Following other fields and disciplines, public administration scholars have recently embraced the practice of pre-registering experimental studies. Despite being cumbersome, such a ‘best-practice’ has the potential to contribute to the process of knowledge creation by: (i) forcing researchers to distinguish predictions, in which data are used to test the possibility that a hypothesis is wrong, from postdictions, in which propositions are used to explain what is observed in the data, and (ii) possibly mitigating publication bias. However, for these epistemic benefits to be observed, a few conditions need to hold, including: (i) registrations being submitted prior to data collection and analysis, (ii) published studies reflecting execution of pre-registered plans, and (iii) deviations from the original plans being transparently reported in published studies. We report findings from a systematic review of pre-registered experiments in public administration to show that, in most cases, these conditions do not hold. We conclude by discussing recommendations on how to make the best out of pre-registrations.
Belardinelli et al. (Thu,) studied this question.