Registered Reports: Peer Review Before Data Collection Fixes Publication Bias

Registered Reports enact a structurally simple but consequentially profound inversion of peer review: the evaluation point shifts from post-result to pre-data-collection. Authors submit their research question, hypotheses, and full methodology for review. If accepted in principle, the journal guarantees publication regardless of outcome. This decouples editorial decisions from result valence, directly attacking the Incentive architecture that drives publication bias, p-hacking, and Selective Reporting — pathologies well-documented across biomedical, psychological, and social sciences.

The empirical signal is already striking. In fields where over 90% of traditionally published findings are positive, approximately 60% of Registered Reports yield null or negative results. This inversion is not evidence of worse science; it is evidence that the filter has been corrected. Early citation analyses further undermine the longstanding editorial objection that null results depress journal impact factors — Registered Reports appear to be cited at comparable or higher rates than conventional papers. The mechanism likely involves methodological quality: pre-data review allows genuine correction of design flaws, statistical power issues, and analytic ambiguities before they become irreversible.

The deeper structural implication concerns institutional redundancy. A Registered Report is functionally identical to a grant application: both present a question, a rationale, and a method for prospective evaluation. This convergence opens the possibility of a unified review process in which funding bodies and journals co-evaluate proposals, simultaneously committing resources and publication space. Such integration would eliminate the well-known problem of funded studies that never reach the literature, closing the loop between investment and accountability in the research enterprise.