Extended definition
A Registered Report is a publication format in which peer review and the acceptance decision occur before data collection. The manuscript moves through two stages: at Stage 1, the author submits the introduction, hypotheses, and a detailed plan of methods and analysis, judged on the merit of the question and the rigor of the design, with no result yet known. Once Stage 1 is approved, the journal grants in-principle acceptance, a commitment to publish the study regardless of what the data show, provided the author follows the protocol. At Stage 2, results and discussion are added to the approved protocol and undergo a second round of review, focused on adherence to the plan and the validity of the conclusions. Chambers and Tzavella (2022) describe the format as a remedy aimed at publication and reporting biases: by deciding publication on the question and the method, not the result, it decouples the luck of the finding from the fate of the article.
When it applies
A Registered Report applies to confirmatory research, with a clear hypothesis and an analysis plan definable before data collection. It is the right form for hypothesis tests, replications, and expensive or longitudinal studies, where the author wants to secure publication regardless of the sign of the result. It applies when the risk of bias is high: fields under strong pressure for positive results benefit from decoupling the finding from acceptance. Soderberg and colleagues (2021) compared Registered Reports with standard-model articles and found clear gains in the rigor of methods and analysis, with no meaningful loss of novelty. It also applies to the career planning of early-career researchers, who gain recognition already at Stage 1 and reduce the risk that a well-designed study goes unpublished for yielding a null result.
When it does not apply
The format does not apply to purely exploratory or inductive research, where hypotheses and analyses emerge from the data: forcing an a priori plan where none exists distorts the work. It does not apply well to studies on already-collected data, nor to designs that depend on serendipitous discovery. It is not a universal remedy: Chambers and Tzavella (2022) note that a Registered Report does not by itself correct execution errors, fraud, or generalization problems. And it does not apply without cost: the two-stage process is slower and more laborious, and the supply of journals that accept the format, though growing, is still limited in many fields. Mistaking the presence of a registered protocol for guaranteed quality is a misreading of the format.
Applications by field
- Psychology and cognitive science: the format’s birthplace; Scheel and colleagues (2021) showed that the share of positive results drops from about 96% in the standard literature to near 44% in Registered Reports.
- Biomedicine and clinical trials: growing use in replication and trials; the confirmatory nature of the RCT fits the format’s design.
- Economics and social sciences: more recent adoption, aimed against p-hacking and publication bias in empirical work.
- Ecology and animal behavior: journals in the field began offering the format to reduce the bias toward surprising findings.
Common pitfalls
The first pitfall is treating in-principle acceptance as unconditional acceptance: it holds only if the author follows the Stage 1 protocol, and undeclared deviations break the commitment. The second is confusing a Registered Report with simple preregistration: preregistration deposits the plan but includes neither prior peer review nor a publication guarantee. The third is submitting an exploratory study as confirmatory, squeezing an after-the-fact hypothesis into an a priori mold. The fourth is underestimating time: the two-stage cycle requires planning the Stage 1 submission well before data collection. The fifth is presenting Stage 2 exploratory analyses as if they were confirmatory; the format allows unregistered findings but requires labeling them clearly as exploratory.