Continuing their reflections on 20 years of research methods support, David and Lily explore multilevel data through examples from agricultural research in West Africa. They discuss how identifying the right levels can help account for variability and reveal important differences that might otherwise be missed.