of analysis (poisson) that you propose, an extra J guess will have far more impact than an extra D guess.

If I recall correctly, your count had 18 J's in 85 total guesses. If we have one less J and everything else remains equal, we do not reject the null hypothesis.
Thanz,

Saying it is flawed simply repeats a bare assertion. What you have to address is why the "J"s become significant. Your claim that a J has more impact than a D is specious. There are 26 letters from which to choose. Statistically, one would expect the counts to be inflated proportionately. If this had happened the J count would not have become significant.

Now Lurker claims to have achieved a similar result with different data. If this is so, your post here is even weaker. Lurker's results would give us two different sets of data in which this J count is significant.

You cannot simply wave it away. Why do the J's inflate so? Explain it.

[ImpyTimpy]

Originally posted by BillHoyt

--snipped--


Evidence is not subjective. It is objective, sir. That people often use subjective criteria does not argue in favor of continuing such an error. Teaching people not to make such errors is part of the JREF mission.

Cheers,

Actually, that's a very interesting thing to say Billy... Objective evidence? Interesting idea. That would suggest that evidence is self explaining, there is no need for analysis or discussion of evidence because the result should be objective. Unfortunetly that's not how it works. When someone is presented with evidence, that evidence is analyzed subjectively. Conclusions are drawn based upon the evidence. Evidence itself doesn't lead to immediate conclusions (as you seem to imply with your "objective evidence" statement).

[T'ai Chi]

Originally posted by BillHoyt

The "J"s are used significantly more than expected.

You didn't answer the question. At all.

JE uses some high-frequency letters more, and some high-frequency letters less than expected.

What do you conclude from that?

[T'ai Chi]

Originally posted by BillHoyt

You can't simply declare the method null and void.

You must address this by identifying a systematic error with the method ..


Yeah, good point. I can use that in another thread..

[Ersby]

Originally posted by T'ai Chi


You didn't answer the question. At all.

JE uses some high-frequency letters more, and some high-frequency letters less than expected.

What do you conclude from that?

Just out of interest, what's wrong with the conclusions I drew?

[BillHoyt]

Originally posted by Tr'oll Chi


You didn't answer the question. At all.

JE uses some high-frequency letters more, and some high-frequency letters less than expected.

What do you conclude from that?

JE uses "J" significantly more than expected. There is nothing that can be concluded from lower-than-expected counts. Period. That is so utterly basic to inferential statistics.

Ya buy em books and ya buy em books and all they do is chew on the covers.

[Thanz]

Originally posted by BillHoyt

Saying it is flawed simply repeats a bare assertion.
No, I have explained why it is flawed. It is flawed no matter what results it gives. It is irrational. It is as flawed as the gap in Dr. Schwartz's screen.

What you have to address is why the "J"s become significant.
No I don't.

Your claim that a J has more impact than a D is specious. There are 26 letters from which to choose.
And I could have picked any or all of them. It doesn't matter. Your counting method is irrational, so the results (whatever they may be) cannot be trusted.
Now Lurker claims to have achieved a similar result with different data. If this is so, your post here is even weaker. Lurker's results would give us two different sets of data in which this J count is significant.
Ever hear of "confirmation bias"? You have to justify your counting method as rational independent of the results that it gets. This you have failed to do.

You cannot simply wave it away. Why do the J's inflate so? Explain it.
In fact I can just wave it away, as the method is wholly irrational. And I am done discussing this with you. I trust that others understand my position, and I have already wasted too much time responding to you. Good day, sir.

[BillHoyt]

Originally posted by Thanz
No, I have explained why it is flawed. It is flawed no matter what results it gives. It is irrational. It is as flawed as the gap in Dr. Schwartz's screen.
With the gap in Schwartz' screen, we can predict an experimental outcome. That outcome is an artificial improvement in JE's hits due to information leakage.

You have made no such prediction for my experimental method. You simply say it is flawed. Yet Lurker claims to have produced similar results with different data. That requires an explanation. Why did my test show "J"s significantly over-used? Why does Lurker claim his test shows "J"s significantly over-used?

Why should the "J"s be so dramatically over-represented? Why did the other 25 letters not inflate equally? How can that be?

Do you claim my results did not happen? Do you claim Lurker's results did not happen? How can you justify declaring them "irrelevant"?

