CASRAI Dictionary

Tag: research misconduct news

  • ORI Second Phase Guidance Documents: 2026 Update for US Institutions

    The ORI second phase guidance documents are a batch of topic-specific PDFs — covering honest error, admissions, pursuing leads, sub-awardee assurances and institutional assessments — that the US Office of Research Integrity released from September 2025 onward to help Public Health Service (PHS)-funded institutions implement the 2024 Final Rule on research misconduct (42 CFR Part 93) ahead of its January 1, 2026 applicable date. Institutions now need to fold this guidance into formal policy before their next Annual Report cycle.

    The Office of Research Integrity (ORI) is the component of the US Department of Health and Human Services (HHS) responsible for overseeing research-misconduct policy and compliance at institutions that receive Public Health Service funding, including funding from the National Institutes of Health (NIH).

    What Is ORI’s Second Phase of Guidance Documents?

    ORI began releasing guidance in June 2025 to support institutions implementing the 2024 Final Rule on Public Health Service Policies on Research Misconduct, published in the Federal Register at 42 CFR Part 93. The first tranche — Writing Policies and Procedures, Sample Policies and Procedures, Small Institution Guidance, and Implementation Guidance — gave institutions templates for compliant policy language.

    The second phase, announced by ORI on 15 September 2025, is topic-specific rather than template-based. It clarifies how institutions must apply the Final Rule to real proceedings: Honest Error, Admissions, Pursuing Leads, Sub-Awardee Assurances, and Assessments. ORI has continued releasing additional topic guidance in further batches through May 2026, all falling under the same phased rollout of Final Rule guidance.

    The Final Rule’s effective date was 1 January 2025 (optional use, by mutual agreement of the parties to a case), and its applicable date was 1 January 2026, after which every PHS-funded institution must use the updated regulation for any new research-misconduct matter.

    What Do the Phased Guidance Documents Cover?

    Across the rollout, ORI has grouped guidance into consistent topic clusters. The table below maps each release window to its documents and subject matter, based on ORI’s official guidance-documents index.

    Release window Guidance documents Institutional focus
    June 2025 (Phase 1) Writing Policies and Procedures; Sample Policies and Procedures; Small Institution Guidance; Implementation Guidance Drafting Final Rule-compliant policy language
    September 2025 (Phase 2) Honest Error; Admissions; Pursuing Leads; Sub-Awardee Assurances; Assessments Distinguishing error from misconduct; handling admissions and sub-awardees
    December 2025 Institutional Record; Research Records; Multiple Institutions Compiling the institutional record; multi-site cases
    March 2026 Confidentiality; Interviews; Subsequent Use Exception Disclosure limits, interview conduct, record-sharing exceptions
    May 2026 State of Mind; Respondents; Institutional Record Best Practices Assessing intent; managing multiple respondents; record organisation

    Two details are easy to miss. First, ORI’s own guidance notes that some documents cross-reference other guidance not yet released at the time of publication — institutions should treat the rollout as a living body of interpretation, not a closed set. Second, the Sub-Awardee Assurances Guidance confirms that every recipient of PHS support, including sub-awardees on collaborative grants, must independently establish and maintain an active research-integrity assurance with ORI — a requirement that is easy to overlook on multi-institution NIH research misconduct cases.

    What Must Institutions Update in Annual Report Submissions?

    Every PHS-funded institution files an Annual Report with ORI: a record of research-misconduct activity from the prior year, combined with an annual assurance that the institution’s policies and procedures comply with 42 CFR Part 93. ORI required institutions to submit an updated assurance reflecting the 2024 Final Rule as part of the 2025 Annual Report, due 30 April 2026 — a deadline that has now passed for most institutions.

    That does not close the compliance question. Guidance covering State of Mind, Respondents and Institutional Record Best Practices was not released until May 2026, after the April assurance deadline. Institutions therefore have unfinished work: policies and procedures need to be revised now, in the second half of 2026, so the underlying documents — not just the assurance statement — genuinely reflect the completed guidance set before the 2026 Annual Report is due on 30 April 2027. Practical year-end priorities include:

    • Revising written policies to incorporate the Honest Error, State of Mind and Admissions frameworks, so investigative committees apply a consistent intent standard.
    • Updating institutional-record procedures against the Institutional Record and Research Records guidance, including retention and compilation practices.
    • Confirming that every sub-awardee on active PHS-funded collaborations holds its own current ORI assurance.
    • Documenting confidentiality and interview procedures consistent with the March 2026 guidance before any new proceeding opens.

    How Does ORI’s Jurisdiction Fit Under Federal Misconduct Rules?

