Hello,
I've been making a decent amount of use of the lme function
recently, and I am aware that I can extract the variance and correlation
components with the VarCorr(model) function. However, I am interested in
testing these components as well...is there a module or function available
for testing these?
I understand there's some debate as to which test is best for a
given situation (e.g., Wald vs. likelihood ratio), as well as some good
arguments that it is perhaps ill-advised to test (variance is something to
be predicted; if there is literally no variance, then the estimate perfectly
predicts the outcome and the scientific question is basically answered; no
statistics necessary), but there are some circumstances in which a
nonsignificant variance component might be useful: For example, when
deciding (in an HLM) whether it is necessary or at all useful to include a
grouping factor.
Thanks in advance.
Adam D. I. Kramer
Ph.D. Candidate, Social and Personality Psychology
University of Oregon
