Skip to contents

Changelog

medRCT 0.1.0

  • An initial public release of medRCT package, version 0.1.0

  • Supports multiple mediators, intermediate confounders, and various variable types, including categorical exposures, and binary or continuous mediators, intermediate confounders, and outcomes.

  • Handles scenarios with a single mediator and/or without intermediate confounders.

  • Provides a shiny app for model assessment.

medRCT 0.1.1

  • Updated the reference for the JOSS paper.

medRCT 0.2.0

  • JOSS paper published (added badge in README)

  • Fix bug when mediator and intermediate confounder are factors

  • Bootstraps that return warnings or errors are now disregarded, and the number of failed bootstraps is reported in the summary output

  • Add argument to specify the effect measure

  • Add package logo

medRCT 0.2.1

  • Added the separation_method argument to handle separation in binomial (binary outcome) models.

    • Options are "brglm" (logistic regression models are fitted using bias reduction methods via the brglm2 package), or "discard" (if separation is detected, the function returns NA. If this occurs during the main estimation, the program stops; if during bootstrapping, the affected bootstrap samples are discarded).

  • Fixed a bug when using custom interaction terms.