MMR-GEE

Code for the simulation study and the analysis of the Pro-CCM data corresponding to draft manuscript "Estimating Marginal Treatment Effect in Cluster Randomized Trials with Multi-level Missing Outcomes". The data that support the findings in this paper are openly available at https://doi.org/10.7910/DVN/IIDE2B.

data-application folder

Contains R code for cleaning data and running the analyses for the Pro-CCM data:

• data-application.R: main code file that (1) cleans and reorganizes the data from the manuscript "Proactive Community Case Management Decreased Malaria Prevalence in Rural Madagascar: Results from a Cluster Randomized Trial" (Ratovoson et. al., 2022) and (2) fits four estimators (CC-GEE, IPW-GEE, MIPW-GEE-EM, and MMR-GEE) to the cleaned data.
• data-application-bootstrap.R: code for obtaining standard error estimates through the cluster bootstrap approach.
• EM.R: helper function for implementing the EM algorithm.
• MR.R: helper function for estimating the multiply robust weights.

simulation folder

Contains R code for the simulation study:

• run-methods.R: main code file for (1) estimating beta parameters and (2) obtaining bootstrap standard error using the cluster bootstrap approach for the simulation.
• data-generation: subfolder includes data-generation1.R, data-generation2.R, data-generation3.R for simulating the data under the Org-Pro-CCM, Alt-1, and Alt-2 design, respectively.
• methods: subfolder includes EM.R for implementing the EM algorithm, MR.R for estimating the multiply robust weights, run-GEE-no-EM.R for implementing CC-GEE, IPW-GEE, MIPW-GEE, and MMR-GEE without the EM algorithm, run-GEE-EM.R for implementing CC-GEE, IPW-GEE, MIPW-GEE, and MMR-GEE with the EM algorithm.