gee_stepper: Stepwise GEE Model Selection
In dewittpe/pstools: Propensity Score Analysis Tools

Description Usage Arguments Value Examples

Starting with a model of y ~ 1, use a forward/backward step selection process to find the preferable fit.

1	gee_stepper(fit, upper)

`fit`	a `geeglm` object. the response, id, data, corstr, ... will be reused in all the the fits called by `gee_stepper`
`upper`	A formula with the rhs containing all the variables that could be the the model.

the "best" model based on QIC from the stepwise selection process.

data(pride)
library(geepack)
geefit <- geepack::geeglm(PCR_RSV ~ SEX + RSVINF + REGION + AGE + ELTATOP + EINZ + EXT, 
                  id = REGION, data = pride, family = stats::binomial())
gee_stepper(geefit, formula(geefit))