Registered Reports

Registered Reports

Reviewed by Faisal Zaidi

A Registered Report is a paper accepted for publication before its results exist. The design and the analysis plan go through peer review first, and if they are sound the journal commits to publishing whatever comes out. That single change removes the incentive that produces a literature made almost entirely of successes.

What it is

The ordinary sequence is: run the study, see the result, write it up, submit. At every stage there is a filter favouring positive findings. Researchers are more likely to write up a result that worked. Reviewers find a positive result more convincing. Editors prefer papers people will cite. None of that requires anyone to behave dishonestly, and all of it distorts the published record in the same direction.

A Registered Report inverts the order. Stage 1 is a submission containing the question, the hypotheses, the design, the sample size and power calculation, and the exact analyses that will be run. Reviewers assess whether the question matters and whether the design could answer it. If accepted in principle, the journal is committed. Stage 2 is the same paper with the results filled in, reviewed for whether the authors did what they said, not for whether the answer is interesting.

The consequence is that a null result is publishable by construction. That is the whole point, and it is why the format is a structural fix rather than an exhortation.

More than three hundred journals now offer it, either as a standing submission route or as part of a special issue. The Center for Open Science's own count includes one-off special issues, so the number is a ceiling rather than a measure of routine adoption.

In effect

The cleanest measurement comes from Anne Scheel, Mitchell Schijen and Daniel Lakens in 2021. They took every psychology Registered Report published up to November 2018, which is seventy-one papers, a complete population rather than a sample. For comparison they drew 152 ordinary articles using the same search method Daniele Fanelli had used in 2010.

In the ordinary literature, 146 of 152 articles concluded that their first hypothesis was supported. That is 96 per cent. Among Registered Reports, 31 of 71 did, which is 44 per cent.

Two things about that gap are usually misreported. "Positive" was not a significance threshold: following Fanelli, the coders recorded whether the authors themselves said their first stated hypothesis was supported, and partial support counted. And more than half the Registered Reports were close replications, which plausibly lowers their hit rate. Excluding those, the comparison becomes roughly 96 per cent against 50 per cent, so the gap narrows and does not close.

The format also produced psychology's most instructive failure. The 2023 paper announcing that rigorous practices deliver 86 per cent replicability was retracted in 2024, partly because its central analyses were not preregistered despite the paper saying they were. That is the format's machinery working, and it is also a reason not to rely on the number.

What it does not say

It does not separate publication bias from Type I error inflation. Scheel and colleagues say so themselves: the gap establishes that something filters the ordinary literature, not how much of the filtering happens at the journal and how much in the analysis.

It does not guarantee adherence. Accepting a plan in advance constrains the analysis; it does not make anyone follow it. Across 459 preregistrations matched to their publications, 52 per cent omitted at least one preregistered hypothesis and 57 per cent added one.

And it is not a general-purpose format. It suits confirmatory hypothesis testing and fits exploratory or descriptive work badly, which is a real limitation given the argument that psychology skipped the descriptive stage in the first place.

Sources

  1. Scheel, A. M., Schijen, M. R. M. J., & Lakens, D. (2021). "An excess of positive results: comparing the standard psychology literature with registered reports." Advances in Methods and Practices in Psychological Science, 4(2). DOI 10.1177/25152459211007467.
  2. Fanelli, D. (2010). PLoS ONE, 5(4), e10068. The search method and the 91.5 per cent baseline.
  3. van den Akker, O. R., et al. (2023). "Selective hypothesis reporting in psychology." AMPPS, 6(3).
  4. Center for Open Science, Registered Reports journal count, retrieved 2026-09-10.
  5. Where this came from
  6. No book in this library is the source of this concept. It entered the vault from What Forty-One Participants Buys, built from primary sources rather than from a book.