What should a preregistration contain?

Abstract

A large amount of variation exists in beliefs about the purpose and benefits of preregistration, making it difficult to implement and evaluate, and limiting its usefulness. Additionally, no single resource exists to describe what a preregistration should contain or how it should be used. In this paper, I describe what an effective preregistration should contain and when it should be used. Specifically, preregistration should 1) restrict as many researcher degrees of freedom as possible, 2) detail all aspects of a study’s method and analysis, 3) detail information on decisions made during the planning stages, and 4) specify how the results will be used and interpreted. Further, a preregistration must be publicly verifiable and permanent. Finally, I argue that pre-registration should be used in any situation where researchers intend to collect data in order to make a claim, description, decision, or inference based on that data. I also note that preregistrations which do not address each of these points do more harm than good by falsely signalling credibility and quality.