[R-sig-ME] lme4 vs nlme

[Nicholas Burgoyne]

Dear all,

I have been using lme4 (1.1-7) extensively for a project, and have become stumped when trying to perform an F-test anova on the output, I want a to jointly set a series of terms as zero (I am using pbkrtest to generate DF values, which I believe are good enough for the test). I see that nlme has this functionality (specifying Terms in the anova function), but I am too far gone down the lme4 route for that.

The lme4 value I think is not generated (I'm not going to say missing, I'm sure there is a good reason for it) is the varFixFac matrix (an attribute of the fixDF in nlme). With this set the anova required should be simple. Is there anyway of defining this matrix using the values generated in the lme4 output?

Any input would be greatly appreciated, many thanks, 

Nicholas Burgoyne
  
Tel. +44 (0)1249 705 450 | Mobile. 
mailto:nburgoyne at mango-solutions.com | www.mango-solutions.com



2 Methuen Park
Chippenham
Wiltshire 
SN14 OGB
UK


--

LEGAL NOTICE\ \ This message is intended for the use of ...{{dropped:18}}