[R-sig-ME] What is the Correct glmmadmb syntax for random repeated measure and block?

[Ben Bolker]

On 13-12-10 08:59 PM, Sheryn Olson wrote:
> Hello everyone,
> 
> In this model:
> 
> strctr.all <- glmmadmb(pellets ~ season * (totl.splgs + totl.trees +
>                                        pctMidCov + lc +
>                                        cc + BAsplgs + BAtrees) +
>                                        offset(ln.days)+(1|plot/stand)+
> (1|year),
>                                        data=hv, family="nbinom")
> 
> Is (1|plot/stand) correct to mean group=stand ?

  There are two spatial grouping levels here -- it sounds like you
should use stand/plot (i.e., stand and plot within stand).
> 
> Is the repeated measurement correctly addressed with (1|year) ?
> 
> There are 3 repeated measures of summer (2010,2011,2012); 3 of winter
> (years 2011,2012,2013) but the 3 measures per season are a "nuisance"
> source of random variation.   The Season variable Is the fixed variable of
> interest.

  You should probably use Year+(1|Season:Year), i.e. Season and Year are
fixed (Year fixed because there aren't enough years to estimate
among-year-variance reliably).

Each plot is sampled multiple times, right?


> Data structure: 2,184 plot records of pellet counts - the response variable.
> In 26 forest stands were subsampled with 20 summer and 10 winter
> Plots/stand for vegetation metrics. Stands are replicates.
> 
> Thanks much,
> Sheryn
>