Papex
Explore
← Blog

Guide17 min read

How to publish a hypothesis without data: journals, preprints and DOIs

Which journals and preprint servers accept a hypothesis without data, from Medical Hypotheses to Octopus, how to get a DOI, and how to cite it.

Papex

You can publish a hypothesis without data, but not everywhere. Journals with a hypothesis article type, such as Medical Hypotheses, Biology Direct, RIO and the Frontiers journals' Hypothesis and Theory articles, review and publish them. Octopus and Preprints.org post them without a journal, while arXiv, bioRxiv and medRxiv turn away research proposals or hypotheses without new data. For a public, dated version with a DOI, Zenodo, OSF and Figshare accept any research output. Wherever it goes, write the hypothesis so that someone else can test it: one clear claim, its predictions, and the result that would show it wrong.

This guide covers where a hypothesis paper can go, what each venue says it accepts, how to get a DOI without a journal, what the paper should contain and how others should cite it. Every policy below was checked on 13 October 2026 on the provider's own pages or, where a page blocked automated access, on its most recent Internet Archive copy, noted in the sources.

Can you publish a hypothesis without data?

Yes, in venues made for it. A hypothesis paper argues from evidence that is already published: it proposes an explanation, shows how it fits what is known and says how it could be tested. Medical Hypotheses' guide for authors, in its latest readable copy from 2024, says "inclusion of extensive new data is not usually acceptable". Ideas in Ecology and Evolution does not publish "papers based primarily on experimental, data-driven studies". Frontiers' Hypothesis and Theory articles "may include original data", but need not.

What you cannot count on is a general preprint server: several of the largest will not post hypotheses without data. So start from what you need most: peer review and journal credit, a fast public version with a DOI, or a place your field already reads.

Which journals publish hypothesis papers?

Each venue below has a named article type for a hypothesis or a new idea. The details are as stated on the venue's own pages.

VenueArticle typeReview before publicationDOICost, as statedNotes
Medical Hypotheses (Elsevier)Hypothesis papers in medicine and biomedical scienceEditor and external reviewers, judging "the premise, originality and plausibility"YesNo fee for subscription publication; USD 3,280 for open accessExtensive new data "not usually acceptable"
Frontiers journals, such as Psychology and NeuroscienceHypothesis and TheoryPeer reviewedYesA-type fee: CHF 1,495 to 3,150, depending on the journalUp to 12,000 words; article types are set per journal
Biology Direct (BMC, Springer Nature)HypothesisSingle-anonymous peer reviewYesGBP 2,190, USD 2,890 or EUR 2,490Open access
Research Ideas and Outcomes, RIO (Pensoft)Research IdeaOpen review you arrange: at least two positive reviews, or an editor's endorsementYes, and a new one for each updated versionEUR 600 up to 20,000 charactersAlso publishes grant proposals
Ideas in Ecology and Evolution (Queen's University)New IdeaPeer reviewed; referees are named in the paperYes$200, or $100 if you arrange the review; $50 if all authors are graduate students or postdocsMust include "a proposal for testing the idea"
Octopus (Jisc and Octopus Publishing CIC)Rationale/HypothesisNone: "no acceptance or rejection"; open reviews afterwardsYes, per publication and per versionFreeLinks to a Research Problem on Octopus; sign-in with ORCID

Fees exclude VAT or local taxes where the venue says so, and several venues offer discounts or waivers. Three notes on the table.

Octopus works differently. It splits research into eight linked publication types, such as Research Problem, Method and Results. A Rationale/Hypothesis "sets out the theoretical basis for a potential solution, or partial solution, to the 'research problem' it is linked to". Each publication is linked to "the one above it in the chain", and Octopus suggests searching for an existing Research Problem first. Nobody accepts or rejects it; others can post open reviews afterwards, and you can mark a hypothesis as equivalent to a preregistration (author guide).

At RIO you arrange the review. For a Research Idea, authors "source and invite at least three or more external peer review(s)" at submission, and reviews are published with the reviewers' names. RIO's site was down for maintenance on 13 October, so this comes from its archived copy of 15 September 2026.

