[R-sig-ME] Fwd: Help with nested model in lme

[Ben Bolker]

Chris Habeck <habeck.cw at ...> writes:

> I am trying to analyze a model with lme, but I could use some clarification
> on the proper syntax for dealing with the nested nature of my experiment.
> I have two fixed factors (soil strength = 4 levels; competition = 4 levels)
> applied separately to 6 replicate pots per treatment combination (n = 96
> pots).  Within the pots I took several measurements
> on two plants per pot.  The
> ultimate goal of the experiment is to understand how the two factors affect
> plant performance. Although some measurements from the two plants
> (subsamples) can be averaged (e.g., height, number of leaves), others that
> may be important covariates in the model cannot be averaged (e.g., time
> since being galled).  So, I need to retain information on individual plants
> even though they are obviously subsamples within pot.  Any advice would be
> appreciated.
> 
> Here is a snippet of my data for context:
> 
  [snip]

> 
and here is a (most likely incorrect) example of how I'd like to model
> these data:
> lme(final.height ~ avg.PAR + weeks.galled + competition*soil.strength,
> random = ~1|pot)
> 

  Actually, this looks about right to me (someone else might chime in).
Your random effect is allowing for among-pot variation in average
response, as it should.
   My only question is why you are averaging the height, number of
leaves within plots: don't these affect the final height on a per-plant
basis?