Here, ladies and gentlemen, it is. Skeptics are called "arrogant". Skeptics are called "closed-minded." And here is what happens when real evidence is brought to bear. The woos declare it 'irrelevant". This is not how science works. This is not how we get to the truth. It is called "reality therapy" and Thanz has left the counselor's office.

[Lurker]

Originally posted by BillHoyt

You have made no such prediction for my experimental method. You simply say it is flawed. Yet Lurker claims to have produced similar results with different data. That requires an explanation. Why did my test show "J"s significantly over-used? Why does Lurker claim his test shows "J"s significantly over-used?


I think there is a misunderstanding. According to my calculation using MY method, yes, J is overused but NOT SIGNIFICANTLY. I will repeat my numbers:

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 <----------

Lurker

[BillHoyt]

Originally posted by Lurker


I think there is a misunderstanding. According to my calculation using MY method, yes, J is overused but NOT SIGNIFICANTLY. I will repeat my numbers:

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 <----------

Lurker
I'm sorry if my post was unclear. With the new data and my counting method you report you found J was significantly overused. So why did your Js rise over four-fold while all other initials rose only two-fold?

[Lurker]

Originally posted by BillHoyt

I'm sorry if my post was unclear. With the new data and my counting method you report you found J was significantly overused. So why did your Js rise over four-fold while all other initials rose only two-fold?

All right. Just trying to keep things clear that they are overused using YOUR counting method, which I believe is the bone of contention Thanz is discussing. I'll let him speculate.

Lurker

[BillHoyt]

Thanz,

Attempt #2:

"Why should the "J"s be so dramatically over-represented? Why did the other 25 letters not inflate equally? How can that be?

Do you claim my results did not happen? Do you claim Lurker's results did not happen? How can you justify declaring them "irrelevant"?"

[Thanz]

Mr. Hoyt - I will answer you one last time. If you don't understand it after this, get help from someone else.

Originally posted by BillHoyt
"Why should the "J"s be so dramatically over-represented? Why did the other 25 letters not inflate equally? How can that be?
Who cares?

You are coming at the problem backwards. First, you must come up with a rational and logically supported data collection method. Then, you analyze the data that you collect. Then you examine the results of the analysis. The data collection method must be rationally supported before any of the other steps are taken.

Your data collection method is not rationally supported. I have explained multiple times why your data collection method is irrational for the analysis we want to perform. I do not also have to analyse the results of the analysis done with your irrational data.

Do you claim my results did not happen?
Nope. I claim that your data collection is flawed, so we cannot rely on your results.
Do you claim Lurker's results did not happen?
Nope. Lurker's results of the method he endorses do not reject the null hypothesis. On your irrational method, it does - but again, we cannot rely on the results of the analysis when the data collection method is irrational.
How can you justify declaring them "irrelevant"?"
Simple. Your data collection is irrational, therefore the results cannot be relied upon, therefore they are irrelevant.

Let's say you collected only data from CO. We know that CO is edited, and cannot be relied upon. But you do it anyway, and come up with "statistically significant" results from that data. Can we rely on it? No, because the underlying data is flawed. It is flawed whether or not we predict the results.

[BillHoyt]

Originally posted by Thanz
Mr. Hoyt - I will answer you one last time. If you don't understand it after this, get help from someone else.


Who cares?

You are coming at the problem backwards. First, you must come up with a rational and logically supported data collection method. Then, you analyze the data that you collect. Then you examine the results of the analysis. The data collection method must be rationally supported before any of the other steps are taken.

Your data collection method is not rationally supported. I have explained multiple times why your data collection method is irrational for the analysis we want to perform. I do not also have to analyse the results of the analysis done with your irrational data.

[B]
Nope. I claim that your data collection is flawed, so we cannot rely on your results.
[B]
Nope. Lurker's results of the method he endorses do not reject the null hypothesis. On your irrational method, it does - but again, we cannot rely on the results of the analysis when the data collection method is irrational.
[B]
Simple. Your data collection is irrational, therefore the results cannot be relied upon, therefore they are irrelevant.

Let's say you collected only data from CO. We know that CO is edited, and cannot be relied upon. But you do it anyway, and come up with "statistically significant" results from that data. Can we rely on it? No, because the underlying data is flawed. It is flawed whether or not we predict the results.