    ORI’s authority is defined by 42 CFR Part 93, which applies to institutions receiving PHS funding — chiefly through the NIH, but also the CDC, FDA and other HHS agencies. Research misconduct according to federal regulations is defined as fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results, and explicitly excludes honest error or legitimate differences of opinion.

    ORI’s own jurisdiction is narrower than institutions sometimes assume: it oversees the institution’s handling of an allegation and can make findings and impose administrative actions, but the initial inquiry and investigation remain the responsibility of the institution itself. This is why the phased guidance — on assessments, interviews, pursuing leads and the institutional record — matters so much: it is ORI’s interpretation of how institutions must run proceedings that ORI does not run directly.

    Common Questions on the Second-Phase Guidance

    What is ORI’s second phase of guidance documents?

    It is a set of topic-specific guidance PDFs — Honest Error, Admissions, Pursuing Leads, Sub-Awardee Assurances and Assessments — that ORI published starting 15 September 2025 to clarify how institutions must apply the 2024 Final Rule (42 CFR Part 93) during actual research-misconduct proceedings.

    What is the Office of Research Integrity’s jurisdiction?

    ORI oversees research-misconduct policy compliance and case findings at institutions receiving Public Health Service funding, including NIH research misconduct matters. Institutions conduct the actual inquiry and investigation; ORI reviews findings and can take administrative action.

    What counts as research misconduct according to federal regulations?

    Under 42 CFR Part 93, research misconduct means fabrication, falsification, or plagiarism in proposing, performing or reviewing research, or in reporting results. Honest error and genuine scientific disagreement are expressly excluded from the definition.

    When is the next ORI Annual Report due?

    Institutions submitted their updated assurance with the 2025 Annual Report by 30 April 2026. The next Annual Report, covering 2026 activity, is due 30 April 2027, giving institutions the remainder of 2026 to align policies with all released guidance.

    Implications for Research Integrity Offices

    For research integrity officers and research administration offices, the practical task has shifted from watching for new PDFs to closing the gap between assurance language and actual policy text. An assurance filed in April 2026 cannot have anticipated May 2026 guidance on State of Mind and Respondents; institutions that treat the assurance as a one-time filing, rather than a standing commitment to keep procedures current, risk a mismatch that surfaces during ORI compliance review.

    The rollout also has downstream implications beyond individual institutions: consistent application of Honest Error and State of Mind standards across hundreds of PHS-funded institutions should, over time, produce more comparable case outcomes and a clearer public record — a goal shared by research-integrity bodies and standards organisations working alongside CASRAI’s originated frameworks for research accountability. Institutions that update procedures methodically now, rather than waiting for the next assurance deadline, will be better positioned when ORI closes remaining cross-references in future releases.

  • Vietnam Research Integrity Crackdown: New Enforcement System

    Vietnam’s Ministry of Science and Technology introduced a binding national framework on 25 May 2026 that replaces voluntary integrity guidance with tiered sanctions – written warnings, retractions, funding claw-backs and permanent project bans – logged on a new National Digital Platform. The Vietnam research integrity crackdown is notable less for its penalty ladder than for what it reveals: a research system that expanded output rapidly is now trying to build enforcement infrastructure – a registry, an investigative chain of command, funder linkage – in a single directive, rather than over the decades it took comparable systems elsewhere.

    Research misconduct enforcement infrastructure is the set of institutions, registries and procedural rules a research system uses to detect, investigate, sanction and publicly record integrity violations. Vietnam’s 25 May 2026 directive is an attempt to construct that infrastructure from a near-standing start.

    What did the Ministry of Science and Technology announce?

    Vietnam’s Ministry of Science and Technology published guidance on 25 May 2026 requiring every science and technology organisation in the country to adopt formal rules against research misconduct and follow a defined investigation-and-sanction process. Before the directive, integrity expectations existed mainly as general principles promoted on a voluntary basis, without a consistent enforcement mechanism across institutions.

    The framework names four categories as the most serious violations: fabricating data, plagiarising others’ work, concealing conflicts of interest, and acts that distort the true nature of research. It also addresses generative-AI misuse directly, classing fabricated AI data or images, and citing unverified AI-generated material as a reference, as sanctionable conduct. Confirmed violations must be logged on the National Digital Platform for Science, Technology and Innovation Management, creating a permanent, searchable record.

    How is Vietnam building enforcement infrastructure from scratch?

