Free scoping read. A written assessment of your manuscript in two business days, at no cost.

Start one
Contents

Academy · Research Publication · Plagiarism: Types & Prevention

Understanding and Tackling Plagiarism in Academic and Research Settings: Categories and Editorial Process

The guidance of the Office of Research Integrity distinguishes several categories of plagiarism, including the less frequently discussed practice of self-plagiarism, or text recycling, and the COPE flowcharts provide editors with a specific, documented decision process for handling each category once it is suspected. This article describes those categories and the editorial process.

Pubrica AcademyWritten by a research integrity writer Reviewed bya publication ethics reviewer Published10 October 2026 Last checked10 October 2026 Reading time11 minutes Sources13 peer-reviewed

Plagiarism in research is the use of another person's words, ideas, data or results without adequate acknowledgment, presented as though they were the author's own original contribution. The U.S. Office of Research Integrity (ORI) and the Committee on Publication Ethics (COPE) both treat it as a serious breach of research integrity, but their guidance also distinguishes between categories that are often conflated in informal discussion: direct copying, paraphrasing without citation, and self-plagiarism, or text recycling (the reuse of substantial portions of one's own previously published text without disclosure), which raises ethical questions distinct from those raised by copying another person's work.

Plagiarism in research at a glance

Definition
Using another person’s words, ideas, data or results without adequate acknowledgment
Main categories
Direct (verbatim), paraphrasing, mosaic, and self-plagiarism or text recycling
How common
11% of 754 manuscripts submitted to one medical journal in 2009–2010 were confirmed as plagiarised12
In surveys
A pooled 1.7% of scientists admitted committing plagiarism; 30% knew of a colleague who had11
Share of retractions
9.8% of 2,047 retracted biomedical and life-science articles7
Reading time
About 11 minutes, with 2 figures and 2 tables

Learning objectives

  • The distinct categories of plagiarism recognised in research-integrity guidance, including self-plagiarism
  • Why self-plagiarism is treated as a genuine ethical issue even though no other author’s original work is copied
  • What the documented COPE process for handling an allegation of plagiarism involves

Common misconceptions

  • “Plagiarism refers only to copying another author’s words or data without citation.” Research-integrity guidance, including the definitions of the Office of Research Integrity and widely used frameworks such as Miguel Roig’s ORI-published guide on avoiding plagiarism and self-plagiarism, explicitly recognises self-plagiarism, namely the substantial reuse of one’s own previously published text, data or figures without disclosure, as a distinct ethical issue, separate from but related to plagiarism of another author’s work.

Categories of plagiarism in research-integrity guidance

Research-integrity guidance divides plagiarism into categories that call for different levels of scrutiny and different corrective actions. Direct or verbatim plagiarism involves copying another author's exact wording without quotation marks or citation; it is the most straightforward category to detect and the least ambiguous to adjudicate. Paraphrasing plagiarism, in which another author's ideas or findings are restated so closely in structure and substance that the specific intellectual contribution of the source is used without proper credit, even though the exact wording differs, is more difficult to detect automatically and requires editorial judgment about whether citation adequately credits the original source. Miguel Roig's widely used guide on avoiding plagiarism, originally developed for the Office of Research Integrity, adds a further distinction with which many researchers are less familiar: mosaic plagiarism, in which phrases and short passages from a source are interspersed with the author's own words without adequate attribution, dispersed throughout a text rather than concentrated in one obviously copied block; standard plagiarism-detection software can sometimes miss this pattern if the matching segments fall below its flagging threshold1.

The main categories of plagiarism, with illustrative examples

CategoryWhat it isIllustrative exampleHow it is usually detected
Direct (verbatim)Another author’s exact wording copied without quotation marks or citationA paragraph from a published Introduction pasted into a new manuscriptReliably flagged by similarity software as a long matching string
ParaphrasingAnother author’s ideas restated so closely in structure that their contribution is used without creditA source’s argument rewritten sentence by sentence with synonyms swapped in and no citationOften missed by software; needs a reader who knows the source
MosaicShort phrases from a source scattered among the author’s own words without attributionSix short phrases from one review spread across a Discussion sectionIndividual matches can fall below the flagging threshold; check the passage-level report
Self-plagiarism (text recycling)Reuse of one’s own published text, data or figures without disclosureMethods and Discussion text carried over from the author’s earlier paper with no note of the overlapFlagged like any other overlap; the editor judges disclosure and copyright
Duplicate publicationThe results of one study published more than once without disclosureThe same trial reported in two journals as if it were two studiesComparison of data, tables and author lists, often by reviewers or after publication

The examples in the table are illustrative and were written for this article.

