A reviewer asks you to add a batch of references, most from one research group. Here is how to tell a fair suggestion from coercion, and what to do next.
What counts as citation manipulation
A reference tells readers where an idea, method, or finding came from. Citation manipulation is adding or requiring references for a different purpose: to raise a person's citation count or a journal's metrics.
For medical journals, the ICMJE Recommendations put the standard in one sentence: references "should not be used by authors, editors, or peer reviewers to promote self-interests."
The practice takes five common forms:
| Practice | Who drives it | What it looks like |
|---|---|---|
| Coercive citation | An editor | A request to cite recent articles from the journal, with no specific gap identified |
| Reviewer self-citation requests | A peer reviewer | A batch of references to one author or group, with little or no rationale |
| Excessive self-citation | Authors | Their own papers cited where they do not support the statement |
| Citation rings, or cartels | Groups of authors or editors | An arrangement to cite each other's work or journals |
| Citation stacking | Two or more journals | One journal citing another at an unusually high rate |
The labels matter less than the question underneath all of them: does this reference support something the paper says? The Committee on Publication Ethics (COPE) has also published a discussion document on citation manipulation, and most publisher ethics policies now name it as misconduct.
Legitimate suggestion or coercion?
Not every request to cite a reviewer's own work is misconduct. Reviewers are invited because they know the field. The editors of Bioinformatics, writing after they banned a reviewer for coercive requests, noted that it is "not unusual that one or a few requested citations may be to the reviewer's own research."
What made their case different was scale and leverage. According to the editorial, the reviewer asked for an average of 35 added citations per review, about 90% of them to the reviewer's own papers, in wording that suggested the recommendation depended on them. The editors argued that a reviewer should explain why the authors would be remiss, or the paper weaker, without each citation.
Medical journals are not exempt. A cross-sectional study of peer reviews at the Journal of Psychosomatic Research, published in 2015, found reviewer self-citation was common, appeared more often in reviews recommending revision or acceptance than in those recommending rejection, and came without a rationale more often than citations to other people's work.
For each requested reference, ask three questions:
- Does it support, contradict, or qualify a specific statement in your manuscript?
- Would a reader be misled, or the paper weaker, without it?
- Can you say where it belongs, having read it?
If all three answers are yes, cite it, whoever wrote it. If not, you can decline. These signs push a request towards coercion:
- Several references share an author, research group, or journal, and the rest of the review is thin
- The only rationale is that "the literature review is incomplete"
- Adding them is linked to the recommendation, or repeated as a condition in the decision letter
- An editor asks for recent articles from the journal without naming a topic
- The papers concern a different population, method, or question from yours
The Bioinformatics editors also proposed treating more than one requested reference per printed journal page as substantial. That is one journal's rule of thumb, not a standard, but it gives you a sense of scale.
How to respond in your revision
Read every suggested paper before you decide. Then answer each request point by point, as you would any other comment; our guide to responding to peer review comments covers the structure. Declining is acceptable. The Bioinformatics editors told authors not to hesitate to say they had considered a suggestion and found the references unnecessary, and to contact the editor if the reviewer objected.
Wording for the response letter:
We thank the reviewer for these suggestions and have read each article. We have added [reference] to the Discussion (page X, lines Y–Z), where it directly addresses [point]. We have not added [references], because they concern [topic] and do not support a specific statement in the manuscript. The existing references [numbers] cover this point.
Wording for the editor, where the submission system allows confidential comments:
Reviewer 2 asked us to add [number] references, most of which share an author group, without linking them to specific statements. We have added the one that is directly relevant and explained our reasons for the others. We would welcome your guidance if these additions are expected as a condition of acceptance.
When the request comes from the editor, reply in the same factual way. Ask which statements the suggested articles should support, add any that fit, and give reasons for the rest. Keep the decision letter and your response. An editor who makes journal citations a condition of acceptance is also telling you something about how that journal is run, which is worth weighing when you choose where to submit next time.
Your own reference list
Senior authors are more likely to be on the other side of this problem. Self-citation itself is normal: your methods paper, your cohort's protocol, your earlier trial. John Ioannidis, writing in the Journal of Psychosomatic Research in 2015, argued that self-citation is usually appropriate and often necessary, while inappropriate self-citation can distort the literature and the metrics built on it.
Publisher policies draw the line at excessive and inappropriate self-citation, and at arrangements between author groups to cite each other. Taylor & Francis adds that references in reviews and other non-research articles should not be unfairly biased towards a particular research group, organization, or journal. Before submission, check:
- Every self-citation supports a specific statement, method, or dataset
- No reference is there because a co-author, collaborator, or editor expects it
- There is no reciprocal citing arrangement, formal or informal
- A review article represents the field, not mainly your group
- Nothing was added to flatter the target journal
The norm is set in the lab. If a junior researcher is told to add citations to a collaborator's papers, they need to know they can refuse and whom to tell; that belongs with the other research integrity responsibilities of a lab lead.
How manipulation is detected
A single request is hard to judge. Patterns across many papers are not, so detection happens at three levels.
