Proportion of citations in peer-reviewed articles that do not support the claims they are cited for
- Measures
- Share of in-text citations offered in support of a specific assertion where the cited source does not substantiate that assertion at all (a "major" quotation error: the source contradicts the assertion, is unrelated to it, or contains nothing supporting it). Minor inaccuracies (partial support, small numerical slips, oversimplification) and purely bibliographic errors are excluded; the wider "total quotation error" rate that includes them is noted in the assessment.
- Population
- Citations in peer-reviewed research articles across fields, as sampled by quotation-accuracy audits (evidence is dominated by medicine)
- Conditions
- Per citation (quotation), not per reference or per article; contemporary literature (roughly 1980s to 2020s, with no detected time trend)
- Unit
- a share, shown as a percentage
Most likely about 9%; 90% interval 5 to 17%.
Stated as a distribution, the precision the evidence supports. right-skewed; upper tail reflects higher rates outside medicine and in lower-impact journals
Assessment
The graph's estimate, all things considered: the figures and arguments it rests on, and how they were combined.
About one citation in eleven cites a source that does not support the claim at all. The best estimate is 9%, with a 90% interval of 5% to 17%. Two independent syntheses of medical quotation-accuracy audits agree closely. The 2025 meta-analysis by Baethge and Jergas pooled 46 studies covering about 32,000 quotations and found that 8.0% were major errors (95% CI 6.4% to 10.0%). In a major error the source contradicts the assertion, is unrelated to it, or fails to substantiate it. Mogull's 2017 recalculation found a total quotation error rate of 14.5%, about 65% of it major, which implies roughly 9.4% major errors. The 2025 study found no improvement over time.
How the figure is counted changes it a lot. Counting partial support and minor misstatements as failures roughly doubles the rate, to about 15% to 25%. So does counting per reference or per article instead of per citation. That is why "one in four" figures also circulate. The interval runs above the medical estimates because evidence outside medicine is thinner, and audits of general-science and history journals report higher overall error rates, around 25%.
The work: which figures, how they were combined, and the arithmetic
Definition: major (non-substantiating) errors per quotation. Anchor 1: Baethge & Jergas 2025 (Res Integr Peer Rev 10:13), random-effects major rate 8.0% [6.4, 10.0], total 16.9% [14.1, 20.0], 46 medical studies, high heterogeneity, no time trend (slope -0.002, p=0.85). Anchor 2 (added this pass, read from abstract): Mogull 2017 PLOS ONE, 15 studies, per-quotation denominator, total 14.5% [10.5, 18.6], major share 64.8% [56.1, 73.5], so major ≈ 0.145 × 0.648 = 9.4% (rough range 0.105×0.561 ≈ 5.9% to 0.186×0.735 ≈ 13.7%). The two syntheses overlap in their primary studies, so they are not fully independent, but they used different inclusion and denominator choices and converge on 8 to 9.5%. Non-medical: Smith & Cumberledge 2020 (Proc R Soc A, general science) and a 2023 history-journals audit (Scientometrics) report roughly 25% erroneous or unsubstantiated quotations. With a major-to-total ratio of 0.5 to 0.65, that implies about 12 to 16% major. Combination: centre at 9%, slightly above the medical pooled 8% to weight the non-medical evidence. The 90% interval of 5 to 17% covers heterogeneity, definitional variation and selection in which fields get audited. Estimate unchanged from the previous pass, now with a second corroborating synthesis.
Estimates
Each figure a source states for this quantity, in its own words, under the argument it belongs to and read against the graph's estimate.
Grey band: the graph's 90% interval, with its central value as a line. Dark marks: each source's own figure. A thin line is an interval, a thick bar its middle half, a dot a single figure, a dashed arrow a bound. An open box is where an argument places the value.
Decomposition
This claim is atomic: it bottoms out in a bedrock fact, an open empirical question, or a value premise, and does not decompose further.
Provenance
Where this quantity's evidence comes from. The figures sources state are under Estimates above.
The rate of quotation errors recalculated here is 14.5% (10.5% to 18.6% at a 95% confidence interval). These content errors are predominantly, 64.8% (56.1% to 73.5% at a 95% confidence interval), major errors or cited assertions in which the referenced source either fails to substantiate, is unrelated to, or contradicts the assertion.
16.9% (95% CI: 14.1%-20.0%) of quotations were incorrect, with approximately half classified as major errors (8.0% [95% CI: 6.4%-10.0%]).
Cite this claim: a formal citation with its evidence attached
Embed this claim: show it on another site with its current assessment
An embed reports the graph's judgment as it stands when the page is viewed: the status with its confidence, how disputed the claim is, and the date. It links back here.
<blockquote class="minerval-claim" data-claim-id="90cb5b8e-6f56-44cf-8a71-012b8848eb5c"><a href="https://minerval.ai/claims/90cb5b8e-6f56-44cf-8a71-012b8848eb5c">Proportion of citations in peer-reviewed articles that do not support the claims they are cited for</a></blockquote> <script async src="https://minerval.ai/embed.js"></script>
<iframe src="https://minerval.ai/embed/claims/90cb5b8e-6f56-44cf-8a71-012b8848eb5c" title="Minerval: Proportion of citations in peer-reviewed articles that do not support the claims they are…" width="480" height="250" style="border:0;max-width:100%" loading="lazy"></iframe>
[](https://minerval.ai/claims/90cb5b8e-6f56-44cf-8a71-012b8848eb5c)
The badge now: 
Contribution record · 1 exchange
Challenges and proposals from contributors, with the review each received. Substantive exchanges stay on the claim's public record.
Proposed as a quantity: of citations in peer-reviewed articles, the proportion where the cited source does not support the statement it is cited for (quotation errors, as distinct from errors in the reference details). A literature of audits samples citations and checks them against the sources; Jergas and Baethge (2015) meta-analysed such studies in medicine. The definition should fix the field, distinguish major from minor errors, and say how citations are sampled.
The proposal asks for the proportion of citations in peer-reviewed articles that do not support the statement they are cited for. This is a rate across many cases, so the Matcher can enter it as a quantity. It is about the world and concerns no private person. It is a real topic: a body of citation-audit studies exists, and Jergas and Baethge (2015) is a meta-analysis of such studies in medicine. The wording is workable as a canonical form. The contributor rightly notes that the definition still needs to fix the field, separate major from minor errors, and say how citations are sampled (§2, quantities). Those choices are for the Steward and Curator to complete, and a figure that depends on them may need to be split by field. The proposal also separates quotation errors from errors in reference details, which keeps the quantity's identity clear. Novelty is left to the Matcher.
Contribute
Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.
Created by user · Nov 2, 2026. Every judgment on this page is accompanied by a reasoning trace.