Self-plagiarism as a genuine ethical issue

Self-plagiarism, also called text recycling, is the reuse of substantial portions of one's own previously published text, data presentation or figures in a new publication without disclosure of the overlap, and it raises a different ethical question from copying another person's work: no one's original intellectual contribution is misappropriated, but the practice can still mislead readers, journals and the broader scientific record. Roig's guidance identifies several specific concerns: self-plagiarized text can create the appearance of a new, independent contribution where none exists, inflating an author's apparent publication output; it can violate the copyright transfer agreement that an author signed with the original publisher, since copyright in the exact wording, and not merely in the ideas, typically transfers to the journal; and when self-plagiarism involves recycling the results of an entire study across multiple papers without disclosure, it risks the related but distinct problem of duplicate publication, which can distort meta-analyses and systematic reviews that treat each publication as an independent data point1. COPE's discussion documents explicitly address text recycling as a category that editors should be prepared to evaluate on its own terms, distinct from, though related to, plagiarism of another author's work2.

Too close: paraphrasing plagiarism

“Brief phrases copied from multiple sources may avoid detection since each match is too small to set off a flag.”

Same structure, same order of ideas and synonyms swapped in, with no citation. This uses the source’s contribution without credit.

Acceptable: own words, with credit

“Because similarity software only reports matches above a minimum length, copying spread thinly across a text may go unflagged (Source, year).”

The idea is explained in new wording and structure, and the source is cited.

Illustrative example, written for this article. The original sentence being paraphrased was: “Short phrases copied from several sources can escape detection because each match is too small to trigger a flag.”

Capabilities and limitations of similarity-detection tools

Automated similarity-detection software, such as Turnitin or iThenticate, compares a submitted text against large databases of previously published and submitted material and flags matching strings of text above a certain length; this makes it effective at detecting direct and verbatim copying but structurally limited with respect to paraphrasing and mosaic plagiarism, in which the matched strings are shorter or the wording has been altered sufficiently to fall below a simple string-matching threshold. The systematic review of academic plagiarism-detection research by Foltýnek, Meuschke, and Gipp found that most widely deployed commercial tools are built primarily around this kind of text-similarity matching and perform considerably less reliably in detecting paraphrased or idea-level plagiarism, translated plagiarism across languages, or plagiarism of structure and argument without matching wording, which are precisely the categories that require human editorial judgment rather than automated flagging for reliable detection4. This limitation directly justifies the graduated, human-reviewed COPE process discussed below, since treating a similarity score as a final verdict would miss serious cases that fall below the detection threshold of the tool while also risking false flags on properly cited material. It is also one of the strongest arguments for pairing automated screening with a human reader who examines the specific passages highlighted by the tool, rather than treating the single overall percentage score as a sufficient basis for a decision.

Studies of real submissions show why. When the Croatian Medical Journal ran every manuscript submitted in 2009–2010 through three detection programs and then checked the results by hand, the software flagged 105 of 754 manuscripts (14%), and manual verification confirmed plagiarism in 85 (11%)12. In a sample of 110 manuscripts submitted to the American Journal of Roentgenology in 2014, 12 (10.9%) contained plagiarism, yet 9 of those 12 had an overall similarity index below 20% once quotations and references were excluded, and 5 had not been detected at submission; two had already been accepted13.

From software flag to confirmed case: all 754 submissions to one journal, 2009–2010

ManuscriptsShare of 754Checked by
Flagged
105
14%
Software
Confirmed
85
11%
Editor
Others
63
8%
Editor
Self
22
3%
Editor

“Others” is plagiarism of another author’s work; “Self” is self-plagiarism. Bar lengths are proportional within each column.

