Robust effect size index analysis using RESI R package
Presenting author: Megan Jones, Department of Biostatistics, Vanderbilt University Medical Center
Co-authored by:
- Kaidi Kang, Department of Biostatistics, Vanderbilt University Medical Center
- Simon Vandekar, Department of Biostatistics, Vanderbilt University Medical Center
Abstract:
Effect size indices are useful parameters that quantify the strength of association and are unaffected by sample size. There are many available effect size parameters and estimators, but it is difficult to compare effect sizes across studies as most are defined for a specific population parameter. We introduced a new effect size measure, the Robust Effect Size Index (RESI), which is advantageous because it is defined from M-estimators, so is not model-specific, uses a robust covariance estimate and can be used in a wide range of models. We recently developed confidence interval procedures for the RESI, useful for quantifying the estimate's precision. We present the RESI R package, which makes it easy to report the RESI and its confidence interval alongside common analyses. The package produces coefficient and ANOVA tables and overall Wald tests for many model types, appending the RESI estimate and confidence interval to each. The package also includes functions for conversion, visualization and interpretation. We will briefly introduce the RESI estimators and confidence interval procedure and walk through practical application with a demonstration of the RESI R package.