OAC 2025 Professional common mistakes
OAC 2025 Professional · Understand the traps and how to avoid them.
Treating 'open access' and 'open science' as synonyms, so the candidate answers governance questions as if OA were the whole of open science and misses the wider components (open data, open methods, open peer review, citizen science, reproducibility).
The two terms are used interchangeably in advocacy material and news coverage, so candidates memorise a single definition and never build the layered model (open science as the umbrella, open access as one pillar). Exam stems that ask about UNESCO's open science concept or about national policy frameworks then get answered with a publishing-only definition.
What to do instead
Build a two-level mental map: open science = umbrella covering open access, open data, open source, open methods, open educational resources, citizen science and reproducibility; open access = free, unrestricted online availability of scholarly outputs with reuse rights. Practise by writing a one-sentence definition of each plus three examples of activities that belong only to open science, and check policy texts (UNESCO 2021 Recommendation, national OA strategies) for which level they operate at.
Describing 'gold' and 'hybrid' as the same model, or calling any article in a subscription journal 'green' because the author posted it, so model-classification questions get the wrong label.
The vocabulary is overlapping and historically inconsistent (gold has meant both 'fully OA journal' and 'paid OA article'), and candidates learn labels from a single blog post rather than from the dimensions that actually distinguish models: where the version of record lives, who pays, and when the article becomes free.
What to do instead
Classify every model on four axes: (a) venue type (fully OA journal vs subscription journal), (b) payment route (APC, no fee, subsidy, transformative agreement), (c) version and location (version of record vs accepted manuscript in a repository), (d) timing (immediate vs embargo). Then place gold, hybrid, diamond/platinum, green, bronze and transformative agreements on those axes and drill with 20 short scenarios until each label is assigned from the axes, not from memory.
Assuming any Creative Commons licence permits commercial reuse, or that CC BY-NC-ND allows text and data mining and repository deposit without restriction, leading to wrong answers on reuse and compliance scenarios.
Candidates remember 'CC means open' and skip the licence-element table (BY, SA, NC, ND), so they cannot reason about the six licences or about what 'open' means under the Budapest/Berlin declarations and the Open Definition, which require commercial reuse and derivative works.
What to do instead
Memorise the four elements and the six combinations, then practise classifying activities as allowed or not for each licence (commercial republication, translation, remixing, TDM, repository deposit, course pack use). Explicitly note that CC BY and CC BY-SA meet the open definition, that NC and ND restrict openness, and that ND blocks adapted versions. Also separate author-rights questions (what the author retained under a publishing agreement) from licence questions (what the reader may do).
Treating a DOI as proof that a record is preserved and citable, or depositing metadata in a repository without a persistent identifier strategy, so questions on PIDs and infrastructure get answered with 'the DOI handles it'.
DOIs are the most visible PID and candidates generalise from them, missing that persistence depends on the registration agency and the committing organisation, that other PIDs (ORCID, ROR, RAID, IGSN, ARK, Handle) identify different entities, and that metadata must be deposited with the identifier to be useful.
What to do instead
Learn a PID-to-entity table: ORCID = person, ROR = research organisation, DOI = output or dataset (via DataCite/Crossref), RAID = project, IGSN = physical sample, ARK/Handle = general digital objects. For each, note who commits to persistence and what metadata is mandatory. Practise diagnosing broken-link scenarios by asking which identifier type is missing or which agency commitment failed, rather than defaulting to DOI.
Confusing FAIR with open, answering that FAIR data must be publicly accessible, or that making data open automatically makes it FAIR.
The acronyms are taught together in RDM training and the 'A' in FAIR (Accessible) is misread as 'open access'. Candidates then fail questions about restricted or sensitive data that can still be FAIR, and about open data that lacks metadata and therefore is not Findable or Reusable.
What to do instead
Anchor the distinction in one sentence: FAIR is about machine-actionable findability, accessibility under stated conditions, interoperability and reuse; openness is about who may access. Work through cases: sensitive patient data under controlled access (FAIR, not open), a CSV on a personal website with no metadata (open, not FAIR). Learn the FAIR sub-criteria and the role of metadata, persistent identifiers and licences in each.