Figure 1. Baždarić et al. (2012) screened every manuscript submitted to the Croatian Medical Journal in 2009–2010 with eTBLAST, CrossCheck and WCopyfind, treating more than 10% text from other sources as suspicious, then verified each flagged manuscript by hand. Not every flag was confirmed. Among self-plagiarised manuscripts, the copied-text rate was highest in the Materials and Methods section.12

Checklist: reading a similarity report

  • Open the source-by-source matches and the highlighted passages; do not stop at the overall percentage.
  • Look at the highest single-source match as well as the total: in the AJR sample, 7 of the 12 plagiarised manuscripts had a top single-source score below 10%.13
  • Check whether the main matching source is the authors’ own earlier work; in the same sample this was the case for 9 of the 12.13
  • Give the Methods section particular attention, which was a common site of duplicated text in both screening studies.1213
  • Confirm whether matched text is correctly quoted and cited, or is a reference list or standard phrasing, before treating it as a concern.
  • Read paraphrased passages against the source for close structure, which software cannot reliably detect.4
  • Record what was checked and why the overlap was judged acceptable or not, so the decision can be explained to the author.

Procedures following a suspicion of plagiarism

COPE publishes a specific flowchart for editors handling suspected plagiarism in a submitted or published manuscript, distinguishing the appropriate response based on factors including whether the manuscript is still under review or already published, and the apparent extent of the overlap2. For a submission still under review, the documented process generally involves the editor contacting the corresponding author for an explanation before reaching a decision, rather than rejecting the manuscript outright on first detection, since legitimate explanations do occur, such as appropriately quoted and cited material that a detection tool has flagged as a false positive. For material already published, COPE's guidance outlines a more formal process that can include contacting the author's institution, requesting a correction or retraction depending on severity, and, in confirmed serious cases, issuing a formal retraction notice explaining the reason to readers, in accordance with the same retraction-guideline principles that COPE applies to other forms of confirmed research misconduct3. The key feature of this documented process is that it is a graduated, evidence-gathering procedure rather than an automatic, single-step response triggered solely by a similarity-detection score.

1A similarity report or a reader flags overlap with an earlier source.A prompt, not a verdict
2The editor reads the flagged passages. Is the material properly quoted and cited, or a known false positive?Yes: no action
3Is the manuscript still under review? The editor contacts the corresponding author for an explanation before deciding.Explanation sought
4Is the paper already published? The editor may contact the author’s institution and, depending on severity, request a correction or issue a retraction notice.Correct or retract
Figure 2. A simplified outline of the graduated process in COPE’s guidance on suspected plagiarism. Consult the COPE flowcharts themselves for the full decision points.23

Variation in the application of plagiarism norms

Research-integrity scholarship also documents that attitudes toward, and training in, proper citation and paraphrasing practice vary across educational systems and disciplines, which bears on how allegations are investigated rather than on whether the underlying standard applies. The Handbook of Academic Integrity, edited by Bretag, brings together research documenting that writers trained in educational systems with less explicit instruction in citation conventions, or writing in a second language, can be disproportionately flagged by similarity-detection tools for practices (such as closely following the sentence structure of a source while translating or paraphrasing) that reflect a gap in formal training in citation norms rather than an intent to deceive5. This body of research does not argue that the underlying standard of properly crediting sources should differ by background; it argues instead that editorial and institutional investigation processes, including COPE's graduated approach, need to distinguish a training gap from deliberate misconduct before concluding which has occurred, since the corrective response (education or formal sanction) differs substantially between the two.

Takeaway for editors and supervisors

Establish first whether a flagged passage reflects a gap in training in citation conventions or an intent to deceive. The standard for crediting sources is the same for everyone, but the right response differs: teaching and revision for the first, a formal process for the second.

The relationship between plagiarism and the wider retraction record

Plagiarism is not an isolated category of misconduct; it is one of the specific, recurring causes identified in large-scale analyses of why papers are retracted after publication. Fang, Steen, and Casadevall's analysis of 2,047 retracted biomedical and life-science articles found that plagiarism accounted for approximately 10% of retractions in their sample, a meaningful share though smaller than that of fraud or duplicate publication, which confirms that journals and institutions treat plagiarism as serious enough to warrant full retraction in a documented, non-trivial share of cases rather than handling it only through less visible corrections7. This connection helps to explain why editorial bodies such as COPE maintain a formal, documented process specifically for plagiarism: the consequence of an unresolved or wrongly dismissed plagiarism case is not merely a corrected citation but a potential retraction, with all the effects on the citing literature that a retraction entails. It is also why journals increasingly conduct similarity screening as a routine step for every submission rather than only when an editor suspects a problem, treating the check as a standard quality-control step rather than an exceptional response reserved for cases that already appear suspicious. Incorporating a similarity check into the intake of every submission, rather than invoking it reactively, also protects authors by identifying an unintentional overlap or an improperly removed citation while it can still be corrected easily before publication, an outcome considerably less costly for all parties than a post-publication correction or retraction.