    Three infrastructure elements distinguish this directive from a simple penalty list. First, a central registry: the National Digital Platform gives Vietnam a national record of confirmed violations, a component many systems only add after years of institution-only files. Second, a distributed investigative chain: heads of individual research organisations, not a single national office, are charged with running periodic inspections, monitoring retraction patterns, and operating complaint-intake procedures. Third, preventive obligations built in at launch – mandatory plagiarism checks before submission, retained raw data and research logs, and disclosure of funding, conflicts of interest and AI use.

    Notably, this formal apparatus is arriving after an informal one had already taken root. Tu Van Duong, a senior researcher at Purdue University, founded a Facebook group dedicated to scientific integrity in Vietnam that has grown to more than 300,000 members, according to Retraction Watch. Duong described the Ministry’s directive as an “important milestone” that converts principles his community had already been discussing informally into “concrete mandates” and a “binding obligation.” Vietnam’s enforcement infrastructure is therefore being built on two tracks at once – a bottom-up integrity community that predates regulation, and a top-down registry that formalises it.

    Vietnam is not the only fast-publishing system doing this in parallel. Daniel Barr, principal research integrity advisor at RMIT University in Australia, has pointed to Thailand’s Research Integrity Network (THRIN) as a comparable case: a recently established cross-institutional body connecting academics, experts and administrators rather than a single statutory regulator. Barr argues that principles-based frameworks like Vietnam’s and Thailand’s are well suited to systems where research spans many disciplines and institutional types – but only if the framework also supports responsible practice, not just punishment.

    How does Vietnam’s build-out compare with established systems?

    Judged purely on infrastructure components rather than penalty severity, Vietnam’s May 2026 framework is compressing into one directive what other systems built over one to three decades.

    System Formal infrastructure dates to Central public registry Investigative model Funder linkage Explicit AI-misuse clause
    United States (ORI) 1989-1992 (PHS oversight bodies consolidated into ORI) Yes – published case summaries Institutional inquiry, ORI oversight of PHS-funded research Federal funding debarment Added later, via updated guidance
    Germany (DFG) 1998-1999 (Rules of Good Scientific Practice, Ombudsman system) Partial – sanctions list Institutional ombudspersons, DFG appeal panel 1-8 year funding exclusion Added later, via updated guidance
    United Kingdom (UKRIO / Concordat) 2006 (UKRIO) / 2012 (Concordat, revised 2019) No – institution-level records only Institutional investigation under UKRI’s Good Research Practice policy UKRI can require investigation as a funding condition Under active revision (2025 UKRI guidance)
    Thailand (THRIN) Recently established No – network coordination role Cross-institutional coordination, not a single regulator Indirect, via member institutions Not yet standardised
    Vietnam (MST framework) 25 May 2026 Yes – National Digital Platform, built at launch Distributed – heads of individual organisations mandated to investigate Funding return and project bans Yes – built in from day one

    Two things stand out. Vietnam is the only system in this comparison to launch a searchable national registry and an explicit AI-misuse clause simultaneously, rather than retrofitting either years after the founding framework. It is also the only one relying entirely on a distributed model – individual institution heads, not a national investigative office – to do the actual case work, which is precisely the part Tu Van Duong warns could leave the framework “merely existing on paper” if implementation and monitoring are not rigorous.

    Answer-first questions on research misconduct

    What are the consequences of misconduct in research?

    Confirmed misconduct can trigger correction or retraction of the publication, loss of funding eligibility, employment termination, and degree revocation. Beyond the individual, it damages institutional credibility, undermines public trust in the literature, and can harm the colleagues and whistleblowers who rely on or expose the affected work.

    What is research misconduct UK?

    UK Research and Innovation defines research misconduct as fabrication, falsification, plagiarism, or other serious deviations from accepted practice in proposing, performing, reviewing or reporting research. UK institutions investigate under their own procedures, guided by the Concordat to Support Research Integrity and UKRI’s Good Research Practice policy.

    What are the 5 unethical practices in research?

    Commonly cited categories are falsification of data, failure to credit others’ contributions, plagiarism, undisclosed conflicts of interest, and biased design or interpretation driven by outside influence. Vietnam’s framework converges on a narrower set of four “most serious” categories, adding concealment of AI use as a fifth, newer concern.

    How serious is misconduct?

    Severity ranges widely: a first procedural lapse may draw only a written warning, while confirmed fabrication or plagiarism can end a career through funding debarment, degree revocation or criminal referral for misuse of public funds. Vietnam’s tiered structure, like the US and German models, scales the response to the offence rather than applying one blanket punishment.

    Implications for institutions – and what happens next

    For research administrators outside Vietnam, the framework is a live test of a proposition many funders are already testing: that integrity infrastructure can be launched all at once, rather than assembled piecemeal. Three things follow.

