Skip to content
Prev 307628 / 398506 Next

glmmPQL and spatial correlation

Luis Huckstadt <lahuckst <at> ucsc.edu> writes:
That's a bit surprising.  When you say the significance of one variable
changes -- does that mean it changes from (e.g.) 0.045 to 0.055, which
crosses the magic line at p=0.05 but is not otherwise meaningful?
This is not going to be easy in any case.  I would look into the
INLA package and AD Model Builder ... (I would also be mildly nervous
about using PQL for binary data -- it might be OK, but this is known
to be more or less the worst case scenario for PQL ...)

  Ben Bolker