Citation indexes. Clarivate's suppression policy for Journal Citation Reports (the May 2021 version) describes two kinds of journal-level distortion. For journal self-citation, it compares a journal's category rank with and without self-citations and examines the share of self-citations in what the journal cites and in what cites it, judged against its own category. For citation stacking, where a donor journal cites a recipient at an unusually high rate concentrated in the Journal Impact Factor window, it measures the donor's share of the recipient's citations and looks for individual items with non-scholarly or near-exclusive references to the recipient. Scopus separately runs a data model that flags journals showing unexpected patterns in citation graphs, publication activity, or author collaborations for re-evaluation.
Editors and publishers. Editors see reports from the same reviewer across many manuscripts. Elsevier's ethics policy, for example, asks editors to review reviewers' suggestions for self-citation and allows them to remove inappropriate citation requests before reports reach authors. The Bioinformatics editors proposed adding a checkbox to submission systems asking whether each reviewer requested citation of their own research.
Authors. You are often the first to see a coercive request. In the Bioinformatics case, the investigation suggested the reviewer had behaved similarly at other journals, and the editors noted that authors seem reluctant to raise these requests with editors, which lets the behaviour continue.
What happens when it is confirmed
The consequences depend on who is responsible:
- Journals. Under Clarivate's policy, journals showing citation distortion do not receive a Journal Impact Factor. In citation stacking, both donor and recipient journals are suppressed for one year and re-evaluated with the next year's data. Clarivate says it reports findings without accusations of wrongdoing, and suppressed journals stay indexed in Web of Science but may be re-evaluated. Scopus can discontinue a journal's coverage after re-evaluation.
- Authors and reviewers. Wiley's policy states that any citation manipulation results in rejection of the manuscript. Elsevier treats citation manipulation as systematic manipulation, which can lead to rejection or retraction, notification of an author's institution, employer, or funder, and refusal of future submissions from repeat or serious violators. Bioinformatics banned its reviewer.
Policies change, so check the current version from the index or publisher involved before relying on any detail.
Where to report, and what to include
Send the report to the person who can act on it:
- A reviewer's request: the handling editor
- An editor's request: the publisher's research integrity or publication ethics team, usually listed on the publisher's website
- Pressure from co-authors or collaborators: your institution's research integrity office
Do not name a reviewer publicly, even if you think you know who it is. The Bioinformatics editors themselves withheld the reviewer's name to protect peer review anonymity. Check the journal's confidentiality terms before sharing review content with anyone outside the process.
A useful report includes:
- The journal, manuscript ID, and dates
- The exact wording of the request
- The requested references, and what they have in common
- Which ones you judged relevant, and why
- Whether acceptance was tied to adding them
Keep it factual. You are describing a request and its effect on your paper, not proving intent; the editor or publisher can compare it with what they already know.
Directive Publications' peer review page lists what reviewers are asked to assess, including the accuracy, currency and adequacy of a manuscript's references. Our author resources cover the rest of what to check before you submit.
Frequently asked questions
Is it legitimate for a reviewer to suggest citing their own papers?
It can be. Reviewers are chosen for their expertise, so their own work is sometimes exactly what a manuscript is missing, and one or two relevant self-citations are not unusual. The request becomes a problem when the references do not support a specific point in your paper, arrive in bulk without a rationale, or are tied to the reviewer's recommendation. The ICMJE Recommendations state that references should not be used by authors, editors, or peer reviewers to promote self-interests.
How should authors respond when an editor asks for citations to the journal?
Ask which statements in your manuscript the suggested articles should support, and add any that genuinely fit after reading them. If the request is generic, such as citing recent papers from the journal without naming a topic, decline politely in your response letter and explain that your existing references already cover those points. If citations are made a condition of acceptance with no scholarly reason, contact the publisher's research integrity or ethics team.
How are citation rings detected?
Mostly through patterns in citation data rather than single papers. Clarivate's published suppression policy for Journal Citation Reports looks at how much of one journal's citations come from another, how concentrated those citations are in the Journal Impact Factor window, and whether individual articles cite the recipient journal almost exclusively. Scopus also runs a model that flags unexpected patterns in citation graphs, and editors can notice reviewers who request the same references across many manuscripts.
Can excessive self-citation lead to penalties for researchers or journals?
Yes. Under Clarivate's suppression policy, a journal whose self-citation or citation stacking distorts its category rank does not receive a Journal Impact Factor that year, and Scopus can discontinue coverage of a journal after re-evaluation. For individuals, published publisher policies include rejecting the manuscript, retracting articles, notifying the author's institution, and refusing submissions from repeat offenders. Self-citation that supports specific statements is normal and is not what these policies target.
Where can authors report suspected citation manipulation?
Report a request from a reviewer to the handling editor, who can look for a pattern across that reviewer's reports. Report a request from an editor to the publisher, usually through the research integrity or publication ethics contact on the publisher's website. If the pressure comes from co-authors or collaborators, your institution's research integrity office is the right first contact.