Skip to content
Prev 4539 / 20628 Next

Structural zeros in lme4

On Tue, Oct 5, 2010 at 1:06 AM, Lydia Belton <lbelton at zoology.up.ac.za> wrote:
I don't know of an easy way of doing that but, as you are the second
person in as many weeks to pose such a question, it is obviously a
situation that we should accommodate.

The glm function uses a modified version of some classical linear
algebra software (Linpack, which predated Lapack) specifically to
handle rank deficient cases.  These arise most commonly from missing
cells, as above.  The way the calculations are done in lmer and glmer
we don't have the option of using that decomposition that detects rank
deficiency.  We may be able to modify the code to do a preliminary
decomposition using the modified Linpack routine and take corrective
action if it is found to be rank deficient.  I regret to say that I
won't be able to do so soon as I am swamped by my teaching obligations
at present.