robumeta: An R-package for robust variance estimation in meta-analysis

TL;DR

The paper introduces the 'robumeta' R-package that implements robust variance estimation methods for meta-regression, effectively handling dependent effect sizes without requiring detailed correlation information.

Contribution

It provides a versatile R package for robust meta-regression that accommodates unknown dependence structures among effect sizes, enhancing meta-analytic accuracy.

Findings

Valid point estimates and standard errors are obtained despite unknown dependencies.

The package supports various weighting schemes for flexible analysis.

Methods are distribution free and applicable to both large and small samples.

Abstract

Meta-regression models are commonly used to synthesize and compare effect sizes. Unfortunately, traditional meta-regression methods are ill-equipped to handle the complex and often unknown correlations among non-independent effect sizes. Robust variance estimation (RVE) is a recently proposed meta-analytic method for dealing with dependent effect sizes. The robumeta package provides functions for performing robust variance meta-regression using both large and small sample RVE estimators under various weighting schemes. These methods are distribution free and provide valid point estimates, standard errors and hypothesis tests even when the degree and structure of dependence between effect sizes is unknown.