Dear R-listers,
I am fitting bivariate mixed models for cost-effectiveness data of cluster
randomized trials using lme in R. So I have individuals nested within
clusters. My response variable is a vector with bivariate response (individual
level costs and effects) stacked into a single column. The covariates in my
models are a constant and a treatment term. They are response-specific, e.g. a
constant for cost (cons.1) and one for effect (cons.2) , a treatment term for
cost (treat.1) and one for effect (treat.2). The syntax I used is this:
lme(yl~
-1+cons.1+treat.1+cons.2+treat.2,
random=~-1+cons.1+cons.2|cidl,
weights=varIdent(form=~1|cons.1),
corr=corCompSymm(form=~1|cidl/indl),
data=datalong)
where cidl = cluster id, indl = individual id and datalong = data (also data
had been stacked before model fitting)
I have been trying to extend this model to allow individual level standard
deviations and correlations to differ by cluster. In my data, there is evidence
that SDs for costs and effects and correlations between costs and effects
differ by cluster. I have worked out the code to account for the former but not
both (see code below).
lme(yl~
-1+cons.1+treat.1+cons.2+treat.2,
random=~-1+cons.1+cons.2|cidl,
weights=varIdent(form=~1|resp.cidl),
corr=corCompSymm(form=~1|cidl/indl),
data=datalong,
control=lmc2) # SANN: 20k iterations
where resp.cidl = response-specific cluster id, lmc2 = a lmcControl object
specifying SANN as optimisation method among other settings
I would like to find out if there is a way to allow for the correlations to
differ by cluster via the corr option. Any suggestion or pointer would be
greatly appreciated.
Many thanks in advance.
Best wishes,
Edmond
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Edmond Ng, MSc CStat CSci
Lecturer in Medical Statistics
Department of Health Services Research and Policy, Faculty of Public Health and
Policy
London School of Hygiene and Tropical Medicine, 15-17 Tavistock Place, London
WC1H 9SH, UK
Tel: +44 (0)20 7927 2366 (direct line: ext 2065) Fax: +44 (0)20 7927 2701
Website: http://www.lshtm.ac.uk/hsru
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