Survey and screening studies give a sense of scale. A meta-analysis of anonymous surveys found that far more scientists report knowing a colleague who has plagiarised than admit doing so themselves11, and journals that screen submissions systematically find plagiarism in roughly one in ten manuscripts.

What peer-reviewed studies have found about how common plagiarism is

StudyWhat was examinedMain finding
Baždarić et al., 201212All 754 manuscripts submitted to the Croatian Medical Journal in 2009–2010, screened by software and verified by hand85 (11%) confirmed as plagiarised: 63 (8%) plagiarism and 22 (3%) self-plagiarism
Taylor, 201713110 of 1,610 manuscripts submitted to the American Journal of Roentgenology in 2014, checked with CrossCheck and manually against ICMJE criteria12 (10.9%) contained plagiarism; source referencing was lacking in 11, and 5 had not been detected at submission
Pupovac and Fanelli, 201511Meta-analysis of 17 anonymous surveys asking scientists about plagiarismPooled 1.7% (95% CI 1.2–2.4) admitted committing plagiarism; 30% (95% CI 17–46) knew of a colleague who had
Fang et al., 20127All 2,047 biomedical and life-science articles indexed in PubMed as retracted on 3 May 2012Plagiarism caused 9.8% of retractions, against 43.4% for fraud or suspected fraud, 14.2% for duplicate publication and 21.3% for error

Key points

Key points

Plagiarism guidance distinguishes direct, paraphrasing and mosaic plagiarism from self-plagiarism, or text recycling, each of which raises a distinct ethical concern, and the documented COPE process for suspected cases is a graduated, evidence-gathering procedure rather than an automatic response to a similarity score.

Practical application

Practical application

  1. Previously published work by the same author should be checked before text, data or figures are reused. Where substantial portions of a prior publication are recycled, the overlap should be disclosed explicitly and it should be confirmed that the reuse does not breach the original copyright transfer agreement.
  2. A high similarity-detection score should be treated as a prompt for investigation rather than a verdict. In accordance with the graduated approach of COPE, it should be established whether flagged overlap consists of properly quoted and cited material before misconduct is assumed.
  3. Paraphrased passages should be checked specifically for mosaic plagiarism. Passages that closely follow the structure and phrasing of a source, even where individual words have been changed, should be reviewed, since this pattern can constitute plagiarism without exact verbatim copying.

Frequently asked questions

Research-integrity guidance treats it as a distinct but genuine ethical issue, since it does not involve misappropriating another person's original work but can still mislead readers about the novelty of a contribution, breach copyright transfer agreements and create duplicate-publication problems for evidence synthesis.

No; properly quoted and attributed material is not plagiarism by definition. The concern addressed by COPE's guidance is specifically unattributed or inadequately attributed overlap, which is why its editorial process includes verifying whether flagged material was in fact properly cited before concluding that misconduct occurred.

It is the practice of interspersing short phrases or passages, copied or closely paraphrased from a source, with the author's own original wording throughout a text, rather than copying one obvious continuous block, which makes the practice more difficult for both human readers and automated similarity-detection tools to detect.

No. COPE's documented process treats a similarity score as a prompt for editorial investigation, including contacting the author for an explanation, rather than an automatic basis for rejection or retraction; properly cited quotations, standard methodological phrasing, and reference lists can all contribute to similarity scores without constituting plagiarism.

Fang, Steen, and Casadevall's analysis of 2,047 retracted biomedical and life-science articles found that plagiarism accounted for roughly 10% of retractions in their sample, a smaller share than fraud or duplicate publication but still sufficient to confirm that plagiarism is treated as grounds for full retraction in a meaningful share of cases.

