Cold Reading Demos at TAM2

[Thanz]

Thanz,

Try with statistics. Not lengthy explanations.

Just statistics.

Why? Because you can't actually refute my explanation?

Aren't you trying to put the cart before the horse here? Don't you need to have a logical and sound experimental method before you analyze the results?

Stats are a tool for analyzing data. Before you can do this, you need to get the data. You need a sound method for collecting the data. I am attacking the soundness of Hoyt's data collection. If his data collection is flawed, the results of any stat analysis based on it cannot be trusted. Let's make sure the data colelction is sound before we argue about the appropriate analysis, shall we?

 

[Thanz]

You call me a liar. You apologize not for that, but for "forgetting." And you want me to cooperate with you. Yes, folks, the American hat trick. Stupid, stubborn and arrogant is a dangerous mix, sir.


Go away.

You are correct. You did address the birds, after a fashion. I should not have called you a liar because I find the explanation lacking. I apologize for calling you a liar.

Now can you please actually address the points I am making?

 

[CFLarsen]

Why? Because you can't actually refute my explanation?

Not at all. This is about statistics, Thanz. Please use statistics in a discussion about statistics.

Aren't you trying to put the cart before the horse here? Don't you need to have a logical and sound experimental method before you analyze the results?

Stats are a tool for analyzing data. Before you can do this, you need to get the data. You need a sound method for collecting the data. I am attacking the soundness of Hoyt's data collection. If his data collection is flawed, the results of any stat analysis based on it cannot be trusted. Let's make sure the data colelction is sound before we argue about the appropriate analysis, shall we?

No, you are not "attacking the soundness of Hoyt's data collection". Remember the Poisson calculator you found on the Internet?

By using that, you have agreed that Hoyt's data are sound. Why else use it??

 

[Thanz]

No, you are not "attacking the soundness of Hoyt's data collection". Remember the Poisson calculator you found on the Internet?

By using that, you have agreed that Hoyt's data are sound. Why else use it??

Are you actually following this discussion? It is all about data collection. I used the Poisson calculator because that is the test Hoyt used. That in no way means I endorse his counting data. I went and did my own count, and came up with different results. That is the heart of the dispute - who has the more correct count.

Let's recap:

Hoyt's method counts "A J connection, like John or Joe" as three separate guesses of the first initial J. The same as if JE had just said "J" connection in 3 separate reading to 3 separate people.

My counting method counts "A J connection, like John or Joe" as one guess for the first initial J, as he is only making one real guess for J.

With that in mind, my count was 9 J's out of 43 guesses. Hoyt's was approximately double, 18 out of 85. Using Poisson, 9 of 43 was not significant (although not very likely either) but 18 of 85 was significant. However, if Hoyt's counting is wrong, his result is wrong.

The whole argument is about counting. Any questions?

 

[CFLarsen]

Are you actually following this discussion? It is all about data collection. I used the Poisson calculator because that is the test Hoyt used. That in no way means I endorse his counting data. I went and did my own count, and came up with different results. That is the heart of the dispute - who has the more correct count.

◊◊◊◊◊◊◊◊! By arguing from the data, using the calculator, you have endorsed his data.

Why use the same calculator, if you thought the data was bad? Were all your posts about the Poisson calculator simply a waste of time?

◊◊◊◊◊◊◊◊, Thanz.

 

[BillHoyt]

Why? Because you can't actually refute my explanation?

Aren't you trying to put the cart before the horse here? Don't you need to have a logical and sound experimental method before you analyze the results?

Stats are a tool for analyzing data. Before you can do this, you need to get the data. You need a sound method for collecting the data. I am attacking the soundness of Hoyt's data collection. If his data collection is flawed, the results of any stat analysis based on it cannot be trusted. Let's make sure the data colelction is sound before we argue about the appropriate analysis, shall we?

I have defined the statistic to be name guesses, and include either bare initials or full names. I apply that method to every name guess, and group the results into "J"s and "non-J"s. Each guess is a random event. There is no doubling, trebling or any other such factor applied to the numerator and denominator, sir. All randomly-uttered letters are counted equally and consistently.

