Re: JACK TOMSKY: an HISTORIAL of nonsense

Richard Atkins wrote


*** Surely this all boils down to the same argument Fisher and Pearson were having nearly a century ago? Luis is considering hypothesis tests as providing a quantitative weight of evidence which can never establish the truth of either the null or the alternative hypothesis, while Jack is using hypothesis tests to make a binary decision. Assuming this is true then I suppose the one remaining question is whether Jack did a power analysis before deciding that N=1 was an adequate sample size? ***

My response

Thank you, Richard by the concern.
In fact, I have not sufficiently documentation, nor time, to study the fight between R.A. Fisher and Pearson. You can easily get that DATA PROCESSING in Research Practice one have no time (and no opportunity) to spend to appreciate such *delicacies*.
I think, (naively?) that the simulation procedure to find Confidence Intervals (the point under discussion) is not against NP theory, or at least against logic, in its broad sense.
However I am not sure. I am not quite concerned.
My job was simply to construct a table of values that will be (perhaps) of some usefulness for those believing in the method I used. Otherwise…, no one is forbidden to sweep them aside.
Concerning H0 and the *equality*, I always believed that it was a shortcut of the more precise question: are the data showing sufficient evidence that we should keep the hypotheses that *the status of the world* not *markedly* changed? Or on contrary, the observed difference is so big that *cannot* be owing to chance?
As a simple minded person I do not really care what the Testing Hypotheses philosophy is.


_____licas (Luis A. Afonso)
.

Relevant Pages
Quantcast