One cannot dismiss data simply because of an a prior declaration they are flawed. Lurker says he repeated the J results with a totally different set of data. If this is right, then there are two different counters, two different sets of data that yield similarly significant results. Your refusal to acknowledge this flies in the face of logic. This includes your own logic, which has yet to identify any flaw in my method that would cause the J-count to inflate so disproportionately relative to all other initials.

"Ye men of Galilee, why stand ye gazing up into heaven?"
-Father Caccini

Cheers,

[Thanz]

Originally posted by BillHoyt


One cannot dismiss data simply because of an a prior declaration they are flawed.
Sure you can. If I want you to count dead crows in a field, and I find out that your count included all dead birds, and that also if you came across a dead bird with it's wing off you counted it as two birds, then I would dismiss the data.

Are you saying that we cannot dismiss data taken from Crossing Over broadcasts?

[BillHoyt]

Originally posted by Thanz

Sure you can. If I want you to count dead crows in a field, and I find out that your count included all dead birds, and that also if you came across a dead bird with it's wing off you counted it as two birds, then I would dismiss the data.
That is a posteriori, not a priori. Your wing example demonstrates that the experimental method was not as advertised. My experimental method was described, and now Lurker claims to have used this method and achieved the same result.

Are you saying that we cannot dismiss data taken from Crossing Over broadcasts?
I did not. We can dismiss the data because we know it is not as advertised. The data are known to be edited and therefore incomplete and we don't know how they are edited. We have no way to verify that the editing process is either representative of the original data or repeatable.


Cheers,

[Thanz]

Originally posted by BillHoyt

That is a posteriori, not a priori. Your wing example demonstrates that the experimental method was not as advertised. My experimental method was described, and now Lurker claims to have used this method and achieved the same result.
If I said to you - I need you to count the dead crows in the field, and you said that you were going to count all the dead birds, and each piece of the dead birds as a separate bird, and then went and did that, the data collection method is still flawed. Just as your data collection method is flawed.

I am not saying that your method was not as described. I am saying that your method, as described, is flawed - just as the dead birds and wings method is flawed.

[BillHoyt]

Originally posted by Thanz

If I said to you - I need you to count the dead crows in the field, and you said that you were going to count all the dead birds, and each piece of the dead birds as a separate bird, and then went and did that, the data collection method is still flawed. Just as your data collection method is flawed.

I am not saying that your method was not as described. I am saying that your method, as described, is flawed - just as the dead birds and wings method is flawed.

You can count bird parts so long as you don't say you counted whole birds. Read some paleontology papers, or simply think about it.

So, how do you explain the fact that Lurker claims to have achieved the same results with my method. Stop avoiding this. Your claimed flaws do not predict inflated "J" counts. How did Lurker get elevated "J" counts?

Cheers,

[Thanz]

Originally posted by BillHoyt

So, how do you explain the fact that Lurker claims to have achieved the same results with my method. Stop avoiding this. Your claimed flaws do not predict inflated "J" counts. How did Lurker get elevated "J" counts?


I haven't been avoiding it - I've been saying it doesn't matter. You count too many Js and you'll end up with inflated J counts.

Why are you focussing on Lurker? He doesn't endorse your counting method. In fact, no one has endorsed your counting method.

Let's say that you are counting all the dead birds in a field to see the proportion of crows to other dead birds. You are counting wing parts and are claiming to be counting dead birds. Then you say to me that unless I can explain why Crows get dismembered more often than other your method of counting parts is valid. It isn't.

You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.

[BillHoyt]

Originally posted by Thanz
I haven't been avoiding it - I've been saying it doesn't matter. You count too many Js and you'll end up with inflated J counts.
Really? How does my method manage to count Js disproportionately?
Why are you focussing on Lurker? He doesn't endorse your counting method. In fact, no one has endorsed your counting method.
Fallacy of the appeal to popularity. My focus is on his claim that he has repeated my results with different data.
Let's say that you are counting all the dead birds in a field to see the proportion of crows to other dead birds. You are counting wing parts and are claiming to be counting dead birds. Then you say to me that unless I can explain why Crows get dismembered more often than other your method of counting parts is valid. It isn't.

