Test with random effects model

[MatM]

Hello everybody, I have yet another question,

When testing for effects with simple aster models, one use Likelihood ratio tests, as described in many TR and in the slides. What about random effect aster models? From the TR692 and from the deck of slides on random effect, I got the impression that one uses the p-values provided by the summary and does not perform LRT. Why is that so? Is the LRT non valid in that case?

Thanks in advance !

[geyer]

This one is easy, so I am answering it first.  The anova.asterOrReaster function (which takes as arguments either results of calls to the aster function or to the reaster function) was not there originally, so the tech reports about reaster don't use it.  But it is there now.  And does (approximate) likelihood ratio tests.