[R-pkgs] New car package with new linear.hypothesis function
Dear R-packages list members,
Last month, Peter Muhlberger posted a message to the r-help list that
suggested, among other things, a desire for a more convenient method of
performing Wald tests for statistical models fit in R.
Thanks largely to Achim Zeleis's help, the linear.hypothesis function in the
car package has been reworked so that (1) it is applicable to any model
object that responds to the generic coef() and vcov() functions (along with
specific methods for lm and glm objects); and (2) permits symbolic
specification of hypotheses as linear equations in the named parameters of
the model. I'd be particularly interested in people's reactions to the new
interface and to learn of any problems that arise
The new version 1.1-0 of the car package (with some other, less significant
changes and additions) is now on CRAN.
Department of Sociology
Canada L8S 4M4