Danielle Navarro (“Prediction, pre-specification and transparency”) SUMMARY: Pre-registration tries to simultaneously satisfy three different goals: to make predictions (i.e., avoiding hindsight bias); to adhere to the statistical assumptions of confirmatory testing; and to provide transparency. Unappreciated is that these different goals require different specifications of pre-registrations. Further, in the case of transparency, pre-registration may not be the best tool for the job.
We should also add to this mix the recent Guest Blog from Nathan Goodman at TRN (“Hold the Bus!”) where he questions what seems like blind faith in pre-registration:
Where is the evidence that pre-registration works? That there are more null results? Is that the correct indicator of good science? Are there, in fact, ANY measurables that will enable us to know that pre-registration is producing better science? Those are good questions to which there have not been a lot of good answers to date.