[R-sig-Geo] model comparisons ruf.fit

[Sarah Papworth]

Hi,
I'm modelling resource use using ruf.fit, but I'm a little confused
about the Materns logLik / LS LogLik given by summary(model).

It seems that Materns logLik changes if I change the values for theta,
but stays fairly stable if I change the predictor variables. LS LogLik
stays the same with various values of theta, but changes when
predictor values do.

Presumably I use Materns logLik to find the best theta values (and
lower Materns' logLik suggests a better model??)
and LS Loglik to calculate AIC to compare the predictor variables?

Thanks in advance for any help,
Sarah
-- 
Imperial College London
http://www.iccs.org.uk/sarahpapworth.htm