[R] "multivariate" mixed models with crossed random effects

Tom Van Dooren vdooren at rulsfb.leidenuniv.nl
Thu Oct 10 14:03:18 CEST 2002


I am currently analysis a dataset of chaffinch song durations.
Per observed song, the durations of two parts (segments) of the song are

One can assume that a random effect of bird and of song type contribute
to duration variation.
With songtype nested within bird, lme does well in estimating random
effects of bird and songtype on durations of both parts of a song, and
of the correlation between bird random effects. I coded this model as


So the mean durations are estimated per song part, I get random effects
of bird and songtype, plus the correlation between random effects on
both parts, and error variances are estimated separately per song part.

Now it is possible to classify songs according to sonogram properties,
such that songtype according to that classification is not nested
anymore within bird (so different birds can sing the same song).
When the songtype and bird random effects are crossed, I modelled
duration of each separate  part of a song as:


In the case of crossed random effects,does anyone know whether it is
possible to have
durations of both song parts in a single model, so that I can estimate
the correlation between random effects of bird or songtype on both
durations, just as the nested model does?

Best regards, much thanks for your help,

Tom Van Dooren
Section Theoretical Evolutionary Biology
Leiden University
The Netherlands

r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list