You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.
You claimed you can't count dismembered crows. You can. It depends on what you are after. Paleontologists regularly count parts of fossils to determine how many animals they represent. You cannot misrepresent what you are counting. I have not misrepresented the count. And now Lurker claims to have used the same method on different data and he has seen the same J count elevation.

Now you cannot wave your arms about and scream the data are flawed. You must tell us how my method inflates the Js, which is your claim.

Cheers,

[Drooper]

Can I add a spanner to this debate. Regardless of the methodology used, I don't think this "letter counting test" can be used to draw any inference - regardless of the methodoloy. The reason?


Well the hypothesis is that a fraudulent psychic will choose names with more commonly used first letters. A high proportion of these will indicate that they are fishing for a match.

However, think of the opposite - the hypothesis that the psychic is genuine. If names beginning with J are more common, then there will be more J people trying to get through or more people trying to get through to people with an association with names beginning with J etc.

I think you could devise a test of this nature, but it would need to be far more sophisticated, calling on data that gives information on the actual distribution of first name letters in the population. Then you would need to run the test across all 26 letters to find a significant deviation from the population distribution.

Does that make sense?

[Thanz]

Originally posted by BillHoyt

Fallacy of the appeal to popularity.
Nope - simply reporting the results of an informal peer review of your methods - which have been found to be illogical and substandard.

You claimed you can't count dismembered crows. You can. It depends on what you are after. Paleontologists regularly count parts of fossils to determine how many animals they represent.
Blah, Blah, Blah. It is an example, doofus. It was clear from my example that you are counting bird parts and counting each part as a distinct and separate bird, even if the wing is broken off the crow and sitting beside the rest of the body.

You cannot misrepresent what you are counting. I have not misrepresented the count.
Actually, you have. You represent that the count is an accurate count of JE's J guesses, and it isn't.

You must tell us how my method inflates the Js, which is your claim.
No, that is not my claim no matter how much you try and shoehorn me into it. My claim is that your counting method and (the data derived from it) is flawed, and therefore cannot be relied upon for anything, no matter what the results are. I have also pointed out that when the analysis is performed with a logical counting method, the null hypothesis is not rejected.

This is not a claim that your method inflates the Js, but it is evidence that your method does that. However, my criticism of your method is independent of the results that it gets. You need to justify your method independently of the results, which you have been unable to do.

I am not arm waving, I am trying to point out a basic truth (which I see you have chosen not to address): You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.

[BillHoyt]

Originally posted by Drooper
Can I add a spanner to this debate. Regardless of the methodology used, I don't think this "letter counting test" can be used to draw any inference - regardless of the methodoloy. The reason?


Well the hypothesis is that a fraudulent psychic will choose names with more commonly used first letters. A high proportion of these will indicate that they are fishing for a match.

However, think of the opposite - the hypothesis that the psychic is genuine. If names beginning with J are more common, then there will be more J people trying to get through or more people trying to get through to people with an association with names beginning with J etc.

I think you could devise a test of this nature, but it would need to be far more sophisticated, calling on data that gives information on the actual distribution of first name letters in the population. Then you would need to run the test across all 26 letters to find a significant deviation from the population distribution.

Does that make sense?

Yes, it does. This is what we did with the J count. J is the most frequent first initial in the population. When we look at JE's use of J, by my counting method, JE uses J significantly more often than would be expected if his name guesses were simply reflective of the population at large.

I would love to have sufficient data to do the broader test you suggest, but JE's unedited transcripts are few and far between.

Cheers,

[BillHoyt]

Originally posted by Thanz

Nope - simply reporting the results of an informal peer review of your methods - which have been found to be illogical and substandard.

Come back when you can explain how the method inflates the Js. Stop the handwaving and look through the telescope.

[BillHoyt]

Originally posted by Thanz

Nope - simply reporting the results of an informal peer review of your methods - which have been found to be illogical and substandard.

Come back when you can explain how the method inflates the Js. Stop the handwaving and look through the telescope.

[NoZed Avenger]

Originally posted by BillHoyt

Yes, it does. This is what we did with the J count. J is the most frequent first initial in the population. When we look at JE's use of J, by my counting method, JE uses J significantly more often than would be expected if his name guesses were simply reflective of the population at large.



