The "Process" of John Edward

[Lurker]

Finally, you might begin to answer your own question about Houston if you turn this gaffe around.

When I look back at my college career I can only think of two teachers that I truly thought were bad teachers. Both of them had the same problem - they refused to answer questions.

Lurker

 

[BillHoyt]

First, what insults? Where in my post did I insult anyone?

Gee, I wonder.

Do you disagree that small sample sizes can produce strange and unreliable results?

Never said that I disagree with that.

If my numbers are wrong, what are the right numbers for Lurker's office and the letter A?

You need to use the cumulative probability density, and to be sure you are accumulating over the tail of interest.

I have already admitted that my knowledge of stats is limited. But even with my limited understanding of stats, I can see that Lurker's office representation of the letter "A" is further away from the norm than JE's guesses of the letter "J". Or are you saying that this is incorrect as well?

I didn't say that. I said, in fact, that this can happen and not be significant. I chose the next letter in the alphabet and had the opposite result. Now what does that mean?

Cheers,

 

[BillHoyt]

When I look back at my college career I can only think of two teachers that I truly thought were bad teachers. Both of them had the same problem - they refused to answer questions.

Lurker

The root of "educate" comes from the Latin, educere. Most skeptics I know decidedly do not want to be spoon-fed information. Now I know everybody's different, of course.

BTW, Randi does NOT write for Skeptical Inquirer. Just thought you'd like to know that.

Cheers,

 

[Thanz]

Gee, I wonder.

I'm serious. I am not aware that I insulted anyone.

Never said that I disagree with that.

Good. Nice to know I am not completely off target.

You need to use the cumulative probability density, and to be sure you are accumulating over the tail of interest.

You bet. As should be painfully obvious, I don't know how to do that. Can you post the numbers please?

I didn't say that. I said, in fact, that this can happen and not be significant. I chose the next letter in the alphabet and had the opposite result. Now what does that mean?

Ah, but I am only concerned with the A-hole in Lurker's office. It is not a B-hole, so I don't need to look at B.

If it can happen in Lurker's office and not be significant, why would it be significant for JE?

 

[Lurker]

Bill:

>BTW, Randi does NOT write for Skeptical Inquirer. Just thought >you'd like to know that.

And this nonsequitor was included because...?

It is funny that other people use the same comaprisons I use. And even the same math I use. To wit:


http://www.talkleft.com/archives/003306.html

"A study of the city's murder rate shows an unlikely factor at the heart of the violence. Chicago's rate is three times that of New York not because of policing, but because of a lack of good, affordable housing."

I don't care about the argument presented here but the math. Clearly they used the same method of comparison that I used. Are they wrong too?

Lurker

 

[Thanz]

Just wanted to say that I found a different Poisson calculator, and I think that it will give me the numbers that BillHoyt uses.

Calculator here

For Lurker's office, the probablity of 6 names with A in 231 is .007559, or less than 1%.

Must be a huge A-hole.

Or, I have screwed something up again.

 

[Lurker]

Oopsy, here is someone else using the same comparison methodology I am using.

"The third group, with a total of 55 cities, contains 8,316,455 people, with 1,741 murders, for an average rate of 20.93 per 100,000 -- roughly twice the California average."

What's that Bill? I can't compare subpopulations to a a census? Isn't that what they are doing here? And they are using what measure of comparison? Percentages? Oh no. they can't do that, can they?

Lurker

 

[Lurker]

Source for previous: http://members.aol.com/gunbancon/Frames/CramerMurder.html

Now Bill, if you want to argue that the confidence interval for a 5% level of significance would be too broad to make inferences on the specific frequencies for each letter I will agree with you. But if you had increased the sample size tremendously my guess is that confidence interval would tighten up considerably and eventually you would get confidence intervals that would not have the possiblity of all letters being practically the same.

And amazingly, I am back to where I first interrupeted this sorry thread. I am totally with Thanz on the opinion that a sample size of 78 would be far too small to have any idea on what the true means are for each letter. If Bill Hoyt would like to differ in opinion, he is welcome but I ouwld then ask him what he thinks a good size would be (and provide the significance too.)

Lurker

 

[BillHoyt]

Oopsy, here is someone else using the same comparison methodology I am using.

"The third group, with a total of 55 cities, contains 8,316,455 people, with 1,741 murders, for an average rate of 20.93 per 100,000 -- roughly twice the California average."

What's that Bill? I can't compare subpopulations to a a census? Isn't that what they are doing here? And they are using what measure of comparison? Percentages? Oh no. they can't do that, can they?

Lurker

Don't strawman me. I never said that. I have made the point six ways from Sunday. Here, for example: "You simply can't compare percentages unless you know the denominators are truly the same."

 

[BillHoyt]

Just wanted to say that I found a different Poisson calculator, and I think that it will give me the numbers that BillHoyt uses.

Calculator here

For Lurker's office, the probablity of 6 names with A in 231 is .007559, or less than 1%.

Must be a huge A-hole.

Or, I have screwed something up again.

Yes, you have. I said we need the cumulative probability function here.

 

[BillHoyt]

Bill:

>BTW, Randi does NOT write for Skeptical Inquirer. Just thought >you'd like to know that.

And this nonsequitor was included because...?

It is funny that other people use the same comaprisons I use. And even the same math I use. To wit:


http://www.talkleft.com/archives/003306.html

"A study of the city's murder rate shows an unlikely factor at the heart of the violence. Chicago's rate is three times that of New York not because of policing, but because of a lack of good, affordable housing."

