The "Process" of John Edward

[T'ai Chi]

A binomial distribution is more appropriate, I think, though it and a Poisson distribution yield, in this instance, similar results.

If JE only guessed high frequency names that start with the letter J, then OK.

However, these analyses ignore the other high frequency letters a, c, d, m, and r. Considering JE uses these letters too in his readings and not just the letter J, why completely ignore these other high frequency letters in the statistical analysis? The analysis offered doesn't seem to fit the situation.

 

[69dodge]

Originally posted by BillHoyt
Re-read your source. With understanding, please. That is not a description of the conditions under which Poisson is valid. Neither is it a description of Poisson as "Binomial lite". If you have real trouble understanding this, I will help you. That is an honest offer, if you approach this honestly. Unfortunately, your approach is part of the problem for me.

Please help me. Nothing Lurker wrote leads me to believe that he misunderstood his source. What do you think is his misunderstanding?

Now let me give you an exercise from Hogg & Craig's Introduction to Mathematical Statistics, third edition. It is from page 98. If you don't have a copy, just ask anyone who's been through Cherry's Stat 231 course here at _____'s exotic dance bar. She used to hold it late in the evenings, or early morning's, depending on your perspective.

"3.23 Let X have a Poisson distribution with mu=100. Use Chebyshev's inequality to determine a lower bound for Pr(75 < X < 125)."

Chebyshev's inequality places an upper bound on the probability that a random variable is far from its mean. Specifically, letting mu be the mean of X, V be the variance of X, and d be the distance that we're interested in, it says that<blockquote>Pr( |X-mu| >= d ) <= V/d².</blockquote>The probability of the complementary event, namely that X is within the distance d of its mean, is therefore greater than 1 - V/d².

In this exercise, d is 25, the mean is 100, and, since the variance of a Poisson random variable equals its mean, V is also 100. So we get<blockquote>Pr(75 < X < 125) > 1 - 100/25² = 21/25.</blockquote>Chebyshev's inequality is nice because it applies to any random variable regardless of its distribution, provided we know its mean and variance. (Well, any random variable that has a mean and variance, anyway. But that's most of them.) However, if we know the exact distribution, as we do here, we can often do much better. Chebyshev's inequality here yields a lower bound of 21/25 = 0.84. The actual value of the probability is about 0.986.

Poisson with mu=100? What justifies that?

I don't understand. It's a textbook exercise. It says that X has a Poisson distribution with mu = 100. So by definition X does have that distribution. What kind of justification are you looking for?

I don't see the relevance of a textbook exercise about Chebyshev's inequality, even one which happens to involve a Poisson distribution, to the question of when in real life it is appropriate to use a Poisson distribution to model a situation we're faced with.

 

[69dodge]

If JE only guessed high frequency names that start with the letter J, then OK.

However, these analyses ignore the other high frequency letters a, c, d, m, and r. Considering JE uses these letters too in his readings and not just the letter J, why completely ignore these other high frequency letters in the statistical analysis? The analysis offered doesn't seem to fit the situation.

I was just comparing a binomial distribution to a Poisson distribution. I think BillHoyt's analysis of J's is ok as far as it goes. Nor do I see anything wrong with the chi-square test you propose. I don't think a great deal of weight should be given to the results of either of them, however, due to their sensitivity to the exact method of tallying the guesses and the lack of agreement about which method is the "correct" one.

 

[Lurker]

69dodge:

Thanks for the analysis. I knew I would not get an answer out of Bill Hoyt as he ignored my most basic questions. Glad to see SOMEONE truly understand stats here. I figured once I started talking integrals I would lose Bill. It seems I was correct.

Regardless, I have not gone through your binomial analysis but how did you calculate the standard deviation for the bionomial distribution?

Lurker

 

[Lurker]

Bill:

It appears you took the coward's way out. You chose not to address any of my questions. I will summarize them here for you:

1. What are the limits of Poisson Distribution use?
2. How does one calculate the error?
3. How did you avoid using N in your Poisson calculation? I specifically see you getting the expected mean of 11.05 by multiplying 85*0.13. We'll ignore your roundoff. Isn't N*p being used here? Show me where I am wrong.
4. Am I correct in saying that as p rises the error using Poisson rises?
5. Was my example { n=100, p=0.9, Use Poisson. What do you get for the probability that the count will be higher than 100? Since I am a nice guy, I will give you the answer. The answer is 1-0.8651=.1349} wrong and if you believe so, provide the evidence of it.

you see, Bill, unless you answer some of these questions, you really are just trying to bully people. You ask questions, turnabout is fair play.

Turn the light on, Bill.

Lurker

 

[BillHoyt]

Well, Bill, how can you even ask?

