Multi-curious: A Multi-Disciplinary Guide to Multiverse Analysis

Multiverse analysis has emerged as a valuable tool for increasing transparency and assessing robustness in empirical research across many scientific disciplines. By systematically exploring multiple defensible analytical paths, researchers can uncover how different decisions in data processing, model specification, and estimation impact results. However, while multiverse analysis holds significant promise, it also presents several challenges that researchers must navigate. This paper draws on interdisciplinary perspectives and knowledge to provide procedural guidance on design, reporting and interpretation of multiverse results. We discuss diverse data features, levels of precision, preregistration, and computational resources. We aim to produce cohesion across disciplines in multiverse analysis practices, perspectives and terminology, and contribute to ongoing efforts to improve the robustness and reproducibility of scientific results across disciplines.