I don't care about the argument presented here but the math. Clearly they used the same method of comparison that I used. Are they wrong too?

Lurker

No, they are not wrong.

 

[Lurker]

Don't strawman me. I never said that. I have made the point six ways from Sunday. Here, for example: "You simply can't compare percentages unless you know the denominators are truly the same."

But Bill, the denominator is NOT the same in the examples provided. Why in the very one you responded to they are comparing the murder rate of a specific city (Population A) to the murder rate of California (Population B). And the devils, they even use percentages to make the comaparison ("roughly twice")!

You had better write a letter to them showing them the error of their ways. Yes, I know it will be a lot of work as it will take you 6-7 letters to try and get across why you are right and they are wrong but certainly they will rewrite their studies in light of your information.

Lurker

 

[BillHoyt]

But Bill, the denominator is NOT the same in the examples provided. Why in the very one you responded to they are comparing the murder rate of a specific city (Population A) to the murder rate of California (Population B). And the devils, they even use percentages to make the comaparison ("roughly twice")!

You had better write a letter to them showing them the error of their ways. Yes, I know it will be a lot of work as it will take you 6-7 letters to try and get across why you are right and they are wrong but certainly they will rewrite their studies in light of your information.

Lurker

As I see it, you have three choices here, Lurker:

o Try not to understand
o Try to understand and agree
o Try to understand and disagree

But until you understand what I am saying, we can't discuss the issue. Now you can use the Aussie web page to give you more clues.

 

[Thanz]

Yes, you have. I said we need the cumulative probability function here.

Well, when I did the calculation, it spat out columns for single and cumulative. Then it had probabilities listed. For your J numbers, it shows a probability of >=18 at 0.033393, with an expected number of 11.05.

For your B numbers, it shows a prob. of <=9 at 0.382435.

Using the same thing for A, with an expected average of 15.015, and a count of 6, it shows a probability of <=6 at 0.007559.

Where did I go wrong? Why are my results consistent for your posted J and B results, but my A results are wrong?

 

[BillHoyt]

Well, when I did the calculation, it spat out columns for single and cumulative. Then it had probabilities listed. For your J numbers, it shows a probability of >=18 at 0.033393, with an expected number of 11.05.

For your B numbers, it shows a prob. of <=9 at 0.382435.

Using the same thing for A, with an expected average of 15.015, and a count of 6, it shows a probability of <=6 at 0.007559.

Where did I go wrong? Why are my results consistent for your posted J and B results, but my A results are wrong?

My mistake. I took your sentence here literally: "For Lurker's office, the probablity of 6 names with A in 231 is .007559, or less than 1%."

Yes, .007... is the tail area for <=6.

Cheers,

 

[Thanz]

My mistake. I took your sentence here literally: "For Lurker's office, the probablity of 6 names with A in 231 is .007559, or less than 1%."

Yes, .007... is the tail area for <=6.

Cheers,

Now, for the million dollar question:

Isn't the statistical support for the A-hole just as strong as your statistical support for JE's cold reading? In fact, isn't the support for the A-hole stronger than support for cold reading?

 

[Lurker]

As I see it, you have three choices here, Lurker:

o Try not to understand
o Try to understand and agree
o Try to understand and disagree

But until you understand what I am saying, we can't discuss the issue. Now you can use the Aussie web page to give you more clues.

I have read the Aussie site and I guess I will fall into category three above. Let me see what the Aussie site says again:

1. Non-demographic differences between the census and sample. Yep, I addressed that already. And since Bill claims my method is wrong mathematically, this does not apply.

2. Use percentages. Yep, I did use percentages, not absolute numbers. Can't see any conflict with this one. Clearly this is where a looming math error would exist but since I DID use % I don't see it.

3. Use specific geography. Again, not a problem with the math itself but how the sample was obtained. Also addressed by me previously.

and so on...

I am unaware of where my MATH is in error. Please be specific here, Bill. We have gone back nad forth quite a bit and you always refrain from being explicit.

thanks!

Lurker

 

[BillHoyt]

Now, for the million dollar question:

Isn't the statistical support for the A-hole just as strong as your statistical support for JE's cold reading? In fact, isn't the support for the A-hole stronger than support for cold reading?

Fascinating display, Thanz.

 

[BillHoyt]

I have read the Aussie site and I guess I will fall into category three above. Let me see what the Aussie site says again:...

Lurker,

Which part of "You can't simply look at 2.6% and 6.5% and say they are 60% different without knowing you are really comparing apples with apples" don't you understand? Do you think you can compare 2.6% of an inch an hour and 6.5% of a foot per hour and say they are 60% different? Do you honestly not see the relevance of that example?

 

[Lurker]

Lurker,

Which part of "You can't simply look at 2.6% and 6.5% and say they are 60% different without knowing you are really comparing apples with apples" don't you understand? Do you think you can compare 2.6% of an inch an hour and 6.5% of a foot per hour and say they are 60% different? Do you honestly not see the relevance of that example?

Again you try to obfuscate with the same poor analogy. My units are PEOPLE, not inches and feet. Not black people and white people.

So NO, I do not see the relevence of your example. If anything, I am starting to think you are being purposely lazy in your generalizations.

Funny how when I quote a study that used the same methodolgy as mine you backed off and said they were correct. Yet you refuse to define the difference. They used POPULATION A as a subset of POPULATION B and used % to show the difference. Why is MY math in error and not theirs? It is a simple question, Bill. I am starting to think you have no answer and will insist on your inch/hour example or apples/oranges example til the cows come home.

Perhaps you are not as strong in math as you think you are...


Lurker