[R-sig-ME] multivariate glmm

[Torsten Hauffe]

Dear list members,

I' m fitting a multivariate time series using mgcv:::gamm. I cannot run a
summary on my fitted object and no gam.check(). Any idea how I can assess
the model fit and diagnostic plots?

Because I'm interested in the general trend of all dependent variables (dv)
over time, I use a manova-like style and not the mgcv:::mvn function. The
latter function seems to estimate individual coefficients for all dv and
not the general temporal trend of all dv together. I use a gamm because of
allows to include temporal autocorrelation, which is not part of the
minimal example below.

library(mgcv)
data(iris)
set.seed(1)

iris$x <- rnorm(n = nrow(iris) # Create independent variable
# Regular manova
Manova <- lm(cbind(Sepal.Length, Sepal.Width, Petal.Length, Petal.Width) ~
x, data = iris)
summary(Manova)
plot(Manova) # Does not work
# Now the gamm
Gamm <- gamm(cbind(Sepal.Length, Sepal.Width, Petal.Length, Petal.Width) ~
s(x), data = iris)
summary(Gamm$lme) # Works
summary(Gamm$gam) # Does not work
gam.check(Gamm$gam) # Does not work

Thank you for any suggestion,
Torsten

	[[alternative HTML version deleted]]