Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Re: fpower


From   Philip Ender <ender97@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: Re: fpower
Date   Sat, 20 Jun 2009 20:55:42 -0700

"CHIDLOW Agnieszka" <A.Chidlow@staffs.ac.uk> wrote:

Please forgive me if you find this request silly (or basic), but I
have been having some nightmares to figure out how to use ex post
"fpower" for anova.
I have read about the power analysis  (e.g. papers, and on
www.ats.ucla.edu), but it looks like I can't do the test because I
don't seem to know how to calculate delta and/or means for my groups.
Hence would really appreciate some help from you.

My anova results are:
(...)
---------------------------------------------

 "Martin Weiss" <martin.weiss1@gmx.de>  replied:

Seems from the explanations at
http://www.ats.ucla.edu/stat/stata/dae/fpower.htm that these values
are mere educated guesses, not results of calculations...

----------------------------------------------

There certainly are educated guesses involved in any power analysis.
In the case of -fpower- the main educated guess is concerning the
delta().  After that -fpower- computes power based on the noncentral
F-distribution.  So there are indeed calculations going on.  The
-simpower- program generates random samples with given means and
standard deviations and computes power by counting the number of
significant results.  I usually suggest at least 1,000 iterations.

As for Agnieszka power question, I would recommend running the
following: "simpower nm year, reps(2000)"

I'm a little confused by the statement that nm is 2 groups.  Does this
mean that there is a categorical response variable?  If so, then
-anova- is the wrong tool to
use.


-- 
Phil Ender
Statistical Consulting Group
UCLA  Academic Technology Services
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index