geefirth: Fitting Firth-type GEE for correlated binary data with...

In mhmondol/geefirthr: Fitting Firth-type GEE for correlated binary data with separation or near-to-separation

Description

geefirth fits GEE with Firth-type penalization to provide bias-corrected finite estimate of the regression coefficient in case of separation or near-to-separation. In addition, it provides bias-correctedsandwich estimate of the standard error of the penalized GEE estimate.

Usage

geefirth(formula, id = id, data = parent.env(),
  corstr = "independence")

Arguments

formula	Similar as geeglm
id	cluster variable and requirs to be sorted
data	data frame that contains the variables mentioned in id and formula
corstr	working correlation structure, "independence" (default), "ar1", "unstr", "exchangeable".

Author(s)

Momenul Haque Mondol [e-mail], M. Shafiqur Rahman [e-mail]

Examples

data(quasiDat)

# Fitting GEE for quasi-separated data

 geefirth(y ~ x + obstime, id=id,
       data=quasiDat, corstr = "exchangeable");

data(nearDat)

# Fitting GEE for near to quasi-separated data

 geefirth(y ~ x + obstime, id=id,
          data=nearDat, corstr = "exchangeable");

mhmondol/geefirthr documentation built on May 20, 2019, 7:25 a.m.