And at the end of the day, an 'inflated J count,' by any methodology on such a small sample . . . . fails to show much of anything.

IMO - and perhaps its because I am not a statistician - its a non-issue. While an interesting mental excercize, it doesn't move the ball forward.


N/A

[Thanz]

Drooper -

I do not know how much of the other thread you have read. It is very long (20+ pages) and discusses the various options for this kind of analysis. I'll try to summarize some of it here for you.

I agree with you that what we would really need is an analysis of all the letters, but as Mr. Hoyt has said, we don't have the data for that. That didn't stop him from jumping on me for asserting that the sample size we have is too small for a meaningful analysis, however.

In any event, the J guesses were compared to the figures in the 1990 census and the distribution of initial letters there. That was the control data used for the general population. It is not perfect, as we don't know if JE's audience is somehow skewed away from the general population, but without knowing that we used the census data as an approximation.

The statistical tool used was the Poisson distribution. Why Poisson you may ask? Well, BillHoyt has suggested it but hasn't explained why it is the appropriate tool for this analysis. I'm not a statistician, so I can't explain the choice of tool to you. Maybe Bill can.

The real dispute has been about the appropriate way to count JE's guesses. When you read the transcripts, you will quickly realize that it is not as straightforward as you may hope. JE rarely just says "I'm getting a J connection here". He usually tosses out some actual names as well.

The first area of controversy is whether we should count "Who is John" as a "J" guess in the same way we would count "I'm getting a J connection here" as a "J" guess. Obviously, the proportion of the population that has the name John is smaller than the proportion of the population that has a name that starts with J. Arguably, counting them as equivalent is a distortion in the primary data. BillHoyt think that this position is malarky, and that every "John" should be counted as a "J".

Next, we need to deal with the situation where JE throws out more than one name at a time. For example, he may say - "I am getting a J connection here, like Jim or John". I say that this should count as one J guess. BillHoyt counts it as three J guesses, essentially counting it the same as if in three seperate readings JE said "I am getting a J connection here". I say that it is one guess, and that equating it with 3 seperate guesses is erroneous.

Bill's first assumption (that each name should be counted the same as a guess of the first letter alone) combined with the second assumption (that each name guess in a string should be counted separately) leads to great distortions in the data, especially if we are comparing it to the general population. He reads "a J connection, like John or Jim" the same as "J, like J or J" which makes no rational or logical sense.

So, now you are sort of up to date on the dispute. Probably more than you wanted to know, but there you go. I apologize if I have wasted your time.

[BillHoyt]

Originally posted by Thanz
I agree with you that what we would really need is an analysis of all the letters, but as Mr. Hoyt has said, we don't have the data for that. That didn't stop him from jumping on me for asserting that the sample size we have is too small for a meaningful analysis, however.

This is what you said, sir:

"I don't know if we can really glean anything significant from such a small sample."

-Thanz (http://host.randi.org/vbulletin/showthread.php?s=&postid=1870044677&highlight=data#post1870044677)

I have demonstrated that exactly such an analysis can be done. Lurker now claims that, not only can it be done, but that another set of data shows the same thing.

[BillHoyt]

Originally posted by Thanz
Bill's first assumption (that each name should be counted the same as a guess of the first letter alone) combined with the second assumption (that each name guess in a string should be counted separately) leads to great distortions in the data, especially if we are comparing it to the general population. He reads "a J connection, like John or Jim" the same as "J, like J or J" which makes no rational or logical sense.
It is not an assumption. It is a conclusion based on evidence from the transcripts. It is a conclusion based on what JE actually does. You continue to dance about two essential facts here:

1)There is nothing in my method that should distort the Js disproportionately. If you think otherwise, then please explain how it does that.

2) Lurker claims to get the same results with a new set of data. This would suggest that my results were not a fluke, and again demand an explanation from you as to how the Js get distorted disproportionately.

[Thanz]

Originally posted by BillHoyt


This is what you said, sir:

"I don't know if we can really glean anything significant from such a small sample."
And I clarified what I meant in subsequent posts. Way to be complete in your presentation.

