The informativeDropout package for R (>3.2.0) implements several models to account for dropout in longitudinal studies.
The package provides companion code for the manuscripts:
Moore, C.M., Carlson, N.E., MaWhinney, S. A Bayesian Natural Cubic B Spline Varying Coefficient Method for Non-Ignorable Dropout, In review.
Moore, C.M., Carlson, N.E., MaWhinney, S. A Dirichlet Process Mixture Model for Non-Ignorable Dropout, In review.
The results in the above manuscript were produced using R version 3.2.4. To reproduce the results, perform the following steps:
- Install R version 3.2.x or higher by following the instructions at http://www.r-project.org
- From the R environment, install and load the "devtools" package
install.packages("devtools")
library(devtools)- Install the "informativeDropout" package directly from Github.com
install_github(repo="kreidles/informativeDropout", ref="develop")- Load the library
library(informativeDropout)Instructions for replicating results in "A Dirichlet Process Mixture Model for Non-Ignorable Dropout"
Install the informativeDropout package as described above.
Run the DP model demo program
demo(ba_dp_paper)The demo will run for 200,000 iterations may take up to 24 hours to run, depending on processor speed.
Instructions for replicating results in "A Bayesian Natural Cubic B Spline Varying Coefficient Method for Non-Ignorable Dropout"
Install the informativeDropout package as described above.
Run the DP model demo program
demo(smmr_spline_paper)