[R-sig-ME] Interactions between fixed and random effects

[Ben Bolker]

  [cc'ing back to r-sig-mixed]

Christopher Eckert wrote:

> Among the populations sampled we do have an expectation of the form
> of the interaction. Populations were sampled across a latitudinal
> gradient and those from further south (these plants are on the
> California coast) should be less affected to a lesser extent by
> drought. Given that, do you think the random effects model is still
> best.

  I would add consider adding latitude in the model as a (continuous)
fixed effect; the random effects are then interpreted as
population-level deviations from the expectation based on the
latitudinal gradient.

 lmer(DryMass~Water*Latitude+(Water|Population))