[R-sig-eco] Repeated measures in multi variate analyses

[Kingsford Jones]

Hi Allan,

To work with repeated measures (longitudinal) data using mixed models
software you need to transform your thinking from 'wide' to 'long'
data format.  Instead of having a single row per individual and a
column for each response value in each time period, the data is
organized such that there is one row per time within each individual,
and a single column for the response variable.  The analysis is no
longer multivariate with as many responses as time periods, but has a
single response containing all response values within all times, and
the grouping and time structure are described on the right-hand-side
of the model.

So, the answer is no, lme does not accept multiple response variables
in a model formula, but yes it can handle repeated measures analysis.
Furthermore it provides likelihood-based estimation of model
parameters, which not only has nice mathematical properties but also
allows estimation temporal correlation structures within individuals
(via the correlation argument in lme), and variance heterogeneity
structures (e.g, between groups or related to a covariate).

hth,

Kingsford Jones




On Mon, Feb 8, 2010 at 12:33 PM, Allan Edelsparre <aedelspa at uoguelph.ca> wrote:
> Hi all,
>
> I am in a situation where I want to ask what is the effect of ecomorph, diet, and family(nested in ecomorph) on five behaviours that are repeated measures.  Because I have repeated measures I was wondering if the mixed effects model command (lme) can handle a MANOVA, or can the MANOVA command handle mixed effects?
>
> Allan
>
> _______________________________________________
> R-sig-ecology mailing list
> R-sig-ecology at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>