I have demonstrated that exactly such an analysis can be done. Lurker now claims that, not only can it be done, but that another set of data shows the same thing.
Whoa there, sparky. Lurker makes no such claim. Here is what Lurker says (top of this page):I think there is a misunderstanding. According to my calculation using MY method, yes, J is overused but NOT SIGNIFICANTLY.
Lurker does not endorse your counting method. He states, rather unequivocally, that the J is not significantly overused.

[Thanz]

Originally posted by BillHoyt

It is not an assumption. It is a conclusion based on evidence from the transcripts. It is a conclusion based on what JE actually does.
Please. If you could rationally support your method you would have done so by now. Your "conclusions" are irrational and illogical.

You continue to dance about two essential facts here:
There is no dancing, and the "facts" are not in any way essential.

1)There is nothing in my method that should distort the Js disproportionately. If you think otherwise, then please explain how it does that.
I don't have to explain this as it is not my claim. My claim is that your method is irrational, and the numbers that you get for all of the letters, J and the rest (which affects the total guess count) are incorrect. They are ALL inaccurate, and nothing meaningful can be gleaned from them. It is you who are asserting that meaningful analysis can be conducted with inaccurate numbers, so long as you get the result you want. :rolleyes:

2) Lurker claims to get the same results with a new set of data. This would suggest that my results were not a fluke, and again demand an explanation from you as to how the Js get distorted disproportionately.
Again, no. What part of "You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method." don't you understand?

[BillHoyt]

Originally posted by Thanz
And I clarified what I meant in subsequent posts. Way to be complete in your presentation.

Oh, I know exactly what you said. And your first comment is quite representative. Here are subsequent comments from you:

"So, I disagree that we should "work with what we got", as what we got is not adequate to give us any sort of accurate results."

http://host.randi.org/vbulletin/showthread.php?s=&postid=1870046074&highlight=data#post1870046074

That has been demonstrated to be untrue. You may disagree with the method, but that has nothing to do with the claim that we can't get any sort of accurate results. Which, it must be pointed out, is different from your first comment.

"How much more specific do I need to get? I don't think that the sample size is big enough. If we just use juninho's numbers for the sake of argument, some letters have a probability of about one in fifty. In 78 guesses, it is not anything meaningful if that letter does not appear. In 700 guess, that is much more relevant.

"Let me ask it this way: If you picked 78 random people off the street, do you think that the name distribution will mirror that of the total population? Or do you think it is possible that it will be skewed? What if you took 1000 people at random, and compared their names? I think you would agree that the 1000 person sample is much more likely to mirror the name distribution of the population in general than the 78 person sample. I don't think that we could conclude much from the 78 person sample. It is just too small."

http://host.randi.org/vbulletin/showthread.php?s=&postid=1870046999&highlight=data#post1870046999

This post makes the erroneous assumption that we need to test the whole distribution of names. We do not. Any portion of them could be used to yield meaningful results.


Whoa there, sparky. Lurker makes no such claim. Here is what Lurker says (top of this page):
Lurker does not endorse your counting method. He states, rather unequivocally, that the J is not significantly overused.
Thanz, I'll thank you to read what I wrote. I said nothing about "endorse". It matters not that Lurker didn't like or doesn't know what to make of the result. I simply said that he claims to have repeated my result with different data.

[BillHoyt]

Originally posted by Thanz
I don't have to explain this as it is not my claim. My claim is that your method is irrational, and the numbers that you get for all of the letters, J and the rest (which affects the total guess count) are incorrect. They are ALL inaccurate, and nothing meaningful can be gleaned from them. It is you who are asserting that meaningful analysis can be conducted with inaccurate numbers, so long as you get the result you want. :rolleyes:

If the method distorts the data, it equally distorts all the data. The net effect is not to alter the significance of the J count.

Stop dodging this issue.

[Thanz]

Originally posted by BillHoyt


If the method distorts the data, it equally distorts all the data. The net effect is not to alter the significance of the J count.

Stop dodging this issue.
There is no reason to believe that the distortion in the data is equal across all letters. It is much better to pick a method that is accurate rather than hoping the distortions are equal. Unless you are saying that data we know to be inaccurate is somehow better if we assume the inaccuracies are equal?

Again, what part of "You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method." don't you get?

Stop dodging this issue.