Two names mislead. The journal called Hypothesis publishes research for health information professionals, not hypotheses in general. Seeds of Science, a community-reviewed journal for "speculative or exploratory research articles", "was closed in February, 2025".

BioEssays, which Wiley-VCH describes as "a peer-reviewed, review-and-discussion journal" with "hypotheses encouraged", is another option. Wiley's author pages blocked automated access, so check its current article types and charges with the journal.

Is Medical Hypotheses a credible journal?

Medical Hypotheses is an Elsevier journal that has used external peer review since 2010; its reviewers judge a hypothesis's premise, originality and plausibility, so a paper there is a reviewed proposal, not a tested result. Its aims and scope say submissions "will be reviewed by the Editor and external reviewers", who judge "the premise, originality and plausibility of the hypotheses submitted". The journal exists, in its own words, to give "radical new ideas and speculations in medicine open-minded consideration".

Its history explains why people ask. Until 2010 it published papers "on the basis of how interesting or radical they are rather than using peer review" (Times Higher Education, March 2010). In July 2009 it published a paper denying the link between HIV and AIDS. After the outcry, Elsevier told the editor to adopt peer review or resign, and that paper and one other were "permanently withdrawn" after five anonymous reviewers, in a process managed by The Lancet, rejected both. The editor refused the changes and was dismissed in May 2010 (THE, May 2010). His successor took on what Elsevier called a "unique" form of peer review (Nature blogs, June 2010).

On the other things people search for: the journal page lists its indexing, including Index Medicus, EMBASE and Science Citation Index. In May 2026 it showed an Impact Factor of 0.8 and a CiteScore of 4.0; both change every year. The main page shows no acceptance rate; it listed instead 3 days from submission to first decision and 81 days to acceptance. Whether a paper there helps you depends on how your field reads it, so ask colleagues before you submit.

Which preprint servers accept hypothesis papers?

Preprints.org lists "Hypothesis" as a suitable article type; arXiv, bioRxiv and medRxiv do not post research proposals or hypotheses without new data. Here is what each server says.

ServerHypothesis without new data?What the policy says
arXivOnly as substantive research, not as a planMay decline "submissions that do not contain original or substantive research, including course projects, research proposals"; "proposals for future research" are typically not accepted
bioRxivNoNot intended for "Hypotheses without new data"
medRxivNoNot intended for "opinion pieces and hypotheses that lack data"
PsyArXivPossibly, as a theoretical contributionWork must be "either research or engages with research in psychology"; proposals that "do not constitute a complete contribution on their own" go in an OSF project
SSRNUnclear"SSRN accepts studies with rigorous methodology and original findings"; frameworks are typically not accepted
Preprints.orgYes"Hypothesis" is among the article types that are "generally suitable"
QeiosNot addressedPosts preprints after AI-assisted checks, gives every work a DOI and invites open, signed reviews

arXiv's test is whether the paper is "original or substantive research", not whether it has data: a worked-out theoretical argument may pass, a plan for future work will not. PsyArXiv checks the authors' publication history when the paper is "an opinion piece, commentary, or other similar work".

If what you have is a plan to test the idea yourself, PsyArXiv sends "preregistrations and study plans" to OSF, where preregistration means posting "a time-stamped, read-only version of your study plan" (OSF Support). RIO publishes research ideas and grant proposals as articles.

Either way, a preprint is not peer review: medRxiv calls its preprints "preliminary reports of work that have not been certified by peer review".

How do I get a DOI for a hypothesis without a journal?

Deposit it in a general repository: Zenodo, OSF and Figshare each mint a DOI without peer review, and each new version gets its own.

Zenodo (hosted by CERN) accepts "all types of research artifacts" in "any status", from "any stage of the research lifecycle" (policies). The first time you publish, it registers two DOIs: one for that version and one "representing all of the versions" (DOI versioning). You can change the files yourself only within 45 days of publishing, and the metadata "is always publicly accessible", even when the files are restricted (create new upload).

OSF says "DOIs are minted for preprints and registrations automatically when they are made public", and you can add one to a public project (OSF Support). Each new version of an OSF preprint gets "its own distinct Digital Object Identifier" (OSF Preprints). A registration is "a frozen version of your project that can never be edited or deleted".

