Date:         Wed, 12 Jan 2000 13:53:24 -0500
Reply-To:     Bob Muenchen <muenchen@UTK.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Bob Muenchen <muenchen@UTK.EDU>
Organization: University of Tennessee, Knoxville
Subject:      Re: Calculating Power "After the Fact"
Content-Type: text/plain; charset=US-ASCII

"Robert C. Saunders" <robert.c.saunders@VANDERBILT.EDU> wrote in message
news:4.2.0.58.20000111181633.00b27570@s.mail.vanderbilt.edu...
> Hi,
>
> Kind of a general stats/SAS question:
>
> Why doesn't the output of a PROC REG (or other regression or hypothesis
> testing procedure) calculate the power of the tests that are
> calculated? ...

That's a good question. JMP offers such calculations, so I suppose they just
haven't gotten around to adding it to SAS. They've only recently provided a
smattering of a-priori power calculations in the Analyst application. I'm
particularly fond of JMP's "least significant number", which is the number
of observations that would produce a specified p-value if the data were to
have the same structure and estimates as the current sample. It's a good
tool to get across the point that given a large enough sample size any
finding is "significant" even though it may have no practical value.

===================================================
  Bob Muenchen (pronounced Min'-chen), Manager
  Statistical & Computational Consulting Center
  U of TN Division of Information Infrastructure
  Voice: (865) 974-5230
  FAX:   (865) 974-4810
  Email: muenchen@utk.edu
  Web:   http://dii.utk.edu/sccc
===================================================