The eefAnalytics R package for the Analysis of Randomised Controlled Trials in Education

The eefAnalytics R package provides functions to analyse randomised controlled trials according to the study design, including simple randomized trials, cluster randomized trials, and multisite trials. The eefAnalytics package enables R users to easily obtain conditional and unconditional Hedges’ g effect sizes and their confidence or credible intervals. A central contribution of this software package is the correct computation of effect size uncertainties incorporating the variance of the denominator. Statistical models are chosen according to design type, with a simple linear regression model for simple randomized trials, and appropriate multi-level models for cluster randomised and multisite trials. Beside the effect sizes and their confidence/credible intervals in both frequentist and Bayesian approaches, the package also facilitates Bayesian, permutation and bootstrap techniques, and provides functions for the visualization of results such as posterior random effects. This article provides an overview of the modelling techniques used as the basis of the eefAnalytics package and demonstrates its application with examples. The package is available from CRAN: https://cran.r-project.org/package=eefAnalytics.