dbc: DBC criterion

In mlatif71/geemisc: Selection of working correlation structure

Description

Calculates deteminant-based criterion (DBC), which can be used to select the most appropriate working correlation struction in generalized estimation equation (gee). Either the arguments of geeglm or its output can be used as arguments of dbc.

Usage

dbc(object = NULL, formula, family, data, id, corstr,
  corstr.toeplitz = FALSE, ...)

Arguments

object	geeglm output object and the current implementation works only for "binomial" and "gaussian" families, and for "independence", "exchangeable", "ar1", "unstructured", and "toeplitz" correlation structures.
formula	see corresponding documentation of glm
family	see corresponding documentation of glm
data	see corresponding documentation of glm
id	see corresponding documentation of geeglm
corstr	see corresponding documentation of geeglm
corstr.toeplitz	an indicator that takes TRUE for "toeplitz" correlation structure
...	further arguments of geeglm to be passed

Value

rerurns dbc criterion value and the output of geeglm

References

Jaman A, Latif AHMM, Bari W, and Wahed A (2016). A determinant‐based criterion for working correlation structure selection in generalized estimating equations. Statistics in Medicine 35(11), 1819-1833.

See Also

geeglm, glm

Examples

library(geepack)
data(ohio)
#
fit <- geeglm(resp ~ age + smoke, id=id, data=ohio,
        family=binomial, corstr="exch")
dbc(fit)
#
# geeglm arguments can also be used as arguments 

fit2 <- dbc(formula = resp ~ age + smoke, id=id, data=ohio,
        family=binomial, corstr="exch")
        

mlatif71/geemisc documentation built on May 14, 2019, 12:56 a.m.