[R-sig-ME] Mixed Model, 2 Levels, Different VarCov Matrix

[will.eagle at gmx.net]

Dear all,

how can I specify a mixed model with two random effects variables with different covariance matrices in R, e.g. with lme() or lmer()?

Example: multiple, correlated disease indicators which are collected repeatedly over time.

patient time instrument value
100 t1 qol1 +0.56
100 t1 qol2 -0.45
100 t2 qol1 +0.67
100 t2 qol2 -0.34
101 ...

For the example above, one could assume a corCAR1() matrix for "time" and an corSymm() matrix for "instrument".

Since I am unsure with the syntax of the functions above, I would really appreciate a code example.

Thanks in advance,

Will