[R-sig-ME] repeated measures and random effects question

[Thierry Onkelinx]

Dear James,

Your models is too complex given the available data. The random part can be
expanded to (1 + year|site) + (1 + year|site:plot) + (1 +
year|site:plot:subplot). The number of data points for each level of the
random effect groups are respectively 48, 24 and 3. I'm not comfortable
fitting a linear trend through only 3 data points... Also consider
http://bbolker.github.io/mixedmodels-misc/glmmFAQ.html#should-i-treat-factor-xxx-as-fixed-or-random,
especially the last paragraph.

I would strongly recommend that you consult a local statistician.

Best regards,



ir. Thierry Onkelinx
Statisticus / Statistician

Vlaamse Overheid / Government of Flanders
INSTITUUT VOOR NATUUR- EN BOSONDERZOEK / RESEARCH INSTITUTE FOR NATURE AND
FOREST
Team Biometrie & Kwaliteitszorg / Team Biometrics & Quality Assurance
thierry.onkelinx op inbo.be
Kliniekstraat 25, B-1070 Brussel
www.inbo.be

///////////////////////////////////////////////////////////////////////////////////////////
To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to say
what the experiment died of. ~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data. ~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey
///////////////////////////////////////////////////////////////////////////////////////////

[image: Van 14 tot en met 19 december 2017 verhuizen we uit onze vestiging
in Brussel naar het Herman Teirlinckgebouw op de site Thurn & Taxis. Vanaf
dan ben je welkom op het nieuwe adres: Havenlaan 88 bus 73, 1000 Brussel.]
<https://overheid.vlaanderen.be/mobiliteitsplan-herman-teirlinckgebouw>
Van 14 tot en met 19 december 2017 verhuizen we uit onze vestiging in
Brussel naar het Herman Teirlinckgebouw op de site Thurn & Taxis.
Vanaf dan ben je welkom op het nieuwe adres: Havenlaan 88 bus 73, 1000
Brussel.

///////////////////////////////////////////////////////////////////////////////////////////
<https://www.inbo.be>

2017-12-08 0:34 GMT+01:00 Moore, James Roy <jmoore25 op wsu.edu>:

> Hello list,
>
> Not sure how to address a list, so let me know if I'm doing something
> wrong.
>
> I'm having a difficult time wrapping my mind around how to account for
> repeated measures and random effects and was directed here.
>
> My experimental design is:
> 4 sites
> 2 plots per site (control and treatment, 10m x 10m)
> 8 subplots per plot (1m x 1m)
> 3 years (each subplot sampled once per year every year)
>
> Response variable: proportional plant cover
> Potential covariates/predictors: other proportional plant cover, herbivore
> damage
> The treatment is herbivore removal so instead of using the factor
> treatment in the model I'm using proportional herbivore damage.
>
> I have to use a package specific for beta distribution with 0's but we can
> talk about this in the context of a glmm using lme4 since that is the
> syntax I'm most familiar with.  My limited understanding leads me to:
> Plant cover ~ herbivore damage + some covariates +(1|site)+(year|site/plot/
> subplot)
>         Random variation between sites and, to account for repeated
> measures, subplots nested within plot nested within site sampled at a
> yearly interval for   3 years.
>
> As an additional curiosity, is there any problem with excluding the
> treatment factor and using proportional herbivore damage instead? Also,
> since site is included in the second random term, is the first term still
> necessary.  There is definitely variation between sites.
>
> Thank you for any assistance you're able to provide,
> James
>
> _______________________________________________
> R-sig-mixed-models op r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>

	[[alternative HTML version deleted]]