Readme file
SERIES
B
 Statistical
Methodology
Semiparametric estimation and inference for distributional and general treatment effects, by J. Cheng, J. Qin and B. Zhang, pages 881–904
Six programs are provided.
Two sets of programs are included. One set is used for general two-arm trials where the control group can get the treatment, including "ChengQinZhang_JRSSB_CausalEffect_General.R", "ChengQinZhang_JRSSB_GoodnessTest_General.R" and "ChengQinZhang_JRSSB_GoodnessGraph_General.R".
Another set is used for single consent design trials where the control group cannot get the treatment, including "ChengQinZhang_JRSSB_CausalEffect_Single.R", "ChengQinZhang_JRSSB_GoodnessTest_Single.R" and "ChengQinZhang_JRSSB_GoodnessGraph_Single.R".
To use the programs, you need to input the data at the beginning of the programs and may change the values of type I error, number of bootstrap samplings, and number of maximum iterations. All the programs are run in R.
"ChengQinZhang_JRSSB_CausalEffect_General.R" or "ChengQinZhang_JRSSB_CausalEffect_Single.R" computes estimates and confidence intervals for parameters discussed in the paper, including CACE, quantile effects, standard ITT and standard IV effects. The potential cdfs of compliers under treatment and control are plotted. It also provides the semiparametric empirical likelihood ratio test statistic and p-value for testing equality of potential densities of compliers under treatment and control.
"ChengQinZhang_JRSSB_GoodnessTest_General.R" or "ChengQinZhang_JRSSB_GoodnessTest_Single.R" computes p values in observed groups for the goodness-of-fit tests on the assumption of exponential tilt density ratio model. A small p value (<0.05) indicates a poor fit of the assumption to the data.
"ChengQinZhang_JRSSB_GoodnessGraph_General.R" or "ChengQinZhang_JRSSB_GoodnessGraph_Single.R" plots estimated empirical cdfs and estimated semiparametric cdfs in observed groups. Two close cdfs indicate a good fit of the assumption of exponential tilt density ratio model to the data.
Jing Cheng
Division of Biostatistics
University of Florida College of Medicine
Suite 5130
1329 SW 16th Street
PO Box 100177
Gainesville
FL 32610
USA
E-mail: jcheng@biost.ufl.edu
|