    • Distributed enforcement needs central audit, not just central logging. A national registry only deters misconduct if institution-level investigations feeding it are consistent; without a body auditing how heads of organisations conduct inspections, the registry risks recording only what gets reported.
    • AI-misuse clauses written in at launch avoid a costly retrofit. The US, German and UK systems all had to amend decades-old frameworks to address generative AI; Vietnam’s framework, and any new system built after 2026, can reasonably be expected to include equivalent language from day one.
    • Grassroots integrity communities are an infrastructure asset. Duong’s 300,000-member group functioned as an informal reporting and awareness network before the Ministry’s framework existed. Institutions building their own research administration policies can treat existing communities of practice as a resource for compliance culture, not just a symptom of prior gaps.

    Vietnam’s National Digital Platform will take time to populate, and consistency across thousands of institutions and provincial science agencies is untested. The most useful signal over the next year will not be the penalty ladder itself but whether cases are actually investigated, logged and made visible – the same accountability gap that has historically limited less-centralised systems. Definitions of the underlying terms – fabrication, falsification, plagiarism, retraction – are tracked in CASRAI’s research integrity dictionary for institutions mapping their own policy language against Vietnam’s.

  • Is Retraction Watch Credible? Methodology Review

    Is Retraction Watch credible? Yes, with qualifications: independent media evaluators rate it “Pro-Science” with high factual reporting, its retraction database is now co-stewarded with Crossref, and it is cited in hundreds of peer-reviewed studies — but a 2025 comparative study flagged metadata error rates, and critics have challenged its editorial judgement and past partnerships. Research offices can cite it as a lead indicator, not a sole source of truth.

    Retraction Watch is a nonprofit blog and database, launched in August 2010 by science journalists Ivan Oransky and Adam Marcus, that tracks and reports on retractions of published scientific papers. It is operated by the Center for Scientific Integrity (CSI), a US 501(c)(3) nonprofit.

    How is the Retraction Watch database compiled?

    The Retraction Watch Database is built through routine, manual monitoring of PubMed, Web of Science, publisher and journal websites, plus reader tips, rather than automated harvesting from a single authoritative feed. It reached 20,000 entries in October 2019 and had grown to over 50,000 entries by January 2024, according to Wikipedia’s sourced timeline.

    In September 2023, Crossref acquired the Retraction Watch Database and committed to opening it fully to the scholarly community, processing and presenting the underlying data alongside its own DOI metadata infrastructure. This shifted the database from a single-organisation project to a shared piece of scholarly-record infrastructure, with Crossref handling data engineering and Retraction Watch continuing curation and reporting.

    A related tool, the Retraction Watch Hijacked Journal Checker, was added in 2022 in partnership with political scientist Anna Abalkina, using methods such as duplicated archives, identical website templates and anomalous citation growth to flag cloned or hijacked journal domains.

    Who funds Retraction Watch?

    Funding is philanthropic and mixed, not subscription-driven, which matters for assessing independence. Retraction Watch’s own FAQ discloses:

    • A $400,000 grant from the John D. and Catherine T. MacArthur Foundation (first major grant, 2014), which funded construction of the database.
    • Multi-year grants from the Laura and John Arnold Foundation and the Leona M. and Harry B. Helmsley Trust (2015–2017).
    • An ongoing WoodNext Foundation grant since 2022 (including a $250,000 award announced in December 2022).
    • Service fees paid as part of the Crossref acquisition, which now sustain the database itself.
    • Individual donations and, since 2025, an Open Philanthropy grant funding CSI’s Medical Evidence Project, plus a “Sleuths in Residence” programme and the Elisabeth Bik Science Integrity Fund.

    None of these funders are publishers, journals or institutions under active Retraction Watch scrutiny, which reduces the most obvious conflict-of-interest risk — but the reliance on a small number of foundation grants and individual mega-donors means funding continuity is not guaranteed in the way a subscription-funded index like Web of Science or Scopus is.

    What do critics say about scope and accuracy?

    Media Bias/Fact Check rates Retraction Watch “Pro-Science” with “High factual reporting” based on its sourcing standards and clean fact-check record — a credibility signal independent of the site’s own claims. Scientific American, Science and the Chronicle of Higher Education have all treated it as authoritative reporting on research misconduct.

