Preregistration is the practice of publicly time-stamping a study’s hypotheses, methods, and analysis plan before data collection begins (or, for secondary-data studies, before the data are accessed for analysis). It creates a fixed, verifiable record of what a research team intended to test and how — so a reader can later tell the difference between a hypothesis that was tested as planned (confirmatory) and one that emerged from looking at the data first (exploratory). Neither is illegitimate on its own; the problem preregistration addresses is when exploratory findings are written up as if they had been predicted in advance, a practice researchers call HARKing (Hypothesizing After the Results are Known).
What preregistration actually commits you to
A preregistration is not a promise to run the study exactly as written and never deviate — it is a promise to be transparent about whether you did. At minimum, a study preregistration should specify:
- Hypotheses — stated specifically enough that a reader could tell in advance which results would support or contradict them.
- Study design and methods — sampling plan, sample size (and how it was determined), measures/variables, and procedure.
- Analysis plan — the specific statistical tests or models to be used, and how key decisions (exclusion criteria, handling of missing data, primary vs. secondary outcomes) will be made.
Registering the analysis plan is frequently treated as a distinct step from registering the study design — CASRAI’s pre-analysis plan entry covers the analysis-specific commitment in more depth, including how it’s used in fields (economics, some areas of political science) where the study design itself may be observational or quasi-experimental rather than a controlled experiment.
If the actual study departs from the registered plan — a common and expected occurrence, not a violation — the accepted practice is to report the deviation explicitly and re-label any analysis that wasn’t pre-specified as exploratory, rather than to quietly drop the registration or present the exploratory result as if it had been predicted.
What preregistration is not
A few common conflations worth separating out:
- It is not ethics approval. Institutional Review Board (IRB) or Research Ethics Committee approval addresses participant protection and is a separate, usually mandatory, process; a preregistration does not substitute for it, though many registries let you reference your ethics approval number.
- It is not the same thing as clinical trial registration, though the two overlap for clinical research. Trial registration on a WHO-recognized primary registry (most commonly ClinicalTrials.gov) is a publication and, in many cases, legal requirement with its own deadlines and a mandated data set. See CASRAI’s Clinical Trial Registration and Reporting Compliance guide for that specific regulatory landscape — a clinical trial registration can function as a preregistration, but general preregistration platforms (below) are used far more broadly, across disciplines that have no clinical trial registry requirement at all.
- It is not a guarantee of publication. A standard preregistration on a general repository is just a timestamped record — the journal you eventually submit to has made no commitment to accept the paper. (Registered Reports, covered below, work differently.)
Common preregistration platforms
OSF Registries
OSF Registries, run by the nonprofit Center for Open Science (COS) on the Open Science Framework, is the most widely used general-purpose preregistration platform across the sciences and social sciences. It offers discipline- and design-specific templates (including the standard Preregistration template and more specialized ones for particular study types), assigns each registration a DOI and immutable, timestamped record, and allows researchers to choose immediate publication or an embargo of up to four years before the registration becomes public. Because OSF is also used for project management, data, and materials, a registration can sit alongside — and later link to — the rest of a project’s OSF workspace.
AsPredicted
AsPredicted is a free preregistration platform run by the Wharton Credibility Lab at the University of Pennsylvania, launched in 2015. Its distinguishing feature is simplicity: researchers answer a short, fixed set of questions about their design and analysis (roughly nine), and the platform generates a brief, time-stamped, easy-to-read PDF. It’s popular in psychology and behavioral economics specifically because the format is fast to complete and fast for a reader (including a peer reviewer) to check against the eventual manuscript. The Wharton Credibility Lab has since extended the same “irresistibly easy” philosophy to related tools (ResearchBox for materials/data, AsCollected for documenting results provenance).
ClinicalTrials.gov and other primary trial registries
For health-related clinical trials, registration happens on a WHO-recognized primary registry rather than a general preregistration platform — most commonly ClinicalTrials.gov (operated by the U.S. National Library of Medicine), but also national/regional registries such as the EU’s CTIS, ISRCTN, or ANZCTR, all discoverable through the WHO International Clinical Trials Registry Platform (ICTRP). This route carries its own timing rules, a mandated 24-item WHO Trial Registration Data Set, and — for many trials — a legal results-reporting obligation, not just a norm. Full detail is in CASRAI’s dedicated clinical trial registration and reporting compliance guide.
Discipline-specific registries
Some fields maintain their own protocol registries rather than relying on general platforms. PROSPERO, maintained by the University of York’s Centre for Reviews and Dissemination, is the standard prospective register for systematic review protocols in health and related fields, and exists specifically to reduce unplanned duplicate reviews and expose changes made to a review’s methods after it started. Discipline-specific registries like this typically ask for more field-appropriate detail than a general template would.
How preregistration relates to Registered Reports
Preregistration and Registered Reports are related but distinct, and the terms are often used loosely as if interchangeable — they aren’t:
- Preregistration is something a researcher does unilaterally on a registry, independent of any journal. No editor or reviewer evaluates the plan in advance, and registering doesn’t obligate anyone to publish the eventual results.
