ABSTRACT
While meta-analyses typically possess a higher power than the primary studies they include, insufficient statistical power can still be an issue. Power analyses can be used to determine how many, and what kind of studies are needed to detect a specific effect size in a meta-analysis. They allow to verify if enough primary studies are available to ascertain an expected effect, or if the study search and eligibility criteria need to be adapted.
This chapter provides a brief refresher on how statistical power is defined, an how it can be calculated for meta-analyses. Formulas for the fixed-effect model, as well as the random-effects model (assuming different degrees of heterogeneity) are provided, including hands-on examples in R. Lastly, the chapter also discusses how power analyses can be conducted for subgroup analysis contrasts.