Figshare makes an item "public with a DOI immediately" when you press Publish, with a free account (What is Figshare?). "Every Figshare item or collection version has a DOI", and the base DOI "always points to the latest public version" (versioning).

Octopus, Preprints.org, Qeios and RIO issue DOIs as well, and arXiv gives every article one. A DOI is a persistent link that makes a deposit findable and citable. It says nothing about quality: Zenodo states that all information is provided "as-is".

What should a hypothesis paper include?

A hypothesis paper should state one testable claim, show where it comes from and say what result would show it wrong. That last part has a long history. Karl Popper took falsifiability as the line between science and non-science: "if a theory is incompatible with possible empirical observations it is scientific", and corroboration should count only when it is "the positive result of a genuinely 'risky' prediction, which might conceivably have been false" (Stanford Encyclopedia of Philosophy). John Platt's "Strong Inference" (Science, 1964) made a method of it: devise rival hypotheses, then a "crucial experiment" whose possible outcomes would each exclude one or more of them. His test for any claim was a question: "But sir, what experiment could disprove your hypothesis?"

The venues above turn this into requirements. Taken together, a good hypothesis paper has:

  1. The claim, in one sentence. Medical Hypotheses: "The hypothesis needs to be set out in explicit detail." Two sentences may mean two hypotheses.
  2. Why it matters. Ideas in Ecology and Evolution asks how the idea "has significant potential to impact on the discipline" (submissions).
  3. What it builds on, and how it differs. Frontiers asks for "accurate presentation of previously posed hypotheses or theories"; Ideas in Ecology and Evolution asks you to "clearly differentiate the idea" from similar published ones.
  4. The evidence for and against. Medical Hypotheses expects "an evaluation of both evidence in support and evidence (apparently) against the hypothesis".
  5. Predictions. What should be observed if the hypothesis is true. Make at least one of them risky in Popper's sense: one that could plainly fail.
  6. What would show it wrong. Name the observation, and where you can, the size of effect that would count. This is Platt's question, answered in advance.
  7. A test someone could run. Ideas in Ecology and Evolution requires "a proposal for testing the idea, even if it is completely impractical with current technology". Say what to measure, in which sample or system, and against what comparison.
  8. Rivals and limits. Name the alternative explanations your test would tell apart, and where the hypothesis does not apply. Medical Hypotheses' guide asks for "the range of problems for which it may not be an appropriate solution".

For format, Frontiers asks for an abstract, an introduction, subsections and a discussion. Medical Hypotheses suggests an abstract, a short introduction, the hypothesis, its evaluation, and its consequences; the introduction "should not be an attempt to review the evidence in detail". If an AI tool helped you develop the idea, say so in the paper; our guide to AI disclosure shows how.

How do you cite a hypothesis or a preprint?

Cite the version a reader can find, with its DOI. In APA 7, a preprint or repository deposit names the server where a journal name would go, and a journal hypothesis article is cited like any other article. In the examples below, the authors, titles and DOIs are placeholders.

  • Preprint, following APA's own PsyArXiv example: Author, A. A., & Author, B. B. (2026). Title of the hypothesis paper. Preprints.org. https://doi.org/10.xxxx/xxxxx
  • Zenodo, OSF or Figshare deposit. APA says to "follow the same format for other preprint archives": Author, A. A. (2026). Title of the hypothesis paper. Zenodo. https://doi.org/10.5281/zenodo.xxxxxxx
  • Journal article. Author, A. A. (2026). Title of the hypothesis article. Journal Name, 12, Article 123456. https://doi.org/10.xxxx/xxxxx. Use the page range instead if the journal has pages; APA adds "Article" before an article number (journal article references).
  • An idea you heard but readers cannot find. Cite it in the text only, as a personal communication: (A. A. Author, personal communication, March 3, 2026). APA allows this "only when a recoverable source is not available", and it has no reference-list entry.

