Cold Reading Demos at TAM2

Jeff Corey said:

One problem is that you have cited the one-tailed probability for Fisher's and the two-tailed probabilities for both chi-square tests. The two-tailed p for Fisher's is .027.
Back to Stat 101 for you.
Also, for applying the Yate's correction with those cell counts.
 
I gave him credit for that because the experts disagree about Yate's in the case of the 2 X 2 contingency table. It's interesting, though, that the average of the ps of the corrected and uncorrected chi-squares is only a bit higher than the p for Fisher's.
 
Jeff Corey said:
I gave him credit for that because the experts disagree about Yate's in the case of the 2 X 2 contingency table. It's interesting, though, that the average of the ps of the corrected and uncorrected chi-squares is only a bit higher than the p for Fisher's.

Sorry, I may not have been clear. I meant that Yate's correction is applied with 2 x 2 s with one or more cells having a count of 5 or less. He's got 25s and 58s as his hypothetical counts.
 
CFLarsen said:

But the statistician comes in after the experiment is done.


As I've already said, statisticians are often used at many stages of an experiment.


I don't need to say it many times, it is true. Your own posts show this.


There is a difference: you believe my own posts show this.


Then pick those specifics that show that a statistician can unconsciously choose the method that will give better results for one group.


The examples of unconscious bias are the same as in any other study. If his is pro-coffee, and the study shows that coffee has negative health effects, he might unconsciously be less critical, for example. He might choose a method (see me and Corey's discussion on Fisher's Exact test vs. Chi-square and Chi-square with Yates correction) that gives a smaller or larger p-value, etc.


Nope. You have not been able to explain how "tripleblind" works.


Wrong. I mentioned exactly what a tripleblind is, and gave an example of how group membership is hidden from the evaluator. After the group membership is hidden, the analysis continues in exactly the same manner as in non-tripleblinded experiments, as I've already said.


You pointed to a supernatural ability to predict the method that will give biased results, that's all.


Uh.... yeah.... I mentioned a "supernatural ability"? You sure about that pops?


In other words, woowoo.

Please find this "supernatural ability" that I mentioned, and post my exact quote regarding it.
 
BillHoyt said:

Well, that's the funny thing, there, Tr'oll. I was addressing the issues when suddenly you ejaculated this outburst:


LOL, you've addressed absolutely nothing on the statistical and mathematical matters at hand.


Do you now wish to withdraw the claim of "consulting people in experimental design"? If not, then please cite the papers where your consultancy was acknowledged by the principal investigators.

I mentioned that because before that, as you conveniently leave out, you made the personal attack suggestion I take some experimental design classes. I will neither retract that I said I've taken experimental design classes, nor will I retract that I've said I consulted as you stupidly wish.
 
I guess I can conclude that there's nothing more anyone has to say about the cold reading demos at TAM. :p

And as for this thread itself, has anyone else lost track of what Bill and T'ai are arguing about? Is it still whether or not you can -really- do a triple blind test?

Or are we back to the usual, "I know more than you do...No you don't...Yes I do....You don't know anything...Yes I do...No you don't...." exchanges that, Bill, I must admit, you are just so good at. Seriously, how many pages can you keep it up? And...is this a new record?

Why do these arguments about methodology never lead to any substantive results (like the previous "J" analysis that never quite got finished)?

Whatever. Sorry to interrupt, whatever this is about. Carry on.... :)
 
T'ai Chi said:
As I've already said, statisticians are often used at many stages of an experiment.

But not in your own example. Can you please try to keep focused here??

T'ai Chi said:
There is a difference: you believe my own posts show this.

No, I leave it to others to decide.

T'ai Chi said:
The examples of unconscious bias are the same as in any other study. If his is pro-coffee, and the study shows that coffee has negative health effects, he might unconsciously be less critical, for example. He might choose a method (see me and Corey's discussion on Fisher's Exact test vs. Chi-square and Chi-square with Yates correction) that gives a smaller or larger p-value, etc.

But how do these methods give a favorable result, if the statistician is choosing them unconsciously? That is what you are asked, and that is what you keep dodging.

T'ai Chi said:
Wrong. I mentioned exactly what a tripleblind is, and gave an example of how group membership is hidden from the evaluator. After the group membership is hidden, the analysis continues in exactly the same manner as in non-tripleblinded experiments, as I've already said.

