How Should Meaningful Evidence Be Generated From Datasets?

Abstract

Datasets are often considered "ideal" when they are large and contain longitudinal and representative data. But even research that uses ideal datasets might not generate high-quality evidence. This article emphasizes the roles that transparency plays in enhancing observational epidemiological findings' credibility and relevance and argues that epidemiological research can produce high-quality evidence even when datasets are not ideal. This article also summarizes strategies for bolstering transparency in key phases of research planning and application.