Versions matter for hypotheses, because the wording often changes. On Zenodo, cite the version DOI when your point depends on what one version said, and the all-versions DOI for the work as a whole. arXiv's DOI "will always point to the latest version", and a replacement does not get a new DOI (arXiv), so add the version number, such as v2, when it matters. Once a journal version appears, APA prefers it: "Ideally, use and cite the final, published version of a work."

Should you share a hypothesis before you can test it?

Often yes, if you want credit for the idea and want it tested. Weigh four things first.

Credit comes through citation. Ronald Vale and Anthony Hyman write that priority "becomes evident through a process in which scientists credit one another's work in papers and at meetings", and that it is "often most visible in the form of citations" (eLife, 2016). A dated, citable hypothesis gives others something to cite; seven famous hypotheses that waited for evidence show how that credit worked out in practice. But their criteria for disclosing a discovery begin with "all of the data along with a written interpretation of the data". Publishing a hypothesis can earn credit for the idea; credit for the evidence goes to whoever does the test.

Someone else may test it first, and that can be the point. Ideas in Ecology and Evolution describes itself as a catalogue from which modellers and empiricists can "'shop' for original ideas and hypotheses" that are "available to be explored, debated and tested". Octopus argues that if data later support your hypothesis, "you still get credit for the hypothesis as the person who published it", though in practice that depends on others citing you. If you would rather run the test yourself before anyone else sees the idea, an embargoed preregistration or a private dated record may suit you better; our guide to protecting a research idea compares those options.

Public usually means permanent. Preprints.org warns that "preprints cannot be completely removed once online". An Octopus publication cannot be deleted "unless there is a breach of our user terms", and an OSF registration "can never be edited or deleted". A hypothesis that turns out wrong stays on the record. That is normal in science: post a new version that says what changed and why.

Patents. A public hypothesis can count as a disclosure. Preprints.org is blunt: "If you intend to register a patent based on the work, do not post a preprint, as it will compromise your application." If the idea could be patentable, talk to your technology-transfer office before it goes anywhere public.

Where a Papex record fits

Papex is a record registry, not a journal or a preprint server. It does no peer review and mints no DOI, so if you need a DOI or journal credit, use one of the venues above. What it adds is a signed, versioned record of the hypothesis and of the evidence others bring to it, which you can keep alongside a DOI deposit or a journal article.

A Papex Hypothesis record has fields for most of the checklist: claim, why it matters, what would show it wrong, expected outcomes, how it could be tested, references, and "builds on" links to exact signed versions of other Papex records. It stays a private draft until you sign. The signature links you to that exact version, and only a salted fingerprint leaves Papex, timestamped with OpenTimestamps and anchored in Bitcoin, with no wallet or cryptocurrency needed. You then keep it private, share it with people you invite, or publish it. A later edit becomes a new version with its own signature and date; earlier versions stay as they were.

Each published version has a permanent address of the form papex.org/id/PXH-…/v/n, and its page gives a citation to copy, with BibTeX, RIS, CFF and JSON-LD under Other formats, as on Papex Labs' educational record PXH-0BADXVXY7QM3. The plain form reads: Author. (2026). Title. Papex Hypothesis Record PXH-…. https://papex.org/id/PXH-…/v/1.

On a public hypothesis, other signed-in people can add supporting evidence or counter-evidence, each with its source, and raise challenges. The page shows these as activity on the record, not as a verdict. A signature and timestamp show who signed which exact version and that it existed by a date; they are not peer review, and they do not show that the hypothesis is original or right. You can start a private draft at papex.org.

Sources

Keep reading

Pauli's neutrino letter and other hypotheses that waited for evidence

From Pauli's neutrino letter to the Higgs boson: how famous hypotheses were shared, how many years the evidence took and how each is credited today.

Note

How to cite ChatGPT, Claude AI and Gemini in APA, MLA and Chicago

APA 7, MLA 9, Chicago, AMA, IEEE and Harvard formats for ChatGPT, Claude and Gemini, from each guide's current advice, and when to disclose instead of cite.

Guide

How to write an AI disclosure statement for a research paper

AI disclosure statement examples and templates for research papers: what to include, where it goes, if grammar help counts, and 11 journal policies compared.

Guide