[BillHoyt]

Originally posted by Thanz

There is no reason to believe that the distortion in the data is equal across all letters. It is much better to pick a method that is accurate rather than hoping the distortions are equal. Unless you are saying that data we know to be inaccurate is somehow better if we assume the inaccuracies are equal?

I have already, exhaustively identified the problems with the counting method you propose. I have pointed out why they are irrational in that they make a demonstrably false assumption. JE does not map a name guess to a person. You presume a process that you wish to test. We are out to test what process JE actually uses. From the transcripts it is clear that he accepts dogs and entire families as acceptable responses to his name guesses.

There is nothing in my method that should cause the Js to inflate to significance. Period. Yet the method yielded that. Apparently, twice now.

Why, Thanz?

[Thanz]

Mr. Hoyt -

You have not demonstrated the rationality of YOUR method. Any of the arguments that you apply to my proposal also apply to your method.

Again:

You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.

Stop ignoring this issue.

[BillHoyt]

Originally posted by Thanz
Mr. Hoyt -

You have not demonstrated the rationality of YOUR method. Any of the arguments that you apply to my proposal also apply to your method.

Again:

You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.

Stop ignoring this issue.

I count multiple JE guesses because JE makes multiple guesses.

1) Why, exactly is it irrational to count all of JE's guesses?
2) Why, exactly, does doing so wrongly indicate he guesses Js significantly more often than is expected?

[Thanz]

Originally posted by BillHoyt


I count multiple JE guesses because JE makes multiple guesses.

1) Why, exactly is it irrational to count all of JE's guesses?
You need me to explain this to you AGAIN? We have been over this ad nausem. Briefly, it is irrational to think that "I am getting a J connection, like Jim or John" is three guesses rather than one guess. He is fishing for one connection, with a J initial.

2) Why, exactly, does doing so wrongly indicate he guesses Js significantly more often than is expected? You can find out on the radio station WGAF. The results are irrelevant when the data is inaccurate.

And now, for the sixth time, in the hopes that it may seep into your thick skull:

You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.

[BillHoyt]

Originally posted by Thanz
You need me to explain this to you AGAIN? We have been over this ad nausem. Briefly, it is irrational to think that "I am getting a J connection, like Jim or John" is three guesses rather than one guess. He is fishing for one connection, with a J initial.
The bottom line, sir, is this:

1. I make no unwarranted assumptions about what the names mean.
2. The assumptions you continue to make about the meaning of the guesses have been demonstrated to be wrong.
3. I simply count JE's guesses, and total them.
4. You call that "irrational", and cite your unwarranted presupposition. That is a circular argument.
5. You ignore the data I presented, claim it is flawed, yet can offer no explanation as to how it would cause the J-counts to become significant.
6. You ignore the data Lurker presented, claim it is flawed, yet can offer no explanation as to how it would cause his J-counts to become significant.
7. You claim that my method of simply counting name guesses needs to be justified. This, again, you cannot provide evidence for. You simply resort to the bald assertion.

Cheers,

[Thanz]

Bill Hoyt -

If it is incorrect to assume that "A J connection, like Jim or John" is one guess because you find one example where he says "an Sh name" and the sitter says "that is my sister's last name", then it is equally incorrect to assume that "Jane or Jean" is two J guesses when the sitter identifies Jane as her sister. And I can assure you, it is much more common for JE to spit out multiple name guesses for one person than on guess for multiple persons.

So, once again Mr. Hoyt, the bottom line is this:

You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.

[BillHoyt]

Originally posted by Thanz
Bill Hoyt -

If it is incorrect to assume that "A J connection, like Jim or John" is one guess because you find one example where he says "an Sh name" and the sitter says "that is my sister's last name", then it is equally incorrect to assume that "Jane or Jean" is two J guesses when the sitter identifies Jane as her sister. And I can assure you, it is much more common for JE to spit out multiple name guesses for one person than on guess for multiple persons.

So, once again Mr. Hoyt, the bottom line is this:

You don't justify your method by pointing at the results. You justify your results by pointing at the soundness of your method.

Unholy crap, Thanz! What twisted pretzel logic is that? I make no assumptions. I simply count names. Period. We are trying to get to JE's process. You make the assumption that all names in one reading refer to one person. I don't make the assumption that each name refers to a different person. I make NO assumptions about the mapping.

:hb: