A Bayesian Approach to Estimating Effect Sizes in Educational Research

In educational research, a first step in the descriptive analysis of data from any psychometric measurement of the performance of subjects in tests at different time points and between groups is to compute relative measures of learning gains and learning achievements. For these effect sizes, there are classical approaches coming from frequentistic statistics like Student's or Welch's $t$-test with their own strengths and weaknesses. In this paper, we propose a purely Bayesian approach for analysing within-group and between-group differences in learning outcomes, taking naturally into account the multilevel structure of the data, as well as heterogeneous variances among time points and groups. We provide a detailed implementation using the brms package in R serving as a wrapper for the probabilistic programming language Stan, facilitating the implementation of these methods in future research by including online supplementary material. We recommend that for a pooled design, one computes an effect size $d_s$, and for a paired design, one should compute two possibly different quantities $d_s$ and $d_z$ to correct for correlations in within-group designs and allowing for comparability across different studies. All these effect sizes are based on ideas coming from Hedge's total effect size $δ_t$ introduced in 2007.