Answering research-assessment questions with journal-level indicators (JIF, quartile) when the scenario asks for responsible evaluation, or treating DORA, Leiden Manifesto and CoARA as interchangeable statements.
Candidates' own academic experience is saturated with journal metrics, and the reform documents are summarised in training as 'don't use the impact factor', so the specific commitments (DORA: no journal-based metrics in hiring/promotion; Leiden: ten principles including field normalisation and no inappropriate composite indicators; CoARA: institutional reform agreements and narrative CVs) blur together.
What to do instead
Make a comparison table with columns: document, year, who signs, core commitment, what it explicitly rejects, and what it recommends instead. Then practise mapping exam scenarios to the specific document and to the concrete alternative (article-level metrics, narrative CV, peer review of outputs, societal impact evidence). Be able to name at least two alternatives to the JIF and explain why field normalisation matters.
Giving a blanket answer on embargoes and deposit ('green OA is always allowed after 12 months') instead of checking the specific funder or institutional policy terms in the scenario.
Many candidates learn one well-known policy (for example a 12-month STEM embargo) and apply it universally, ignoring that funders differ in embargo length, version allowed, licence requirements and repository location, and that institutional and funder mandates can both apply with the stricter one governing.
What to do instead
Practise a policy-comparison drill: for each major mandate type, record (a) what must be deposited, (b) which version, (c) maximum embargo, (d) licence required, (e) where, (f) when. Then work composite scenarios where a funder and an institution both apply and decide which requirement binds. Always answer from the stated policy in the stem, and flag in your notes that mandates change, so the method is 'read the terms', not 'recall the number'.
Designing advocacy or training answers around 'explain the benefits of OA' only, without segmenting the audience, diagnosing barriers, or specifying a measurable outcome, so the answer reads as generic promotion.
Advocacy is often encountered as campaigning material, and candidates reproduce the same benefits list for researchers, librarians, administrators and policymakers. Exam scenarios, however, reward audience analysis, behaviour-change framing and evaluation of the intervention.
What to do instead
Use a planning template for every advocacy scenario: audience and their incentives, specific barrier (time, confusion about rights, APC budget, fear of quality, discipline norms), intervention matched to that barrier (workshop, rights-check service, repository deposit workflow, transformative agreement guidance), channel, and success measure. Practise producing two different 150-word plans for a researcher versus a faculty dean on the same topic, and state the evaluation indicator for each.
Treating equity as a single 'developing countries are exempt from APCs' fact, and missing other dimensions such as language, disability, early-career and precariously employed researchers, indigenous knowledge, and the equity effects of read-and-publish agreements.
The dominant equity narrative in OA news is the APC waiver, so candidates stop there. They then cannot answer questions about who is excluded by English-language dominance, by APC-based models even with waivers (authors without institutional affiliation), or by transformative agreements that cover only participating institutions.
What to do instead
Build an equity lens checklist: geography and income, language, disability and accessibility standards, career stage and employment security, discipline funding levels, gender and care responsibilities, indigenous and local knowledge. For each OA model (APC gold, diamond, green, transformative agreement), write who benefits and who is excluded. Use this checklist to answer scenario questions and to critique policy statements rather than reciting the waiver fact.
Mixing up the roles of repositories and publishers in the scholarly record - for example saying a preprint server provides peer review, or that depositing in an institutional repository counts as publication - so infrastructure and workflow questions get the wrong actor.
The ecosystem is described in shorthand ('post it in a repository'), and candidates collapse distinct functions: registration, certification, dissemination and preservation (the four functions of the scholarly record). Without that functional model, preprint servers, repositories, overlay journals and publishers all blur together.
What to do instead
Learn the four functions (registration, certification/peer review, dissemination, archiving/preservation) and map each actor to the functions it performs: preprint server = registration and dissemination; overlay journal = certification over preprints; institutional repository = dissemination and often preservation; publisher = all four. Then answer infrastructure questions by asking which function the scenario needs and which actor provides it, and note preservation commitments (CLOCKSS, Portico, LOCKSS) separately.