geeUOmega: Modified gee function to output extra objects for saws

In saws: Small-Sample Adjustments for Wald Tests Using Sandwich Estimators

Modified gee function to output extra objects for saws

Description

This function is normally not to be called directly, but one should usually use mgee (see warning below).

This function takes output from the gee function from the gee package and creates a score matrix (i.e., estimating equation) and information array (i.e., minus the derivative of the estimating equation). Note the function creates the X matrix assuming the data set is the same as it was for the original call to gee, see Warning section.

Usage

geeUOmega(geeOutput)

Arguments

geeOutput

object of class gee, output from gee function

Value

A gee object with two extra elements to the list, u and omega (see saws).

Warning

It is safer to use the mgee function, which internally calls gee then geeUOmega. If you do not use mgee, and instead call geeUOmega directly, there could be a problem if the data set has been changed after the initial gee call. This is because the model matrix (i.e., the X matrix) is not saved as part of the gee object, we must recreate it from the gee call. So it is created assuming that the data argument in gee means the same thing that it did when gee was called. So if you change the data set between the original gee call and using the geeUOmega function, there may be problems.

Note

The function recalculates the fitted.values and the residuals to the gee object, since in gee (version 4.13-18 at least) the fited.values and residuals can be wrong if there is an offset or if y is a matrix (as in the binomial model).

Author(s)

M.P. Fay, with some lines copied from gee function

See Also

gee,mgee

Examples

##  example from gee help
data(warpbreaks)
geeout<-gee(breaks~tension,id=wool,data=warpbreaks,corstr="exchangeable")
guo<-geeUOmega(geeout)
saws(guo)

saws documentation built on June 24, 2022, 1:07 a.m.