[R] Repeated measures design for GAM? - corrected question...

[Simon Wood]

Diederik,

The `gamm' call looks fine provided `park' is nested in `count'. You are 
basically adding a random effect for each survey, and a random effect for 
each `park' within survey to the the smooth effects. 

best,
Simon

ps. I assume there are more than 18 observations in the full dataset --- the 
model sturcture is a bit too complicated otherwise...

On Wednesday 28 January 2009 19:44, Strubbe Diederik wrote:
> Dear Simon,
> Many thanks for pointing me to the GAMM! For clarification, Bird_abundance
> are breeding densities ( e.g. 1.25 BP/ha, 2.20 BP/ha,...) and count is just
> the actual survey(e.g. first_survey,...). The dataset looks like
> Bird_abundance	Study_area	YEAR	COUNT	X1	X2	X3	X4
> 0.15	area_1	2004	first_survey	…	…	…	…
> 1.26	area_1	2004	second_survey	…	…	…	…
> 2.47	area_1	2005	third_survey	…	…	…	…
> 0.00	area_1	2005	fourth_survey	…	…	…	…
> 0.23	area_1	2006	fifht_survey	…	…	…	…
> 2.64	area_1	2006	sixth_survey	…	…	…	…
> 4.14	area_2	2004	first_survey	…	…	…	…
> 5.00	area_2	2004	second_survey	…	…	…	…
> 6.80	area_2	2005	third_survey	…	…	…	…
> 0.15	area_2	2005	fourth_survey	…	…	…	…
> 0.25	area_2	2006	fifht_survey	…	…	…	…
> 2.36	area_2	2006	sixth_survey	…	…	…	…
> 2.59	area_3	2004	first_survey	…	…	…	…
> 6.31	area_3	2004	second_survey	…	…	…	…
> 0.15	area_3	2005	third_survey	…	…	…	…
> 2.85	area_3	2005	fourth_survey	…	…	…	…
> 2.48	area_3	2006	fifht_survey	…	…	…	…
> 1.23	area_3	2006	sixth_survey	…	…	…	…
> …	…	…	…	…	…	…	…
>
> Am I correct in assuming the following is a valid syntax for this repeated
> measures design?:
>
> model <-gamm(Bird_abundance ~ YEAR + s(X1)+ s(X2)+ s(X3)+
> s(X4),random=list(count=~1,park=~1))
>
> best wishes and thanks again,
>
> Diederik
>
>
>
>
>
> Diederik Strubbe
> Evolutionary Ecology Group
> Department of Biology, University of Antwerp
> Universiteitsplein 1
> B-2610 Antwerp, Belgium
> http://webhost.ua.ac.be/deco
> tel : 32 3 820 23 85
>
>
>
> -----Original Message-----
> From: Strubbe Diederik
> Sent: Wed 28-1-2009 18:09
> To: r-help at R-project.org
> Subject: Repeated measures design for GAM? - corrected question...
>
> Dear all,
>
> I have a question on the use of GAM with repeated measures. My dataset is
> as follows: - a number of study areas where bird abundance has been
> determined. Counts have been performed in 3 consecutive years and there
> were 2 counts per year (i.e. in total 6 counts). - a number of
> environmental predictors that do not change over year Xi). When using a
> GLM, a repeated measures design would like: (for example)
>
> lme(Bird_abundance = study_area + count +year+ X1 + X2 + X3,random =
> ~count|study_area).
>
> However, I have found no analogue design for a GAM. For now, I have
> averaged my bird abundances but I wondered whether a more subtle and
> elegant strategy exists...?
>
> Many thanks,
>
>
> Diederik
>
> Diederik Strubbe
> Evolutionary Ecology Group
> Department of Biology, University of Antwerp
> Universiteitsplein 1
> B-2610 Antwerp, Belgium
> http://webhost.ua.ac.be/deco
> tel : 32 3 820 23 85
>
>
>
>
> 	[[alternative HTML version deleted]]

-- 
> Simon Wood, Mathematical Sciences, University of Bath, Bath, BA2 7AY UK
> +44 1225 386603  www.maths.bath.ac.uk/~sw283