OSF — Preregistration and Open Science Projects

OSF hosts preregistrations, project files, and preprints free, from a non-profit. The timestamped, uneditable registration is what makes confirmatory claims credible.

Official Site https://osf.io
Category Data
Pricing Free
Rating ★★★★★ (5/5)

What OSF is

The Open Science Framework is a free research platform from the Center for Open Science, a non-profit. It hosts projects — files, wikis, and collaborators — mints DOIs, serves preprints, connects to your existing storage, and, most importantly, produces preregistrations: timestamped, permanently uneditable records of what you planned to do before you did it.

That last function is the reason OSF matters more than its feature list suggests. The distinction between a hypothesis you committed to in advance and one you formed after seeing the data is invisible in a published paper, and it is the distinction that determines whether a p-value means anything. A preregistration makes it visible and checkable. Everything else OSF does is useful; this one thing is infrastructure for a specific and serious methodological problem.

Why researchers use it

  • Preregistration that cannot be edited — a frozen, timestamped, citable record, which is what makes “we predicted this” a verifiable claim rather than an assertion.
  • Registered Reports — the format where a journal accepts the study on the basis of its design, before results exist, which addresses publication bias at the root.
  • Free DOIs for materials and data — questionnaires, stimuli, analysis plans, and datasets, all citable.
  • Storage connectors rather than migration — link Dropbox, Google Drive, GitHub, or Box and OSF becomes the project’s index without you moving files.
  • A non-profit steward — the Center for Open Science exists to do this, which is a different set of incentives from a company that happens to.

Where it fits in a research workflow

OSF wraps a study from before it starts until after it publishes. Preregistration comes first, files accumulate through the project, and at publication the project becomes public and its DOI goes in the paper.

Its relationship to neighbouring tools is complementary rather than competitive, and worth being explicit about because people expect it to do more than it does. Code belongs in GitHub — OSF’s versioning is basic and will not replace Git. Sensitive participant data belongs in REDCap or an institutional enclave, not in OSF’s storage. Software releases and large derived datasets are better archived on Zenodo. What OSF uniquely provides is the record of the study as a study: what was planned, what was collected, what was analysed, in one citable place.

Getting started

An hour, and the sequence is the point.

  1. Create a project for your current study and connect your existing storage rather than moving files into OSF. The connectors mean adoption costs nothing in reorganisation.
  2. Preregister before data collection, using one of the templates. This is not a formality you can backfill — a registration created after you have seen the data is worse than none, because it claims something untrue.
  3. Be specific in the registration: exact hypotheses, exact exclusion criteria, exact primary outcome, exact analysis. A vague preregistration protects nothing and is transparently defensive to anyone who reads it.
  4. The step people skip: register the analysis plan, not just the hypotheses. The garden of forking paths is mostly in the analysis — which covariates, which exclusions, which test — and a registration that leaves those open has not constrained the thing that needed constraining.

OSF vs the alternatives

AlternativeDoes it betterPick it if
ZenodoArchival DOIs with a stronger long-term preservation commitmentYou are depositing a finished output
GitHubReal version control, code review, and collaboration on codeThe artefact is software
REDCapCompliant capture of identifiable participant dataYou are collecting the data, not describing the study
AsPredictedA faster, lighter preregistration for simple designsYou want nine questions, not a full project

OSF is the only one of these that hosts a preregistration as part of a full project record. That combination is what it is for.

Cost, licensing, and your data

Free, with storage caps per project that are generous for materials and analysis outputs and insufficient for large raw data — which the storage connectors are designed to handle by keeping big files where they already live. The Center for Open Science is a non-profit funded by foundations and grants.

Two governance points. First, OSF is a public-science platform, and its defaults and culture assume eventual openness — appropriate for materials and analysis plans, wrong for identifiable participant data, which belongs in a controlled system. Second, and more subtly: a preregistration is permanent and uneditable by design. That is exactly the property you want, and it means a registration containing an error, an identifying detail, or a hypothesis you regret is not something you can quietly fix. Read it once more before you freeze it.

The honest review

Strengths. OSF made preregistration free, easy, and normal, and in doing so changed what is expected in several fields. The uneditable timestamped record is a small technical feature carrying enormous methodological weight: it converts a claim about your own intentions into evidence. That journals and funders now recognise OSF registrations means the incentive and the tool point the same way.

Limitations. Version control is basic — adequate for documents, inadequate for code, and people who expect Git-like behaviour are disappointed. Storage caps mean large datasets need a connector or a different repository. The interface is functional rather than pleasant, and organising a complex project inside it takes deliberate effort. And the deepest limitation is not the tool’s: a preregistration written vaguely enough to accommodate any result performs openness while constraining nothing, and OSF cannot stop you doing that.

Verdict. Adopt it for any confirmatory study, in any field where preregistration is recognised — it is free, it is expected, and a public OSF project is the strongest reproducibility signal a paper can carry. Skip it as a code repository or a data enclave, where dedicated tools are better. The condition that flips the answer is study type: exploratory work does not need registration, and pretending otherwise devalues the practice for everyone.

When NOT to use this Do not write a preregistration vague enough to be safe. A registration that leaves the primary outcome, the exclusion criteria, or the analysis specification open constrains nothing, and a reader who compares it to your paper will see that immediately — it performs rigour while providing none, which is worse than not registering. If you are not willing to commit to specifics, the honest move is to label the study exploratory and say so. And do not put identifiable participant data on a platform whose defaults assume openness.

Common questions

Is OSF free?

Yes — free, from the non-profit Center for Open Science, with per-project storage caps. Larger files are handled by connecting your existing Dropbox, Drive, GitHub, or Box storage rather than uploading.

Can I edit an OSF preregistration after submitting it?

No, and that is the point. A registration is timestamped and frozen, which is what makes it evidence. You can create a new registration or post an amendment, but the original stays visible — so read it carefully before you submit.

OSF or Zenodo?

OSF for the study as a study: preregistration, project structure, materials, and collaborators. Zenodo for archiving finished outputs with a stronger long-term preservation commitment. Many projects use both — OSF during the work, Zenodo for the version of record.