What we find is that the "J"s become significant. That is, JE shows a clear bias to utter "J" name examples. If you still believe that this technique can create "J"s as artifacts, explain how. If you believe all the multiples are artifactual, explain how the "J"s are propelled into significance when, in fact, they ought to have regressed to the mean.

Now if you are still foolhardy enough to repeat your inane assertion that "we don't know if it favours one variable over all others as we have only ever tested the J," then I refer you to your own specious coin flip example. Explain how the head flips don't compete with the tail flips.

You will need to do plenty of head-over-tail flips yourself to get around that last inanity.

 

[Jeff Corey]

Moi, I thought the Poisson calculator was a bit fishy.

 

[Thanz]

Bullsh*t! By arguing from the data, using the calculator, you have endorsed his data.

Why use the same calculator, if you thought the data was bad? Were all your posts about the Poisson calculator simply a waste of time?

Bullsh*t, Thanz.

Please, Mr. Larsen, you are making a fool of yourself. And you may want to edit your language before a mod has to.

The Poisson calculator is a statistical tool. It can be applied to a bunch of different data sets. Agreeing to use the tool that Hoyt put forth is in no way an endorsement of the data set he compiled.

There were a bunch of transcripts posted from LKL by Renata. Is it your position that only Mr. Hoyt can collect data from those transcripts? That only he can count guesses?

Or is it your position that only Hoyt can use the Poisson stat tool? that makes no sense either. Both Mr. Hoyt and I are trying to count the same thing - JE guesses. We disagree on the best method to do this.

That does not mean that we must also disagree on the stat method used to analyze it. Hoyt suggested the Poisson method, and I have used it for consistency. I have also asked him why Poisson is the best tool here, but he hasn't answered that yet.

I do not see why you think using the same math means I agree to use his counting numbers, unless you were under the impression that we were counting different transcripts. The whole dispute revolves around the fact that the same math process yields different results from the different counts from the same underlying transcripts.

 

[BillHoyt]

Moi, I thought the Poisson calculator was a bit fishy.

Yeah, but a good shad was timed by all.

 

[CFLarsen]

Thanz,

Did you use the Poisson calculator on data you knew were flawed, yes or no?

 

[Clancie]

Thanz,

You've fought the good fight here, but it will never get results beyond the fact that you have effectively made your point in several different ways, all of which are being ignored by Bill and Claus.

Claus obviously doesn't understand statistics -or- the flawed premise of Bill's counting method. He's going to just keep trying to badger you about something he can't do himself (unless he gets a PM about it, maybe).

And Bill, imo, actually knows he's wrong but will never admit it.

 

[Thanz]

I have defined the statistic to be name guesses, and include either bare initials or full names. I apply that method to every name guess, and group the results into "J"s and "non-J"s. Each guess is a random event. There is no doubling, trebling or any other such factor applied to the numerator and denominator, sir. All randomly-uttered letters are counted equally and consistently.

I guess this means you won't address my actual simple questions. We have been around and around this many times, but you still refuse to answer the most basic questions. I ask that you please directly address the points I have made to you in the bulleted list on p. 16 and in my expanded bird example above.

 

[Thanz]

Thanz,

Did you use the Poisson calculator on data you knew were flawed, yes or no?

No. I used it on data that I collected without what I see as Hoyt's flaws. I then compared the result to Hoyt's, who used the same calculator. Hoyt applied the calculator to his flawed data.

 

[CFLarsen]

Thanz,

You've fought the good fight here, but it will never get results beyond the fact that you have effectively made your point in several different ways, all of which are being ignored by Bill and Claus.

Absolutely not. You confuse "ignoring" with "rebuttal".

Claus obviously doesn't understand statistics -or- the flawed premise of Bill's counting method. He's going to just keep trying to badger you about something he can't do himself (unless he gets a PM about it, maybe).

I don't claim to understand it, Clancie. But I do try to understand it. I don't, however, see any calculations from you.

No. I used it on data that I collected without what I see as Hoyt's flaws. I then compared the result to Hoyt's, who used the same calculator. Hoyt applied the calculator to his flawed data.