Bullsh1t. You have not been able to explain what "tripleblind" is, except by pointing to some supernatural ability.

T'ai Chi said:
Uh.... yeah.... I mentioned a "supernatural ability"? You sure about that pops?

How else would you explain how a statistician can unconsciously choose the method that shows his own biases?

T'ai Chi said:
Please find this "supernatural ability" that I mentioned, and post my exact quote regarding it.

Oh, stop this silly game: You are pointing to a supernatural ability, which you have been unable to explain how works.

Clancie has tried this before as well, and so has Steve Grenard. It won't work here.

Just explain how it works, T'ai Chi:

How does a statistician choose - unconsciously - the method that shows his own biases?

It is painfully clear that you are hopping from one foot to another, in order to avoid this question.

(You may try to report me to the mods for "bullying". It might work. Then again, it might not. Wanna take your chances?)
 
Clancie said:
I guess I can conclude that there's nothing more anyone has to say about the cold reading demos at TAM. :p

You may spare that condescending tongue of yours. You were not there, you have no first-hand knowledge of what went on there, and you had no intention of getting it. You chose a backseat, and that's where you are most comfortable.

Clancie said:
(As for this thread itself, has anyone else lost track of what Bill and T'ai are arguing about? Is it still whether or not you can -really- do a triple blind test? Or are we back to the usual, "I know more than you do...No you don't...Yes I do....You don't know anything...Yes I do...No you don't...." exchanges that, Bill, I must admit, you are just so good at. Seriously, how many pages can you keep it up? And...is this a new record?)

If you could read, you would not need to ask that question. It is about T'ai Chi and his claims of statisticians having supernatural powers.

Clancie said:
Whatever. Sorry to interrupt, whatever this is about. Carry on.... :)

You are most welcome. When do you have something substantial to offer this forum?
 
Jeff Corey said:

One problem is that you have cited the one-tailed probability for Fisher's and the two-tailed probabilities for both chi-square tests. The two-tailed p for Fisher's is .027.
Back to Stat 101 for you.

:rolleyes: That's not a stat. knowledge error, that is a looking at the wrong row for output error. Oops.

Let's change things slightly, and my original point still stands:

For example, consider the table of hypothectical data:

58, 25
23, 24

and the hypothetical alpha level is .02.

Using your example of Fisher's Exact and chi-square test:

From Fisher's Exact (2 tails this time!), we get a p-value of .0237. From the chi-square test we get a p-value of .0179. Only in the chi-square case do we technically have statistical significance however.

You like?
 
T'ai Chi said:


:rolleyes: That's not a stat. knowledge error, that is a looking at the wrong row for output error. Oops.

Let's change things slightly, and my original point still stands:

For example, consider the table of hypothectical data:

58, 25
23, 24

and the hypothetical alpha level is .02.

Using your example of Fisher's Exact and chi-square test:

From Fisher's Exact (2 tails this time!), we get a p-value of .0237. From the chi-square test we get a p-value of .0179. With a Yates correction on that chi-square test we get an even higher p-value of .0293. Only in the chi-square case do we technically have statistical significance however.

You like?

It's not a popularity test.

You are getting slaughtered here, again and again. You keep changing the stakes (depending on what you have recently dug up on Google), and expect to be taken seriously.
 
CFLarsen said:

But not in your own example. Can you please try to keep focused here??


As I already mentioned many times, my example was to show how group membership can be blinded in regards to the evaluator. You will not be successful in trying to make it so I cannot refer to what statisticians do in real life.


No, I leave it to others to decide.


"Others" like Bill? Get real.


But how do these methods give a favorable result, if the statistician is choosing them unconsciously? That is what you are asked, and that is what you keep dodging.


If you do not know what a p-value is, you won't understand. To those who do know what p-values are, it was pretty clear.


Bullsh1t.


Now now...


You have not been able to explain what "tripleblind" is, except by pointing to some supernatural ability.


Entirely bogus.

I directly said what a triple blind experiment is and gave an example of how group membership is blinded. This "supernatural ability" you refer to is your own invention, and if you are referring to unconscious bias as "supernatural" you should be ashamed as there is nothing supernatural about it.


How else would you explain how a statistician can unconsciously choose the method that shows his own biases?


Uh, nothing "supernatural" with bias. I'm surprised you think there is.


Oh, stop this silly game: You are pointing to a supernatural ability, which you have been unable to explain how works.


