[R-sig-ME] Mixed models with loglog distibution

[Alan Haynes]

Dear list members,

I am currently trying to analyse some data from an decomposition experiment
where we exclude herbivores and consider the effect of that on
decomposition rates.

Unfortunately the data produces residuals which are not normally
distributed and appear to correspond with a loglog distribution.
I have attached some graphical diagnostic plots of the residuals for your
perusal (distribution plots of the original data are very similar).

This leads me to my question....Is there some way of running a mixed model
with a loglog distribution?
I have found that the 'reliaR' package has the usual d*foo*, r*foo* etc
functions but specifying family = Loglog doesnt seem to work

Any suggestions?


Many thanks in advance,

Alan Haynes







--------------------------------------------------
Email: aghaynes at gmail.com
Mobile: +41794385586
Skype: aghaynes
-------------- next part --------------
A non-text attachment was scrubbed...
Name: model_residuals.pdf
Type: application/pdf
Size: 82132 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mixed-models/attachments/20111206/3552a687/attachment-0003.pdf>