What to Do When a Source in Your Paper Is Retracted

You are checking your references when a journal page labels a source “retracted.” Do not simply delete the reference or replace its citation. Find the claim that depends on it, read the actual notice, and revise the argument around the evidence you can still defend.

Direct answer

When a source in your paper is retracted, verify the source’s identity and read the publisher’s notice before deciding what to change. Trace the affected findings through your paper, then remove, narrow, or replace claims according to what the remaining evidence supports. You may discuss a retracted work transparently as an object of analysis, but you should not rely on its affected findings as though the retraction had not occurred. If you have already submitted the assignment, contact your instructor about the appropriate process.

What does the notice actually change?

The notice tells you what happened to the publication; your revision must address what that change means for your argument. A database warning is a reason to investigate, not a substitute for reading the explanation.

Start by matching the flagged record to the source you used: title, authors, journal, and DOI where available. Check whether you have the published article, a manuscript version, or a separate notice. Crossref’s versioning guidance explains how editorially significant updates are documented through linked correction or retraction notices. A different file or newer upload does not, by itself, establish that the problem has been resolved.

Distinguish the status before changing your draft:

• Correction: identifies an error and supplies corrected information. Determine whether the correction affects the passage, result, or interpretation you used.
• Expression of concern: alerts readers to concerns about a publication. It is not the same status as a retraction, but it calls for examining whether your claim depends on the questioned material.
• Retraction: formally identifies a work as retracted. Read the reasons and scope rather than assuming that every sentence is false or that the problem is merely cosmetic.

The National Library of Medicine’s Errata, Retractions, and Other Linked Citations in PubMed describes these linked notice types. Its policy also states that retraction indexing does not distinguish honest error from scientific misconduct. Do not translate a retraction label into an accusation of fraud.

Before: What breaks when the supporting source is retracted?

The problem becomes clearer when you follow a source into a paragraph rather than treating it as an isolated reference.

Hypothetical case (sample text for demonstration only)

Imagine a student-authored policy paper about evening academic advising. A program evaluation supplies its claim about degree completion. In this invented scenario, the publisher retracts that evaluation because duplicate student records made the completion comparison unreliable.

Separate interview evidence remains available. Participants described evening appointments as easier to attend around their jobs, but those interviews did not measure degree completion. No real study or notice is being described here.

Before (sample text for demonstration only)

The university should expand evening advising because it improves degree completion for working students. The program evaluation supports this outcome claim, while interview accounts show why the policy could work: students described evening appointments as easier to attend around their jobs. Together, these findings establish evening advising as a retention strategy. The university should therefore treat expanded hours as an evidence-backed route to helping working students finish their degrees.

The paragraph’s recommendation depends on a particular result: improved degree completion. Once the hypothetical notice undermines that comparison, the interview accounts cannot quietly take over its job. They concern reported access, not whether students completed their degrees.

This is the Citation Swap Error: changing the reference attached to a sentence while leaving an unsupported claim intact. Recognize it by comparing the old claim with the replacement source’s actual findings. A shared topic is not enough.

After: What can the remaining evidence support?

Repair the claim before repairing the citation. In this scenario, the remaining evidence permits a narrower argument about access—not the original argument about completion.

After (sample text for demonstration only)

Evening advising deserves consideration as an access measure, but the evidence available in this draft does not establish that it improves degree completion. Interview participants described evening appointments as easier to attend around work; those accounts support a claim about their reported access, not graduation outcomes. The evaluation previously used to support the completion claim was retracted because duplicate records made its comparison unreliable. It therefore cannot carry that claim here. The recommendation should be narrowed to assessing demand for evening access, with any claim about degree completion left unresolved until suitable evidence is available.

The revised passage does not prove that evening advising is ineffective. It says that this draft lacks adequate support for the completion claim. That distinction prevents an unsupported positive claim from becoming an equally unsupported negative claim.

Citations are omitted from these invented passages. In a real paper, cite the evidence you retain and the actual notice if you discuss its explanation.

What changed between the passages?

The revision changes the reasoning, not just the source label. Its notice-to-claim trace is:

Notice’s finding → affected result → dependent claim → available evidence → revised conclusion.

In the hypothetical case, duplicate records undermine the completion comparison. The draft uses that comparison to claim that evening advising improves completion. The remaining interviews address appointment access. The conclusion therefore moves from endorsing a completion strategy to considering an access measure.

Notice the boundary: the retraction notice establishes the publication-status problem and explains its stated reason. It does not supply new evidence about whether evening advising works. The interviews supply limited evidence about participants’ reported experiences. The student must decide what argument those materials justify.

Which wording should you cut and replace?

The following replacements apply to the same hypothetical case (sample text for demonstration only):

Cut: “Together, these findings establish evening advising as a retention strategy.”

Replace with: “The interview accounts support a claim about reported access, not degree completion.”

Cut: “The university should therefore treat expanded hours as an evidence-backed route to helping working students finish their degrees.”

Replace with: “The university could assess demand for evening appointments without claiming that expanded hours improve completion.”

These are different claims, not softened versions of the original. Adding “may” to an unsupported outcome claim does not repair its evidence.

A replacement source could support a stronger conclusion, but only if its findings fit the claim. Compare the population, setting, outcome measured, method, and limitations. Evidence about satisfaction, attendance, or convenience cannot automatically support a conclusion about completion.

If you can identify the affected sources but cannot see how the argument should be reorganized, Research Planning & Editing can support evidence organization and source integration. That help does not certify a study’s scientific validity or transfer responsibility for the argument away from you.

How do you find every claim that depends on the source?

