Optimal design and statistical power for multilevel experiments investigating main, mediation and moderation effects