But what's the point of comparing Poisson(correct data) with Poisson(flawed data)? Isn't that simply a waste of time?

 

[BillHoyt]

I have also asked him why Poisson is the best tool here, but he hasn't answered that yet.

You're batting 1.000 here, Thanz. Again, and again and again, answered.

I don't like to repeat myself, Thanz. Neither do I enjoy false accusations of lies and deflections. Save that crap for woo fora. Here, stick to facts, and get them straight.

 

[Thanz]

But what's the point of comparing Poisson(correct data) with Poisson(flawed data)? Isn't that simply a waste of time?

In general, I think I would agree with you - we should only look at the Poisson(correct data). The trouble is that Hoyt claims that he has the correct data, and so do I.

Hoyt did an analysis that claimed a statistically significant result. I did my own count, and found a statistically non-significant result. If my count were also significant, I doubt there would have been as much of an argument. But Hoyt is claiming that the analysis of the transcripts shows something that, in my opinion, it does not. In order to make that assertion I need to show my count and the analysis that comes to a different conclusion.

 

[Thanz]

You're batting 1.000 here, Thanz. Again, and again and again, answered.

I don't like to repeat myself, Thanz. Neither do I enjoy false accusations of lies and deflections. Save that crap for woo fora. Here, stick to facts, and get them straight.

Read back over my posts in this thread. I stated on page 16:

You only posted a general description of Poisson. you never connected that process to JE or explained why JE should be considered a poisson process.

You have now pointed me to that general post. Whoopee. Care to connect it to JE? That is all I am asking. I asked in the other thread too, but you never did answer me. Are you going to explain it or not?

Are you going to answer the other points and questions I have asked in this thread?

 

[CFLarsen]

In general, I think I would agree with you - we should only look at the Poisson(correct data). The trouble is that Hoyt claims that he has the correct data, and so do I.

Hoyt did an analysis that claimed a statistically significant result. I did my own count, and found a statistically non-significant result. If my count were also significant, I doubt there would have been as much of an argument. But Hoyt is claiming that the analysis of the transcripts shows something that, in my opinion, it does not. In order to make that assertion I need to show my count and the analysis that comes to a different conclusion.

Yes, you don't need to reiterate what you did. You need to explain why you did it. It was based on flawed data, right? (Well, in your opinion)

What could such a comparison possibly prove?

 

[BillHoyt]

In general, I think I would agree with you - we should only look at the Poisson(correct data). The trouble is that Hoyt claims that he has the correct data, and so do I.
.

No, Thanz. The trouble is you have both a different approach to the counting and an analysis that you neither understand nor believe it. Your broad-spectrum assault sir has trapped you in a contradiction. You simultaneously claim you have a vaild analysis that doesn't show significance and claim you don't think the data can be modeled Poisson.

It is one way or the other, if you agree Poisson is valid, you can claim you valid results. If you don't, you can't.

The lady or the tiger, sir.

 

[Thanz]

No, Thanz. The trouble is you have both a different approach to the counting and an analysis that you neither understand nor believe it. Your broad-spectrum assault sir has trapped you in a contradiction. You simultaneously claim you have a vaild analysis that doesn't show significance and claim you don't think the data can be modeled Poisson.

It is one way or the other, if you agree Poisson is valid, you can claim you valid results. If you don't, you can't.

The lady or the tiger, sir.

Huh? I haven't claimed that Poisson is not valid. I think that an analysis of one letter is not much help, but that is a different issue and I don't think that we want to get into that.

No, on the contrary, for the purposes of our discussion I have been assuming that Poisson was a valid tool. I have asked you why it is an appropriate tool - as I seem to recall others questioning it - but I really just wanted to understand why it is the appropriate tool if that is indeed the case.

In the original thread, I quoted your general description and said this:

Mr Hoyt -

I am still fuzzy as to why Poisson is appropriate for the analysis we are doing here. Could you perhaps specifically apply your general description of Poisson (quoted above) to the specific analysis we are doing here? What about the JE guesses makes you think it is a "Poisson Process"?

Much appreciated.

It was a request for knowledge, and you chose not to explain it.

None of which is really relevant to the issue of whether your count makes logical sense.