Re: Fisher vs Neyman&Pearson
From: B. D. McCullough (bdmccullough_at_drexel.edu)
Date: 07/10/04
- Next message: Osher Doctorow: "Stat Dependence = P(X-->Y) + (x o y) - (1 + FY)"
- Previous message: Sergey Tarima: "Re: observations in different scales"
- In reply to: Mark Von Tress: "Fisher vs Neyman&Pearson"
- Messages sorted by: [ date ] [ thread ]
Date: 9 Jul 2004 20:29:58 -0700
vontressms@cs.com (Mark Von Tress) wrote in message news:<84afc88b.0407061250.2ba8d2a2@posting.google.com>...
<snip>
> Are there experiments where adjustments for multiplicity are not
> appropriate because the Fisher approach to p-values is really more
> relevant?
Yes. See section 14.5 (Fisher vs Neyman-Pearson) of
Spanos' book, _Probability Theory and Statistical Inference_,
as well as Chapter 15 which justifies (theoretically)
an approach to multiple testing that often is (mis)construed
as "data mining" (testing more than one hypothesis on the
same set of data). For simulation study that supports
the theory set forth by Spanos, see Hoover and Perez
in a recent (2 or 3 years) issue of The Econometrics Journal.
Regards,
Bruce McCullough
- Next message: Osher Doctorow: "Stat Dependence = P(X-->Y) + (x o y) - (1 + FY)"
- Previous message: Sergey Tarima: "Re: observations in different scales"
- In reply to: Mark Von Tress: "Fisher vs Neyman&Pearson"
- Messages sorted by: [ date ] [ thread ]
