Luis A. Afonso
Posts:
4,277
From:
LIsbon (Portugal)
Registered:
2/16/05
|
|
Re: Two nick-picks concerning Significance Tests
Posted:
Jun 22, 2012 8:39 AM
|
|
It couldn?t be otherwise . . .
I wonder how much it?s DEMANDED from Null Hypotheses Significance Tests (NHST) those that completely forget how strongly the results/decisions depend from the information data carries . . . Contingency is the cleft-word.
Plain Statistics users (and not sufficiently learned statisticians), regularly since Tukey intend that H0=0 (or other set value) never occurs in nature AND ARE RIGHT. The problem is that NHST were not discovered to answer that question. NOT AT ALL.
Two questions, absolutely, are on contrary, solved in rigor.
Does differ:
__1___The estimated/observed parameter u significantly from the value u0, or not?
__2___The difference observed between two values, of the same parameters is significantly different from w0, or not?
Sure the algorithms leading to the result do differ with the parameter in iest and WHAT IS KNOWN relevant from the Populations features.
See:
Minimum *secure* difference between two means
Posted: May 4, 2006 3:47 AM, (relevance 100%)
Luis A. Afonso
|
|
