g.dicho: The estimator for the g-formula that utilizes data only...

In mcl868/causalinmisdata: Causal inference in missing data

The estimator for the g-formula that utilizes data only complete cases

Usage

g.dicho(mmodels, exposure, data, ...)

Arguments

mmodels	The models correspond to the models the outcome.
exposure	The time-varying exposure.
data	Data.

Value

coef	The coefficients in marginal strucktural model.
ExpectEstimate	The expected value of the potential outcome.
mmodels	The mmodels that have been used for modeling data.
exposure	The exposure of the analysis.
N	The sample size of data.
NCC	The sample size of complete cases of data. In case of no missing values NCC is equal to N
...

Author(s)

Thomas Maltesen [e-mail]

References

put references to the literature/web site here

Examples

## Not run: 
model1 <- Y ~ L0 + A0 + L1 + A1 + L2 + A2 + L2*A2
model2 <- L2 ~ A1 + L1 + A0 + L0
model3 <- L1 ~ A0 + L0 + A0*L0


estimationSG<-lapply(1:loop,function(iiii) g.dicho(mmodels=c(model1,model2,model3),
                                                   exposure=c("A0","A1","A2"),
                                                   data=DataSetFull[[iiii]])$coef)
round(listMean(estimationSG),3)
     (Intercept)    A0    A1    A2 A0*A1  A0*A2  A1*A2 A0*A1*A2
Est.      -0.007 6.009 3.996 5.002     0 -2.004 -1.005        0


estimationSG.NA<-lapply(1:loop,function(iiii) g.dicho(mmodels=c(model1,model2,model3),
                                                      exposure=c("A0","A1","A2"),
                                                      data=DataSetMonotone[[iiii]])$coef)
round(listMean(estimationSG.NA),3)
     (Intercept)    A0    A1    A2 A0*A1  A0*A2  A1*A2 A0*A1*A2
Est.      -0.249 6.443 3.555 5.743     0 -2.551 -0.782        0

## End(Not run)

mcl868/causalinmisdata documentation built on March 5, 2024, 8:22 a.m.