After all, one hardly needs to be a statistician to see that if JE says "I'm getting a 'J' name, like John, Jenny, Joan" and I count that as "1" guess of 'J' and you count that sentence as showing 4 separate guesses of 'J'...both totals--the total overall number of guesses, as well as the guesses of letter 'J'--will be greater for your results than mine.

Which is exactly what your tally of the 1998 LKL readings shows. [/B]

Very good, Clancie. Now what happens as that denominator increases? What happens as all those "R", "Ronnie", "Reginald" and "DA name like Danny or David" get tallied. What is the number fed into the analysis, Clancie, and what happens?

The denominator goes up, lowering the "J" frequency and threatening to wash it out to insignificance. I applied the technique equally to "J"s and to "D"s and "M"s and "R"s, "B"s and all the letters. I did not distort the data as you keep trying to insinuate. I tried to resolve the problem of JE's blathering all over the place.

 

[neofight]

originally posted by CFLarsen:
neofight,

Did you - or did you not - change your account of the Malibu Shrimp reading?

Concerning anything substantive? No. The only thing that I conceded was that it appeared JE might have been wrong in a part of his interpretation about why the "Malibu Shrimp" recipe was kept "secret".

He interpreted it as meaning that it was secretly based upon one of Deborah's mother's recipes, which he believed was what he was being shown him by her mom, and Deborah said the real reason it was kept secret was because of the questionable clams that she and her friend used when making it, although she did admit that the recipe was based, at least loosely, upon her mom's recipe.

Instig8R feels JE pressured Deborah into admitting that fact, and considers this is a monumental matter. I feel that it could indeed be true, that the recipe was loosely based on one of Deborah's mother's recipes, since most daughters do tend to learn some cooking from their moms.

In any case, I also feel that this whole matter was blown way out of proportion by Instig8R, since the balance of that reading stands, and it was a good reading, with a lot of excellent, accurate hits, and in no way does any of that hinge upon this one, rather irrelevant point.......neo

 

[69dodge]

Regardless, I have not gone through your binomial analysis but how did you calculate the standard deviation for the binomial distribution?

I'm not sure what you're asking. I didn't calculate the standard deviation of the binomial distribution, as there was no need to.

In any case, the variance of a binomial distribution with parameters n and p is np(1 - p), and the standard deviation is, as always, the square root of the variance.

 

[Clancie]

Posted by Bill Hoyt

Very good, Clancie. Now what happens as that denominator increases? What happens as all those "R", "Ronnie", "Reginald" and "DA name like Danny or David" get tallied. What is the number fed into the analysis, Clancie, and what happens?

The denominator goes up, lowering the "J" frequency and threatening to wash it out to insignificance. I applied the technique equally to "J"s and to "D"s and "M"s and "R"s, "B"s and all the letters. I did not distort the data as you keep trying to insinuate. I tried to resolve the problem of JE's blathering all over the place.

I'm just amazed at how, no matter what it is, Bill, you always insist on being right.

The problem with counting like that is two-fold: (1) "I get a 'J' name, like John, Joanne, Joan, Jenny" is [I[not[/I] difficult to count--as you say it is. It's easy....JE is guessing one 'J' name. The "problem" you "resolve" is one of your own creation, Bill. It's very clear and consistent to use this one guess on 'J', method throughout the readings. What you've done is the one that creates problems and misrepresentations of his actual number and pattern of guesses.

(2) Since he uses 'J's more frequently than other letters, what if he also lists names after saying a 'J' more frequently than other letters? You say you apply the technique equally, but it's not a technique you want to apply, if you're really interested in seeing the pattern of how he uses each letter.

Doing the same thing for the other letters doesn't solve the problem, since JE is inconsistent in doing this (and, of course, the technique will pad results for letters he uses more often). It just makes it look as if your overall sample of JE guesses of letters is bigger than it really is and skews the frequency even more, since JE is inconsistent--does it more with some letters than others and not at all with most.

More to the point, though, "I'm getting a J name like Jenny, John, Joan" is one guess, Bill, only one 'J' name.

 

[BillHoyt]

Would someone please explain to Lurker why Poisson is not in error in his example? Will someone please help him understand there is no "N" in the Poisson equation?

 

[BillHoyt]

I'm just amazed at how, no matter what it is, Bill, you always insist on being right.

The problem with counting like that is two-fold: (1) "I get a 'J' name, like John, Joanne, Joan, Jenny" is [I[not[/I] difficult to count--as you say it is. It's easy....JE is guessing one 'J' name. The "problem" you "resolve" is one of your own creation, Bill. It's very clear and consistent to use this one guess on 'J', method throughout the readings. What you've done is the one that creates problems and misrepresentations of his actual number and pattern of guesses.

(2) Since he uses 'J's more frequently than other letters, what if he also lists names after saying a 'J' more frequently than other letters? You say you apply the technique equally, but it's not a technique you want to apply, if you're really interested in seeing the pattern of how he uses each letter.

