Even in the case of simple tests for differences among three
univariate means, one needs more information before such a
question can be answered. One has to know the amount of
variability within samples and how different the means are
expected to be. In addition one needs to know something about the
distribution and the desired amount of statistical power in
making such tests.

There is a mathematical lower limit if one is going to use
methods such as a multivariate analysis of variance or a
multivariate T-squared test. One needs to have the degrees of
freedom of the pooled within-sample covariance matrix equal to or
larger than the number of shape variables (2p-4 for 2D shape
data). This constraint can be relaxed if one uses a permutation
test instead.

----------------------------------------------
F. James Rohlf, Distinguished Professor
Dept. Ecology & Evolution, Stony Brook University
Stony Brook, NY 11794-5245
http://life.bio.sunysb.edu/ee/rohlf




> -----Original Message-----
> From: morphmet [mailto:[EMAIL PROTECTED] 
> Sent: Friday, October 27, 2006 3:52 PM
> To: [email protected]
> Subject: Minimum sample size
> 
> Hi List,
> 
>      I've got an elementary question. I need to know how to 
> estimate the minimum sample size neccesary for exploring 
> morphometric differences (if any) among three geographically 
> separated groups of a fish using multivariate analysis. At 
> this moment I haven't got any preliminary sampling, so I 
> don´t know nothing about the variability of the measurements.
>      If somebody could give me a hand with this problem, I'll 
> be really thankful.
> 
> Regards
> 
> Rodrigo
> 
> 
> **************************************************************
> **********
> Rodrigo Veas
> Departamento de Oceanografía
> Casilla 160-C, Cabina 10
> Universidad de Concepción
> Concepción
> CHILE
> 
> [EMAIL PROTECTED]
> fono: +56 41 2207386 - 2203532
> fax:  +56 41 2256571
> 
> 
> **************************************************************
> ***********
> 
> 
> 
> --
> Replies will be sent to the list.
> For more information visit http://www.morphometrics.org
> 




-- 
Replies will be sent to the list.
For more information visit http://www.morphometrics.org

Reply via email to