    Even so, three distinct lines of criticism recur in the literature and commentary:

    • Metadata accuracy. A 2025 comparative study of retraction metadata found that, for a small sample of retracted articles, the Retraction Watch Database had a higher rate of metadata errors than PubMed and Web of Science, and recommended cross-checking retraction status across multiple sources rather than relying on any single database.
    • Scope and framing. Because the database and blog focus specifically on retractions, some commentators argue this can distort public perception of how common outright misconduct is, since not every retraction indicates fraud (many are for honest error, duplication or authorship disputes) and the coverage skews toward life-science journals that founders Oransky and Marcus know best.
    • Editorial judgement. Researcher Jaime A. Teixeira da Silva has published pointed critiques accusing Retraction Watch of “moral superior arrogance” and inconsistent standards, and has specifically criticised its past support for Jeffrey Beall’s “predatory publisher” list, which was itself controversial and discontinued.

    Retraction Watch has also publicly retracted its own 2013 advice on reporting misconduct after readers pointed out flaws — a transparency move that cuts both ways: it demonstrates a correction culture, but it also confirms that even the watchdog has published guidance it later deemed wrong.

    What Retraction Watch is, and is not, a reliable source for
    Use case Reliability Notes
    Flagging that a specific paper has a retraction notice High Cross-check against the journal’s own erratum/retraction page and Crossref metadata
    Tracking trends in retraction volume over time Moderate Reflects detection and reporting effort as much as underlying misconduct rates
    Ranking authors or institutions via the “leaderboard” Use with caution Author-level counts can conflate honest-error corrections with proven fraud; always read the underlying case notes
    Sole evidentiary basis for a formal misconduct finding Not appropriate Retraction Watch is a journalism and data project, not an investigative body with due-process authority

    Answer-first Q&A

    Who funds Retraction Watch?

    Retraction Watch is funded by philanthropic grants — historically the MacArthur Foundation, Arnold Foundation and Helmsley Trust, currently the WoodNext Foundation — plus service fees from its 2023 database acquisition by Crossref and individual donations, not by subscriptions, advertising or the publishers it covers.

    Do retracted studies still get cited?

    Yes. Research on “citation pollution” during the COVID-19 pandemic and elsewhere found that retracted papers continue to be cited — often uncritically — for years after retraction, because citation databases and reference managers do not always surface retraction status prominently, which is a core reason Retraction Watch and Crossref now integrate retraction flags into shared metadata.

    Is retraction good or bad?

    Retraction is generally a sign the self-correcting mechanism of science is working, not proof of malicious intent. Most retractions stem from honest error, duplication or authorship disputes rather than fraud; treating every retraction as a scandal — rather than as the scientific record being corrected — misreads what the mechanism is for.

    What is the most trustworthy source of scientific information for retraction status?

    No single source is definitive. Best practice, echoed by the 2025 metadata-accuracy study, is to cross-check the Retraction Watch Database against the publisher’s own retraction notice, Crossref metadata and PubMed before citing retraction status in a policy document, grant decision or institutional finding.

    How should research offices cite Retraction Watch in policy?

    Research-integrity offices, ORCID-linked institutional systems and funder compliance teams increasingly use Retraction Watch as a screening signal, not a determination of fact. Appropriate use in policy documents includes citing it to justify further review (e.g., “this output appears in the Retraction Watch Database and warrants verification against the publisher record before renewal decisions”), and inappropriate use includes citing a Retraction Watch leaderboard entry as standalone evidence of misconduct in a disciplinary or funding-eligibility decision.

    This distinction matters more, not less, now that Retraction Watch data flows through Crossref’s shared infrastructure: institutions integrating retraction checks into publication or grant workflows should document that they cross-reference at least one other source, consistent with COPE guidance on handling suspected research misconduct and with ICMJE recommendations on correcting the literature.

    Frameworks that formally define contributor roles and provenance — such as the CRediT taxonomy, which CASRAI originated in 2014 and which NISO now stewards as ANSI/NISO Z39.104-2022 — are a useful complement here, because clear role attribution makes it easier to establish who is accountable for a retracted claim rather than relying on retraction counts alone.

    What this means going forward

    Retraction Watch’s trajectory — from a two-person blog in 2010 to a Crossref-integrated data source cited in scores of peer-reviewed studies — reflects a broader shift toward treating retraction status as structured, machine-readable scholarly metadata rather than a reputational footnote. That shift raises the stakes on accuracy: as more institutional workflows query the database programmatically via Crossref, the metadata-error findings from 2025 become more consequential than when the site was read manually by journalists.

    For institutions, the practical takeaway is straightforward: treat Retraction Watch as a high-value discovery layer with a documented, disclosed funding model and a strong independent factual-reporting rating, while building cross-verification against publisher and Crossref records into any policy or compliance process that relies on it.