You fail to intimidate me with the tactic of attributing something to me then arguing against me with what you invented. I only referred to conscious or unconscious bias (hence the need for blinding). You are pathetically attempting to make it appear as I claimed this is a "supernatural ability". It simply won't work Claus.


Clancie has tried this before as well, and so has Steve Grenard. It won't work here.


When you can't argue points Claus, try guilt by association!


Just explain how it works, T'ai Chi:


Already done. If you don't understand that isn't my problem.


how does a statistician choose - unconsciously - the method that shows his own biases?


Already done.


It is painfully clear that you are hopping from one foot to another, in order to avoid this question.


Nope. In fact I've already answered it.


(You may try to report me to the mods for "bullying". It might work. Then again, it might not. Wanna take your chances?)

Bullying isn't against the forum rules.
 
CFLarsen said:

You may spare that condescending tongue of yours.


:rolleyes:


If you could read, you would not need to ask that question. It is about T'ai Chi and his claims of statisticians having supernatural powers.


You are being highly dishonest here, as:

I did not claim that statisticians have "supernatural powers" (or any powers at all).

If I did, please paste my EXACT quote Claus. Surely it is easy for you.

I simply said that evaluators can suffer, as anyone can, from conscious or unconscious bias. If you are claiming that bias is supernatural, please quit wasting this threads' time.
 
CFLarsen said:

It's not a popularity test.


Address the mathematics and statistics. Address the updated Fisher's Exact vs. Chi square p-values. Thanks.


You are getting slaughtered here, again and again.


If I'm "getting slaughtered", then I should apply for the million, as I seemed to have been totally unharmed.

Again, address the mathematics and statistics.


You keep changing the stakes (depending on what you have recently dug up on Google), and expect to be taken seriously.

Address the mathematics and statistics.

And in anticipation of your next round of personal attacks: Address the mathematics and statistics.

Thanky.
 
T'ai Chi, can you describe a triple-blind experiment or not? Stop this bickering and deliver something. Take Claus' ammo away from him.
 
T'ai Chi said:
As I already mentioned many times, my example was to show how group membership can be blinded in regards to the evaluator. You will not be successful in trying to make it so I cannot refer to what statisticians do in real life.

Then I would suggest that you start addressing your own example and explain how a statistician can - unconsciously - choose the method that shows his own biases?

T'ai Chi said:
"Others" like Bill? Get real.

Whatever. Spare your ad hominems. Just explain how a statistician can - unconsciously - choose the method that shows his own biases.

T'ai Chi said:
If you do not know what a p-value is, you won't understand. To those who do know what p-values are, it was pretty clear.

I have noticed that Jeff Corey and Bill have taken you to task for your lack of statistical knowledge. A university professor and a strip club bouncer.

T'ai Chi said:
Now now...

No, let's cut the crap, shall we? Let's get down to the nitty-gritty.

T'ai Chi said:
Entirely bogus.

Entirely true.

T'ai Chi said:
I directly said what a triple blind experiment is and gave an example of how group membership is blinded. This "supernatural ability" you refer to is your own invention, and if you are referring to unconscious bias as "supernatural" you should be ashamed as there is nothing supernatural about it.

Are you calling an ability to choose - unconsciously - the method that shows a statistician his own biases for natural?

T'ai Chi said:
Uh, nothing "supernatural" with bias. I'm surprised you think there is.

It's not "me". It's simply what any rational human being would think.

T'ai Chi said:
You fail to intimidate me with the tactic of attributing something to me then arguing against me with what you invented. I only referred to conscious or unconscious bias (hence the need for blinding). You are pathetically attempting to make it appear as I claimed this is a "supernatural ability". It simply won't work Claus.

Whatever. I am still waiting for you to explain how a statistician can - unconsciously - choose the method that shows his own biases.

T'ai Chi said:
When you can't argue points Claus, try guilt by association!

No, I am merely pointing out that you use the exact same methods.

T'ai Chi said:
Already done. If you don't understand that isn't my problem.

BEEP! Yes, it is. You made the claim, that a statistician can - unconsciously - choose the method that shows his own biases. You have yet to answer.

T'ai Chi said:
Already done.

You are most welcome to show me wrong. Come on, show me - the editor of SkepticReport, Maker of Lists, wrong!

T'ai Chi said:
Nope. In fact I've already answered it.

You are most welcome to show me wrong. Come on, show me - the editor of SkepticReport, Maker of Lists, wrong!

