Skip to content
Prev 5666 / 20628 Next

Getting out of using a mixed model if "nested" data points are not actually correlated?

On 03/22/2011 08:50 AM, M.S.Muller wrote:
Three possibilities: (1) use glmmADMB with zeroinfl=TRUE (at the
moment it will only handle a single random effect, but it seems that
your problem doesn't require more); (2) use  MCMCglmm (see the course
notes for an example; (3) as you suggest, fit the model and then show a
lack of statistically significant correlations in the residuals.

  I would probably try #1 first.
Message: 1