[R-sig-eco] Power analysis on mixed model

[Simon Blomberg]

You can do it. It is probably easiest to do it by simulation. Simulate
data at several effect sizes (ie when H0 is false) and sample sizes and
count the proportion of times that the model correctly rejects H0. The
big problem is dealing with the random effects. You could treat the
random effects as fixed at the values in your initial analysis, in which
case your power analysis will only be relevant to the original
population, or you can simulate the random effects by drawing them from
a candidate distribution, for more general applicability.

But if this is a post-hoc power analysis, you shouldn't do it at all.
See: 

HOENIG, J. M. & HEISEY, D. M. The Abuse of Power: The Pervasive Fallacy
of Power Calculations for Data Analysis. The American Statistician,
2001, 55, 19-24 

Cheers,

Simon.

On Thu, 2008-12-18 at 10:23 +1100, Caitlin Vertigan wrote:
> Hi all,  I've just completed an analysis of short-tailed shearwater breeding
> success in relation to human disturbance using a mixed model.  I'd like to
> do a power analysis on this now but am unaware of any way to do it.   Is it
> possible? Does anyone have any ideas?
> Thanks,
> Caitlin
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-ecology mailing list
> R-sig-ecology at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
-- 
Simon Blomberg, BSc (Hons), PhD, MAppStat. 
Lecturer and Consultant Statistician 
Faculty of Biological and Chemical Sciences 
The University of Queensland 
St. Lucia Queensland 4072 
Australia
Room 320 Goddard Building (8)
T: +61 7 3365 2506
http://www.uq.edu.au/~uqsblomb
email: S.Blomberg1_at_uq.edu.au

Policies:
1.  I will NOT analyse your data for you.
2.  Your deadline is your problem.

The combination of some data and an aching desire for 
an answer does not ensure that a reasonable answer can 
be extracted from a given body of data. - John Tukey.