Re: The EXACT difference of two proportions

Everybody knows this DEFINITION (quoting from the
web)

- A test statistic is a quantity CALCULATED FROM THE
SAMPLE(s). Its value I use to decide whether or not
the null hypotheses should be rejected in the
hypotheses test. The choice of a test statistic will
depend on the assumed probability model and the
hypotheses under question.

The Readers should compare with Reef Fish´s comment
(quoting).

*** But you DO NOT have a test statistic, and your p
is NOT the
p-value. You made up a probability of an statement
involving
two UNKNOWN but FIXED parameter values p1 and p2.
Pr( p1 <= p2 ) is either 1 or 0!***

I ask: WHO IS THE ABSOLUTE IGNORANT IN THIS MATTER?

________licas (Luis A. Afonso)



Luis, your assumed probability model is your test statistic. You can't do that. The model, as well as the hypotheses, have to be independent of the test statsitic.

Jack
.