T'ai Chi said:
Bullying isn't against the forum rules.

Then, stop complaining. And answer the question: How can a statistician - unconsciously - choose the method that shows his own biases?

T'ai Chi said:
You are being highly dishonest here, as:

I did not claim that statisticians have "supernatural powers" (or any powers at all).

Oh, yes you did!

You claimed, that a statistician could - unconsciously - choose the method that shows his own biases. That is not "supernatural"? Then, what is?

T'ai Chi said:
If I did, please paste my EXACT quote Claus. Surely it is easy for you.

Don't try this "Clancie" method. It won't work. Read your own claim, T'ai Chi!

T'ai Chi said:
I simply said that evaluators can suffer, as anyone can, from conscious or unconscious bias. If you are claiming that bias is supernatural, please quit wasting this threads' time.

But it is supernatural. You see, you have not been able to explain how a statistician is able to unconsciously choose the method that would show his own biases.

T'ai Chi said:
Address the mathematics and statistics. Thanks.

Oh, I can read. So can people.

T'ai Chi said:
Address the mathematics and statistics. Thanks.

Oh, I can read. So can people.

T'ai Chi said:
Address the mathematics and statistics. Thanks.

Oh, I can read. So can people.

T'ai Chi said:
And in anticipation of your next round of personal attacks: Address the mathematics and statistics. Thanky.

Oh, I can read. So can people.

Now, please stop this fiddling, and answer the question:

How can a statistician - unconsciously - choose the method that shows his own biases?
 
Originally posted by T'ai Chi
"That's not a stat. knowledge error, that is a looking at the wrong row for output error. Oops"

Sorry, Mr. B.S and M.S. in stat, you failed the test.

"Stat. knowledge" should include the ability to discern the right row and report the results correctly.
Also, there is a chi-square and Fisher's board that I accessed in 8 seconds through Google that gave all those p values, both one-tailed and two, in two+ minutes, after checkng the input and calculation three times.
Better check the value of .0218 for the non-Yates computation of chi-square ( two tailed). I got .0219.

Was your average grade in your stat classes near the median of the class? Or were you the standard deviation? It means no differential to me.
 
T'ai Chi said:
LOL, you've addressed absolutely nothing on the statistical and mathematical matters at hand.
Au contraire, poser bear. I have specifically and conspicously raised the most essential matters. You wish to sweep them under the rug. Switching a priori design decisions to a posteriori is a significant statistical error. Particularly if, as you claim, it can be done after an analysis in order to cook the research books. Your response to this has been simply bald assertions. When grilled on that, you ejaculated your claim to authority. That, sir, is now an issue.
I mentioned that because before that, as you conveniently leave out, you made the personal attack suggestion I take some experimental design classes. I will neither retract that I said I've taken experimental design classes, nor will I retract that I've said I consulted as you stupidly wish.
Don't retract it, then, but show us the evidence. Cite the papers wherein the principal researchers acknowledge you or thank you for your assistance. It is done all the time. In fact, it is tremendously disrespectful not to do this. If you consulted on a paper, you were acknowledged. Where can we find those papers? Ladies Home Journal?
 
TLN said:
T'ai Chi, can you describe a triple-blind experiment or not? Stop this bickering and deliver something. Take Claus' ammo away from him.

I hope it's ok if I don't hold my breath until this happens.
 
jj said:
I hope it's ok if I don't hold my breath until this happens.

Please don't. Removing dead bodies is not a favorite pastime of anyone outside the Cosa Nostra.
 
CFLarsen said:

How can a statistician - unconsciously - choose the method that shows his own biases?
Heck, I can answer that!

Let's take Mr. Hoyt as an example. Some time ago there was a thread in which the letter J was analyzed for frequency against the general population. Mr. Hoyt chose a Poisson distribution test, but never explained why it was appropriate for the analysis. But let's just take his test as appropriate.

He then embarked on a counting method of J's that defied logic and came up with a statistically significant result, using the Poisson test. When anybody else used the same raw data and went through counting guesses and letter J's, the results were not statistically significant.

So, did Mr. Hoyt consciously choose a method to skew the results in his favour? If he did, he is intellectually dishonest and a fraudster.

Did he unconsciously choose a method that would skew the results in his favour? If that is what happened, then you have a great example of bias on the statisticians part altering the results.

Get it?
 

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom