Can a Frequently Repeated Claim Have Surprisingly Little Direct Evidence?

If everyone keeps saying it, shouldn't there be a lot of evidence behind it?

You encounter the same claim in papers, reviews, introductions, and discussion sections. Authors describe it as “well established,” “widely recognized,” or simply state it without much qualification. The proposition has become part of the background knowledge of the field.

It is natural to assume that such repetition reflects a large body of direct evidence. Sometimes it does. But a claim can also become highly visible because many papers repeat, summarize, or cite it while only a much smaller number of studies have actually tested the proposition directly.

The number of times a claim appears in the literature and the amount of evidence supporting that claim are therefore related only imperfectly.

A large literature around a claim can rest on a small evidential core

Claims and evidence can grow at very different rates

Start by separating papers about a claim from papers that test it

A literature can contain many papers discussing the same proposition without all of those papers providing direct evidence for it.

Some may be reviews. Others may mention the proposition in their introductions. Some may use it to justify a new study without testing the claim itself. Still others may discuss its implications, develop theory around it, or cite another source for the statement.

When estimating the evidential base, the first useful question is therefore not “How many papers mention this?” but “Which papers actually contain data or analysis that directly bear on this proposition?”

Ten citations do not necessarily represent ten tests

Suppose a review says that an educational intervention improves retention and provides ten citations. Those ten references could represent ten independent intervention studies. But they could also include two empirical studies, three reviews, two theoretical papers, a commentary, and two papers that themselves cite the same original experiment.

Counting references alone cannot distinguish these structures.

This is why later papers repeatedly citing one another can create the appearance of a larger evidential base than actually exists.

A claim can generate more scholarship without generating more tests

Once a proposition becomes influential, researchers can build around it. They may investigate consequences, mechanisms, applications, ethical implications, implementation, or related constructs. The literature surrounding the claim can therefore expand substantially even if relatively few studies continue to test the foundational proposition itself.

That growth is not inherently problematic. Research necessarily develops by treating some previous findings as provisional foundations. The problem occurs when the size of the surrounding literature is mistaken for the amount of direct evidence supporting the foundation.

Citation networks can multiply the appearance of support

One primary study can be cited by several papers. Those papers can then be cited by many more. Over time, a branching citation cascade may develop around the original work.

The network can become large even though the number of primary studies addressing the specific claim remains small. A bibliometric network measures relationships among publications. It does not automatically tell you how many independent observations substantiate a proposition.

Secondary sources can conceal how narrow the evidence base is

Reviews and textbooks can make research easier to navigate by synthesizing large literatures. But when researchers cite these summaries without checking what evidence lies beneath a particular claim, the distinction between synthesis and primary support can become difficult to see.

Paper A may report the original evidence. Review B summarizes Paper A. Papers C, D, and E cite Review B. Later papers cite C, D, or E. A reader entering the literature years later can encounter the claim in many locations even though the relevant evidential path repeatedly converges on Paper A.

This is one reason important propositions sometimes warrant being traced back toward their original evidential source.

Repetition can also strengthen the claim

The problem becomes more consequential if the proposition changes while spreading. An early study might report a tentative association. Later authors may call it an effect. Still later papers may describe the effect as established.

Now two processes are occurring simultaneously: the number of papers repeating the proposition is increasing, and the wording is becoming more certain.

This is one way citation distortion can spread through a literature. The claim can accumulate rhetorical authority faster than it accumulates independent empirical support.

There is empirical evidence that citation networks can create this appearance

Greenberg's 2009 analysis of a specific biomedical belief provides an unusually detailed example. He examined 242 papers and 675 citations concerning a claim about beta amyloid in inclusion body myositis and identified 220,553 citation paths supporting the belief. His network analysis distinguished papers containing data relevant to the claim from papers that merely cited or discussed it.

Greenberg concluded that citation bias, amplification, and several forms of citation-based invention contributed to what he termed “unfounded authority.” His analysis also found that papers containing no data addressing the claim could amplify it through citation. This was a case study of one biomedical literature, not evidence that every widely repeated research claim develops in the same way.

Its broader methodological lesson is nevertheless useful: the apparent volume of supporting literature can differ substantially from the volume of direct evidence.

Quotation errors can help inaccurate claims persist

A repeated claim can also become detached from the source that supposedly supports it. Research on quotation accuracy in medical publishing has documented substantial rates of references that do not accurately support the claims attached to them.

A 2025 systematic review and meta-analysis covering 46 studies and approximately 32,000 quotations or references estimated that 16.9% were incorrect and 8.0% involved major errors. The studies were heterogeneous, so these figures should not be generalized to every discipline or literature.

An earlier meta-analysis likewise found quotation errors to be a meaningful problem and specifically noted that indirect referencing can allow errors to propagate.

These findings do not show that frequently repeated claims are generally unsupported. They show why citation repetition alone is insufficient evidence that the underlying source-to-claim relationships have been independently verified.

Little direct evidence does not automatically mean weak evidence

Be careful not to replace one counting error with another. A claim supported by only two direct studies is not necessarily weaker than a claim supported by twenty studies. Study design, sample size, measurement quality, bias, precision, consistency, and relevance all matter.

A single large, well-conducted study can sometimes provide more informative evidence than numerous small or poorly designed studies. Conversely, one striking study may provide a fragile foundation for a broad claim.

The number of independent studies is therefore one dimension of the evidence base, not a complete measure of evidential strength.

Direct evidence can also disagree

Finding the primary studies is only the beginning. They may not all point in the same direction.

A widely repeated claim might originate from one positive study while later null or contradictory studies receive less attention. Alternatively, several independent studies may converge strongly, making the familiar claim reasonably well supported despite a smaller evidence base than you initially assumed.

The goal is not to expose repetition as suspicious. It is to reconstruct the evidence accurately enough to know what level of confidence the literature actually warrants.

How twenty papers can reduce to two direct studies

The twenty papers have not become irrelevant. They may contribute theory, context, implementation knowledge, or synthesis. What changed is your estimate of how much direct empirical evidence supports one particular proposition.

A familiar claim is not necessarily a well-tested claim

Audit the evidential core of claims that carry substantial weight

You do not need to reconstruct the evidence ancestry of every familiar statement. But when a repeated claim becomes central to your research question, theoretical rationale, interpretation, or recommendation, it can be worth finding out what lies underneath the repetition.

A useful literature review does more than report how commonly a proposition appears. It helps readers understand what kind of evidence produced that apparent consensus.

How to find the direct evidence behind a repeated claim

Questions about repeated claims and direct evidence

Repetition tells you how far a claim has travelled, not how much evidence created it

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.