At the beginning of a study, researchers formulate a research question and hypotheses and draft the study design, defining the scope of the study, the planned methods and analyses, and other relevant aspects. Preregistration should take place before data collection begins. For existing (secondary) data, it should take place before the data are accessed or analysed. This documents, in a way others can verify, that the research plans were defined before the results were known. Aspects that are not settled until later can be added to the registration during the ongoing study. Such additions should be clearly marked as such.

Preregistration can take place either as a private “local preregistration” or on a platform. Registering on a platform is preferable, because the platform keeps a verifiable record of when the registration was made and documents any later changes. Appropriate platforms are, for example, AsPredicted, the Open Science Framework (OSF) or the AEA RCT Registry for randomised controlled trials. These platforms offer ready-made templates that guide researchers through the process and help ensure that no relevant points are overlooked. A preregistration does not need to be long. However, the more specific it is, the easier it is to see which analyses were planned from the outset (confirmatory) and which were added later (exploratory). The OSF offers ready-made documents. AsPredicted requires answers to nine questions about the planned research design and analyses. The AEA RCT Registry requires several structured mandatory fields on the study design and planned analyses.

Whether and when a preregistration becomes publicly visible depends on the platform and the chosen settings. Many platforms allow the registration to be kept private at first (embargo) and released later, for example when the results are submitted or published. What matters is that the time of registration is documented. Once the preregistration is public, other researchers and the interested public can view and comment on it. A DOI can also be assigned, making the preregistration permanently citable and traceable.

The researchers are then expected to conduct the study according to the registered plan. Preregistration does not in itself prevent subsequent changes or the selective publication of results. It does, however, make it visible and verifiable whether and where the study deviated from the plan. Once a registration is public, it also shows that a study was planned, even if its results are never published. Deviations are not problematic in themselves, but they should be disclosed and justified in the publication (see Change Possibilities After a Pre-Registration). Once the results are available, they can be compared against the registered plan, reviewed or replicated.