remiod: Reference-based Controlled Multiple Imputation of Longitudinal Binary and Ordinal Outcomes with non-ignorable missingness

The package remiod provides functionality to perform controlled multiple imputation of binary and ordinal response in the Bayesian framework. Implemented are (generalized) linear regression models for binary data and cumulative logistic models for ordered categorical data (Wang and Liu 2022). It is also possible to fit multiple models of mixed types simultaneously. Missing values in (if present) will be imputed automatically.

remiod has two algorithmic backend. One is JAGS, with which the function performs some preprocessing of the data and creates a JAGS model, which will then automatically be passed to JAGS with the help of the R package rjags. The another is based on the method proposed by Tang (Tang 2018).

Besides the main modelling functions, remiod also provides functions to summarize and visualize results.

Installation

you can install remiod from GitHub:

# install.packages("remotes")
remotes::install_github("xsswang/remiod")

Main functions

remiod provides the following main functions:

remiod                      #processing data and implementing MCMC sampling
extract_MIdata              #extract imputed data sets

Currently, methods remiod implements include missing at random , jump-to-reference , copy reference , and delta adjustment . For , argument should follow to specify a numerical values used in delta adjustment

Functions summary(), coef(), and mcmclot() provide a summary of the posterior distribution and its visualization.

Minimal Example

data(schizow)

test = remiod(formula = y6 ~ tx + y0 + y1 + y3, data = schizow,
              trtvar = 'tx', algorithm = 'jags', method="MAR",
              ord_cov_dummy = FALSE, n.adapt = 10, n.chains = 1,
              n.iter = 100, thin = 2, warn = FALSE, seed = 1234)

extdt = extract_MIdata(object=test, method="J2R",mi.setting=NULL, M=10, minspace=2)

Reference