[R-pkgs] New package 'drm' for repeated categorical data analysis
jtjokine at cc.helsinki.fi
Thu Feb 15 12:30:47 CET 2007
A new package 'drm', version 0.5-4, is available on CRAN.
The drm package provides functions for marginal regression analysis of
repeated (or otherwise clustered) binary, ordinal and nominal responses.
This package can be considered as a likelihood-based alternative to GEE
approach for marginal regression. In addition to regression modelling,
several temporal and latent variable models for the associations between
repeated responses can be specified. In other words, the package provides
joint regression and association modelling for repeated categorical data.
For longitudinal studies with dropout, the package also provides a
possibility to model the dropout mechanism by adding a selection model on
top of the joint regression and association model. This can be used to
explore the effect of dropout on the regression and association parameters
when dropout is considered nonignorable.
The novelty of the proposed approach is that there exists an explicit
solution for the joint distribution, and the parameterisation has an
inherent unit-sum constraint, which substantially facilitate maximum
likelihood fitting for datasets with large cluster sizes. For an
application to a binary response with 12 repeated measurements, see the
help-file of function drm.
For more information, see:
Bug reports, comments or suggestions are welcome.
More information about the R-packages