Meta-analysis is the synthesis of information from multiple primary studies of similar design. A useful meta-analytic tool is meta-regression, wich serves to assess the relation between one or more study-level covariates and the observed effect size in a study.
Robustness is the ability of a statistical method to cope with violation of model assumptions. Robustness is of particular importance in meta-analysis, and various reasons for this are presented. Specific issues of robustness in the context of meta-analysis are raised. A robust estimator for meta-regression, termed meta.lts, is introduced, and its usefulness is demonstrated using simulation studies.