The danger of classical hypothesis and significance tests [was Re: MADLY AMUSED]
- From: "S. F. Thomas" <thomas7243@xxxxxxxxxxxxx>
- Date: Wed, 28 May 2008 14:52:58 -0400
illywhacker wrote:
On May 28, 7:24 pm, "Luis A. Afonso" <lic...@xxxxxxxxxxx> wrote:^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^__1__Two pairs teams: by one side Jack Tomsky – John Smith argue that there’s no way to test Hypotheses without data (in particular using Monte Carlo), by the other side illywhacker and me (QUOTE: He is correct that one can use sampling in the way that he suggests to perform tests, Posted: May 27, 2008 9:29 AM), along with a large sample of statisticians that it could be very well.
__2__It´s generally known that the Null Hypotheses cannot be proved true, Jack Tomsky - John Smith are full engaged to find out the simple problem: * prove a coin is far *.
(For us the statement * accept the null * is only permissible as a shorthand for to the correct * it wasn’t found (from the available data) sufficient evidence to reject H0 *).
I should add that in any case this conversation is moot. Confidence
intervals and classical hypothesis and significance tests are an ad
hoc waste of time, and are even dangerous, because they can appear to
work even when they have in fact broken down completely.^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Pardon me for jumping in. Would you care to give an example of this claim? I'm by no means a defender of the classical paradigm, but I'm not aware of the positive danger you claim. Please elaborate.
illywhacker;
Regards,
S. F. Thomas
.
- Follow-Ups:
- References:
- MADLY AMUSED
- From: Luis A. Afonso
- Re: MADLY AMUSED
- From: illywhacker
- MADLY AMUSED
- Prev by Date: Re: MADLY AMUSED
- Next by Date: Re: MLE
- Previous by thread: Re: MADLY AMUSED
- Next by thread: Re: The danger of classical hypothesis and significance tests [was Re: MADLY AMUSED]
- Index(es):
Relevant Pages
|