References

  1. Roig M. Avoiding Plagiarism, Self-Plagiarism, and Other Questionable Writing Practices: A Guide to Ethical Writing. U.S. Office of Research Integrity. read 10 October 2026. View ORI Guide to Ethical Writing
  2. Committee on Publication Ethics (COPE). Plagiarism and Text Recycling Discussion Document. read 10 October 2026. View COPE Discussion Document
  3. Committee on Publication Ethics (COPE). Guidance for Editors: Research, Audit and Service Evaluations and COPE Flowcharts. read 10 October 2026. View COPE Flowcharts
  4. Foltýnek T, Meuschke N, Gipp B. Academic Plagiarism Detection: A Systematic Literature Review. ACM Computing Surveys. 2019;52(6):112. read 10 October 2026. View on ACM Computing Surveys
  5. Bretag T, ed. Handbook of Academic Integrity. Springer; 2016. read 10 October 2026. View publisher page
  6. International Committee of Medical Journal Editors. Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. read 10 October 2026. View ICMJE Recommendations
  7. Fang FC, Steen RG, Casadevall A. Misconduct accounts for the majority of retracted scientific publications. Proceedings of the National Academy of Sciences. 2012;109(42):17028-17033. read 10 October 2026. View on PNAS
  8. Memon AR. Similarity and plagiarism in scholarly journal submissions: bringing clarity to the concept for authors, reviewers and editors. Journal of Korean Medical Science. 2020;35(27):e217. read 10 October 2026. View on Journal of Korean Medical Science
  9. Office of Research Integrity, U.S. Department of Health and Human Services. Definition of Research Misconduct. read 10 October 2026. View ORI Definition of Research Misconduct
  10. Committee on Publication Ethics (COPE). Core Practices. read 10 October 2026. View COPE Core Practices
  11. Pupovac V, Fanelli D. Scientists admitting to plagiarism: a meta-analysis of surveys. Science and Engineering Ethics. 2015;21(5):1331-1352. read 13 November 2026. View on Science and Engineering Ethics
  12. Baždarić K, Bilić-Zulle L, Brumini G, Petrovečki M. Prevalence of plagiarism in recent submissions to the Croatian Medical Journal. Science and Engineering Ethics. 2012;18(2):223-239. read 13 November 2026. View on Science and Engineering Ethics
  13. Taylor DB. Plagiarism in manuscripts submitted to the AJR: development of an optimal screening algorithm and management pathways. American Journal of Roentgenology. 2017;208(4):712-720. read 13 November 2026. View on American Journal of Roentgenology

Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.

How to cite this article

Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.

Pubrica Academy. (2026, October 10). Understanding and Tackling Plagiarism in Academic and Research Settings: Categories and Editorial Process. Pubrica. https://pubrica.com/academy/study-guide/understanding-and-tackling-plagiarism-in-academic-and-research-settings/

This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.

Plagiarism Self-Check Before Submission

This checklist should be completed before submitting any manuscript built on prior drafts, sources or the author’s own earlier publications.

[ ] Checked all paraphrased passages for close structural similarity to their source, not just wording
[ ] Checked for mosaic patterns: short unattributed phrases scattered through otherwise original text
[ ] Checked whether any text, data, or figures are reused from your own prior publications
[ ] Disclosed any self-reuse and confirmed it does not breach a prior copyright transfer agreement
[ ] Confirmed every quotation is marked and cited, and every paraphrase is cited to its source

Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.

How Pubrica can help

Pubrica’s publication ethics team conducts similarity checks alongside editorial review to distinguish properly attributed overlap from genuine risk of plagiarism, including the less frequently checked categories of mosaic plagiarism and self-plagiarism, before a manuscript is submitted.

Find something else

Search the Pubrica Academy

Access is free and requires no account. Enter a topic, or open the full library to filter by category, sub-topic or resource type.

Checks for categories of plagiarism that a basic similarity scan may miss

A scoping assessment is free and takes two business days. It provides a written assessment of what the work requires and which specialist would undertake it.

Get a free quote

Regulations

Check your project against this rule

  1. 1Your work
  2. 2Your details
  3. 3Send
What do you need?

Material you send is held under signed confidentiality, with access limited to the assigned team and encrypted transfer. It is not used to train anything and is not passed to third-party tools. Ask for the agreement before you send anything you would rather cover first.