Search beyond the bibliography. Look for the author’s name, distinctive title words, quoted language, and the finding you borrowed. Inspect footnotes, endnotes, tables, captions, appendices, and any sentence that carries the source’s result without repeating its citation.

Then follow the argument forward. A source might appear in a literature-review paragraph while its conclusion reappears in the introduction, recommendation, abstract, or final section. Removing the cited sentence will not fix those downstream claims.

Use this scaffold to make a notice-to-claim record for your own paper:

• Source identity: [Exact publication and version used.]
• Notice: [Publisher’s notice and its stated reason or concern.]
• Use in my draft: [Passage, quotation, finding, method, or interpretation borrowed.]
• Dependency: [What becomes unsupported if this use is removed?]
• Remaining evidence: [What other sources support, including their limits.]
• Decision: [Retain with explanation, replace, narrow, or remove—and why.]

Keep notice language separate from your interpretation. If the notice does not explain whether a particular finding remains usable, record that uncertainty instead of supplying your own reassurance.

Can your claim pass the Citation-Off Test?

Temporarily set aside the affected source and read the dependent paragraph. For each factual or analytical claim, identify the remaining evidence that supports its actual wording.

Ask whether that evidence addresses the same outcome and context. Then read the paragraph’s concluding sentence. Does it still follow, or does it rely on the removed result?

A claim fails the Citation-Off Test when its apparent support consists of background information, a related topic, or evidence answering a different question. Your next move is to find suitable support, narrow the claim, or remove it—not hide the gap behind a new citation.

Where should you check publication status?

Use the publisher’s article page and actual notice as the starting point. Status tools can help you locate updates, but they do not replace reading them.

Crossref’s Crossmark guidance explains how Crossmark exposes publication updates, including corrections and retractions. Crossref explicitly warns that Crossmark is not a guarantee. A badge does not certify the research as reliable, and an absent badge is not a verdict about the work.

For sources within its coverage, PubMed can help connect an original article with related notices. The PubMed User Guide documents linked citations and retraction-related search information. Use those links to locate the relevant records, then read the notice itself.

PubMed is not a universal database for every discipline. Failure to find a notice there does not clear a source from another field. Likewise, an older downloaded PDF or an unchanged reference-manager entry should not settle a question about the publication’s current status.

If the notice or its technical explanation is inaccessible or unclear, ask a librarian, instructor, or subject specialist for help. Do not invent the missing explanation.

When can a retracted source remain in your paper?

A retracted publication can remain relevant when the paper examines the publication itself: its reception, influence, claims, or subsequent retraction. In that use, the work is an object of analysis rather than unqualified support for the disputed finding.

Make the distinction visible where it matters. Identify the retracted status, cite the notice when explaining it, and state what you are using the original publication to demonstrate. A warning buried in the reference list does not qualify a confident claim in the body.

Assignment context changes the decision. A history-of-science paper may need the original article to explain a controversy. A clinical evidence review needs to address whether its recommendations depend on affected findings. A methods discussion should not assume that a procedure remains defensible merely because the notice focuses on a different issue.

Your instructor, review protocol, or disciplinary requirements may determine how such material should be handled. Citation formatting cannot make that evidence judgment for you.

Common questions

Must I delete every mention of a retracted source?

No. Remove unsupported reliance, not necessarily every mention. Keep the work when its retraction or historical role is relevant to your analysis, and explain its status clearly.

If you remove all discussion of it, reconcile the reference list with the requirements of your citation style and assignment. If you retain discussion of the article or notice, preserve the records needed to identify what you actually used.

Can I replace it with a source that reaches the same conclusion?

Only after checking that the replacement supports your claim. A matching conclusion is not enough if the source examines a different population, measures a different outcome, or uses a method that cannot justify your wording.

Read the replacement and revise accordingly. Do not cite an article you have not examined merely because its abstract appears convenient.

What if the source has a correction or expression of concern instead?

Do not relabel it as retracted. Read the specific notice and examine the overlap with your use.

A correction to material you never relied on may leave your claim unchanged. A correction to the result you quoted requires attention. An expression of concern signals unresolved concerns; seek suitable alternative support where possible and ask your instructor how to handle any remaining dependency.

What if another article cites the retracted finding?

An indirect citation does not restore the finding. Check whether the later article provides independent evidence or merely repeats the affected result.

If your draft relies on that repeated result, trace the dependency back to the original source. This can reveal a Citation Swap Error even when the retracted article never appears in your own reference list.

What if I already submitted the assignment?

Contact your instructor through the course’s normal process. Identify the source, provide the notice, explain which part of your paper may be affected, and ask whether they want a correction or revised file.

Do not silently replace a submitted document or assume a universal resubmission procedure. The appropriate response depends on the assignment and institutional process.

Before You Submit

Check the repaired argument before polishing the reference list:

• The notice belongs to the exact source and version you used.
• Your account of the notice does not add unsupported allegations or explanations.
• Affected claims have been removed, narrowed, or supported with evidence you have examined.
• The introduction, abstract, recommendations, and conclusion reflect those changes.
• Any retained discussion of the retracted work identifies its status and purpose.
• Unresolved technical questions have been raised with the appropriate instructor or specialist.

Once the evidence decisions are settled, use the guide to fixing citation problems before submission to reconcile citations, notes, and references. If that cleanup needs assistance, citation formatting support belongs at this stage—not in place of deciding what the evidence supports.

The repair is complete when the argument fits the defensible evidence, not when the warning disappears from your bibliography.

For help organizing that repair or reviewing source integration, start an order with your draft, assignment instructions, and the actual notice. Specify the support you need, keep the evidence decisions visible, and review the revisions yourself before submission.