Doing the same thing for the other letters doesn't solve the problem, since JE is inconsistent in doing this (and, of course, the technique will pad results for letters he uses more often). It just makes it look as if your overall sample of JE guesses of letters is bigger than it really is and skews the frequency even more, since JE is inconsistent--does it more with some letters than others and not at all with most.

More to the point, though, "I'm getting a J name like Jenny, John, Joan" is one guess, Bill, only one 'J' name. [/B]

And the spice names? And "Helen/Ellen"? And the case where he gives a single initial and says there are two different people? And The "C or K name"? It goes on and on and on. What do you do with those?

You continue to these. You also continue to see an "inflation" of "J" and ignore the "inflation" of "R" and "B" and "D" and "M".

 

[Lurker]

Bill, Bill, Bill,

Still refusing to answer direct questions. Isn't it a tad hypocritical when you accuse others of not answering questions?

Bill, how did YOU calculate the expected value for your "J" analysis again? Again, in your own words:

-------------------------------------------------------------------------------
Originally posted by BillHoyt


...According to the US census data presented earlier, "J" surnames are 13.36% of the total population. In this analysis of 85 JE name guesses, I counted 18 "J" names. I calculated the expected number of "J"s (formally, the "expectation function") as 11.05.

I used the Poisson function to model the population. With an expected mean of 11.05, ...
--------------------------------------------------------------------------------

So now you are claiming n makes no appearance in the formula yet YOU used it in your analysis, did you not?

You going to answer any of the questions directed towards you or are you going to continue to run and divert?

Lurker

 

[Clancie]

According to the US census data presented earlier, "J" surnames are 13.36% of the total population

Surnames? Is this a misprint?

 

[Lurker]

Clancie:

Man, I REALLY hope that was just a typo. Yikes!

Lurker

 

[Lurker]

The results were a revised total guess count of 85. I then tallied the "J"s separately. I picked the "J"s because they are the most frequent initial. According to the US census data presented earlier, "J" surnames are 13.36% of the total population. In this analysis of 85 JE name guesses, I counted 18 "J" names. I calculated the expected number of "J"s (formally, the "expectation function") as 11.05.

Oh, my. I presume you meant first name, not surname. Right Bill? Please say this is a simple typo so our analyses were not a total waste of time.

Thanks!

Lurker

 

[BillHoyt]

Yes, I meant forenames. My error. Thanks for the correction.

 

[Walter Wayne]

And the spice names? And "Helen/Ellen"? And the case where he gives a single initial and says there are two different people? And The "C or K name"? It goes on and on and on. What do you do with those?

You continue to these. You also continue to see an "inflation" of "J" and ignore the "inflation" of "R" and "B" and "D" and "M".

Helln/Ellen and C or K both are in the "not J" bin. Spice names are more difficult, but since it is not a guess of the initial it probably doesn't belong within our sample.

It would be more difficult if he gave a J or G reading, in which case it clearly belongs with the test, but does not fit easily into the "J" bin or the not "J" bin.

Walt

 

[Clancie]

Posted by Walter Wayne

Helln/Ellen and C or K both are in the "not J" bin. Spice names are more difficult, but since it is not a guess of the initial it probably doesn't belong within our sample.

This all seems very logical and easy to follow, doesn't it, Bill?

It would be more difficult if he gave a J or G reading, in which case it clearly belongs with the test, but does not fit easily into the "J" bin or the not "J" bin.

I agree, but fortunately there was only one example like that in these readings so I just counted it as a "J", since he did mention the "J or G" sound, and we're not tallying "G".

 

[Thanz]

I went back to Renata's LKL thread and did some totalling of my own. I tried to take the approach that each person that JE was trying to show a connection with should be counted as one guess. So, if he said 'a "J" name, jim or john", I counted one J and one guess.

There were a couple of exceptions to this, typically where he tries to widen the net. So, if he says "Helen", I count one guess. If he then expands it to "Helen or Ellen", I count it as another guess - for the E. Similarly, I count J or G as one J guess and one G guess as he is trying to expand the net.

Needless to say, I get numbers substantially different than BillHoyt's. In the LKL readings posted by Renata, I counted 43 guesses, of which 9 were J. If I plug this into the Poisson calculator, I get a probability of >= 9 of .128, which means that we cannot reject the null hypothesis.

 

[CFLarsen]

Concerning anything substantive? No. The only thing that I conceded was that it appeared JE might have been wrong in a part of his interpretation about why the "Malibu Shrimp" recipe was kept "secret".

No, that is not correct. You also changed your mind about the Entenmann guess. Instig8r will probably fill us in with additional changes.

But OK, you changed your account of the Malibu Shrimp reading, because of editing of content.

Do you think it is impossible that this happens on a regular basis? Or do you consider editing of content "blown way out of proportion"?

In fact, is there any limit to what JE can get away with, without you sitting up and taking notice?