- A Registered Report is a journal publication format built around the same idea, but with peer review moved earlier: a Stage 1 manuscript (introduction, methods, planned analysis) is peer reviewed and can receive in-principle acceptance before the study is run. The journal then commits to publishing the completed study in Stage 2 regardless of whether the results are positive, negative, or null — the editorial decision is about the quality of the question and method, not the outcome. Over 300 journals across disciplines from psychology to medicine to environmental science now offer Registered Reports as a submission track, according to the Center for Open Science.
In practice, a Registered Report’s Stage 1 protocol typically is the preregistration — it’s usually deposited on a registry (often OSF) at the point of in-principle acceptance. So the two aren’t competing options; a Registered Report is one specific, journal-mediated way of formalizing a preregistration and attaching a publication commitment to it. A closer side-by-side comparison of Registered Reports as a publication format is planned as a dedicated CASRAI resource.
Why funders and journals increasingly require or reward it
Preregistration moved from a niche practice to an expected one largely through two mechanisms:
- The TOP Guidelines. The Transparency and Openness Promotion (TOP) Guidelines, developed by the Center for Open Science and released in 2015 (substantially revised in 2025), give journals and funders a modular framework of standards — including Study Registration and Analysis Plan as two of the core practices — each implementable at increasing levels of stringency, from simply disclosing whether a study was registered up to requiring and independently verifying it. More than 5,000 journals and organizations, including the American Psychological Association, have signed on as TOP signatories.
- Direct journal and funder policy. Beyond TOP, individual journals increasingly ask authors to state whether a study was preregistered and to link the registration; some award Open Practices Badges for it. For clinical research specifically, registration is tied to legal and editorial requirements (ICMJE policy, FDAAA 801, NIH’s dissemination policy) that long predate the broader preregistration movement — see the clinical trial guide linked above for those specifics.
The underlying rationale funders and journals give is consistent across all of these: preregistration makes it possible to distinguish confirmatory from exploratory findings after the fact, which directly addresses two of the most-cited contributors to the reproducibility crisis in the empirical literature — selective reporting of only the results that “worked,” and HARKing. Neither requires malicious intent to occur; both are addressed by a plan that exists, verifiably, before the data do.
How to preregister a study, step by step
- Pick a platform based on your field and study type: OSF Registries for most disciplines, AsPredicted for a fast, simple design, a primary clinical trial registry for health research involving human participants, or a discipline-specific registry (e.g., PROSPERO for a systematic review) where one exists.
- Draft the plan using a template. Most platforms provide one — answer every question concretely enough that someone else could tell, after seeing your results, whether you’d followed the plan.
- Decide on visibility. Most platforms allow an embargo (OSF Registries: up to four years) so the registration isn’t public until you’re ready — useful for competitive fields, but the timestamp itself is still fixed at submission regardless of when the content becomes visible.
- Submit and receive a persistent identifier. OSF Registries and most clinical trial registries assign a DOI or registry number you can cite.
- Cite the registration in the eventual manuscript — including its identifier and a statement about any deviations from the registered plan — so reviewers and readers can check the paper against it directly.
Frequently asked questions
Is preregistration required, or just encouraged?
It depends on the context. For clinical trials, registration is frequently a legal and/or publication requirement (see the clinical trial guide above). Outside clinical research, preregistration is rarely mandatory by default, but is increasingly required or strongly incentivized by specific journals, funders, and TOP-signatory policies — check the individual journal’s or funder’s author/applicant guidelines rather than assuming a blanket rule.
Can I preregister a study using data I already have?
Not in the traditional sense — preregistration’s evidentiary value comes from the plan existing before the relevant data are seen. For work with existing (secondary) datasets, the accepted equivalent is registering the analysis plan before that specific analysis is run on the data, sometimes called a “pre-analysis plan” (see the linked term above) — and being explicit in the write-up about what was and wasn’t already known about the dataset at the time of registration.
What happens if my study doesn’t go as planned?
Report it. Deviating from a registered plan is normal and does not invalidate the registration — what matters is disclosing the deviation and clearly separating any resulting exploratory analysis from the pre-specified confirmatory analysis, rather than silently rewriting the study’s narrative around what was actually found.
Does preregistration slow down or restrict research?
It adds an upfront planning step, but it doesn’t prevent exploratory work — it just requires labeling it honestly. Most researchers who adopt it report the main cost is the discipline of writing a specific-enough plan before starting, not any restriction on what they’re later allowed to analyze or report.
Where does preregistration fit alongside a data management plan?
They’re complementary and often prepared around the same time, but answer different questions: a Data Management Plan covers how data will be handled, stored, and shared; a preregistration covers what you’re testing and how you’ll analyze it. Some funders now expect both.
For the broader scholarly-publishing landscape this fits into, see CASRAI’s Scholarly Publishing hub, which also links to related guides on choosing a preprint server and writing a data availability statement.
