InvestigationMedia
Why does a repeated claim eventually seem true?
Encountering a claim several times can increase its credibility, whether it is true or false. This experimental effect does not turn every repetition into persuasion: its magnitude depends on the knowledge brought to bear, the context and the corrections provided. In children as well as in schools, evidence exists but remains partial.
A claim encountered before seems, on average, truer than a comparable new claim. Yet repetition provides no additional evidence. It is this gap between familiarity and evidentiary value that psychologists call the illusory truth effect. It applies to true statements as well as false statements: it changes their perceived truth, not their accuracy.12
What can be retained
- Established
- Established
- Established
- ReportedThe 2026 meta-analysis estimates the corrected effect at g = 0.37.7
This estimate comes from a single synthesis, whose authors point to substantial heterogeneity.
- Disputed
Established Shown by verifiable documents or data.
Reported Stated by an identified source, not independently confirmed.
Disputed Contradicted by other documented evidence.
1977: repeating without adding evidence
Researchers Lynn Hasher, David Goldstein and Thomas Toppino describe the foundational experiment in the scientific journal Journal of Verbal Learning and Verbal Behavior in 1977. According to their article, the analysis covers 40 students, after excluding ten people absent from at least one session. Participants assess 60 claims per session, across three sessions spaced two weeks apart. The published article mentions no specific funding.1
According to Hasher, Goldstein and Toppino, twenty claims recur across all three sessions; the others are new. The statements are plausible, but chosen so that students are unlikely to know their truth status precisely. Validity judgments increase for repeated claims, both true and false, unlike new claims.1
Researchers Aumyo Hassan and Sarah J. Barber find this increase again in two experiments published in 2021 in the scientific journal Cognitive Research: Principles and Implications. Their study focuses mainly on frequency: up to 9 presentations in one experiment and 27 in the other, before a judgment one week later. According to their results, additional presentations continue to increase perceived truth, with progressively smaller gains.510
On another scale, the authors of the systematic review published in 2026 in the scientific journal Nature Communications bring together 182 studies, 366 effect sizes and 31,184 participants. According to this meta-analysis, after correcting for effects associated with small studies, the average effect is small: g = 0.37, with a 95% confidence interval from 0.30 to 0.44. This figure is a standardized difference between conditions, not a percentage of people convinced.7
Estimates also depend on what is being compared. The authors of this meta-analysis report a historical estimate of d = 0.39 when comparing the same statement between exposure and test, versus d = 0.53 when comparing repeated and new statements at test. While this synthesis gives d = 0.53, the authors of the systematic mapping published in the scientific journal Psychonomic Bulletin & Review report d = 0.50 for a comparison between statements under a random-effects model. These different presentations do not make it possible to attribute the gap precisely to a single cause.711
Effect sizes to compare with caution
- Disputed
Disputed Contradicted by other documented evidence.
Reading fluency can become a cue to truth
The main explanation involves processing fluency: a previously encountered statement becomes easier to process, and this ease can serve as a cue to accuracy. Researchers Lisa K. Fazio, Nadia M. Brashier, B. Keith Payne and Elizabeth J. Marsh examine this mechanism in two experiments published in 2015 in the scientific journal Journal of Experimental Psychology: General. Hassan and Barber also present it as a central explanation of the effect.3122
According to Fazio, Brashier, Payne and Marsh, repetition can increase perceived truth even when participants know the correct answer. Their conclusion concerns the use of knowledge at the time of judgment: possessing correct information does not guarantee that it will outweigh the felt ease of processing.3
This protection is nevertheless not identical across all populations. According to Fazio and her colleagues, the effect can appear despite correct knowledge, whereas researchers Nadia M. Brashier, Sharda Umanath, Roberto Cabeza and Elizabeth J. Marsh, in a study published in 2017 in the scientific journal Psychology and Aging, observe among the older adults studied an effect only when relevant knowledge is lacking. The populations and protocols differ: these results do not permit the conclusion that knowledge is always useless.313
Familiarity should not be confused with conscious memory of a previous encounter either. According to researcher Frederick T. Bacon, in his experiment published in 1979 in the scientific journal Journal of Experimental Psychology: Human Learning and Memory, the increase in credibility depended on subjective recognition of the repetition. While Bacon reports this experimental result, researchers Emma L. Henderson, Daniel J. Simons and Dale J. Barr, authors of the longitudinal study published in 2021 in the scientific journal Journal of Cognition, set out a model allowing familiarity without conscious memory of the exposure. This model, on its own, does not constitute a demonstration of a subliminal effect.149
The authors of a working paper from the University of Freiburg, devoted to the cognitive processes underlying the truth effect, for their part break down the decision mechanisms at work and report that some mechanisms associated with repetition decreased as the retention interval increased. This paper announces public access to its data and code; its editorial status is not specified.15
Finally, liking a brand more and believing a claim more are two different outcomes. According to the authors of a marketing study on advertising wearout, hosted on Duke University’s website, repetition consistently improves memory for the advertisement or brand and may affect brand preference. These measures do not demonstrate that consumers judge a factual claim about the product to be more accurate.16
The limits of the explanations
- Disputed
- Disputed
- Unknown
Disputed Contradicted by other documented evidence.
Unknown No available source can establish it.
Two encounters may be enough, without a universal timeline
It is not necessary to encounter a claim dozens of times. The main gain generally appears at the second encounter, after which additional repetitions produce smaller increases. This pattern is found in the study by Hassan and Barber and in separate experiments devoted to validity judgments of repeated rumours.56
According to Henderson, Simons and Barr, whose registered report was published in 2021 in the Journal of Cognition and deposited in the University of Glasgow’s institutional repository, the effect appears at each of the four intervals tested: immediately, after one day, one week and one month. The study includes 608 participants, of whom 567 were analysed, and observes a decline as the delay increases. This is a single experiment, not several independent replications.817
The trajectory remains contested. According to the authors of this longitudinal study, the effect declines with the delay, whereas the 2010 meta-analysis they discuss found no general relationship between delay and effect size. They also report an earlier experiment in which the effect on false claims was present after one week, but had disappeared after one and three months. An effect observed at one month in one protocol therefore does not establish a duration that applies in all situations.89
In children
The effect is not limited to adults. According to researchers Lisa K. Fazio and Carrie L. Sherry, authors of the study The effect of repetition on truth judgments across development, published in the peer-reviewed scientific journal Psychological Science, it is observed from age 5 onward, as well as among adults: repeated claims are judged truer than new ones. The authors of the 2026 meta-analysis report the same finding.187
The same study is among those finding that relevant factual knowledge does not systematically protect against repetition. This finding concerns truth judgments collected in an experimental setting. It does not describe what happens in a classroom, and the delays studied in this literature are generally short.18711
Children do not judge only repeated sentences: they learn largely through what adults tell them. According to researchers Paul L. Harris and Melissa A. Koenig, in an article published in the scientific journal Child Development on trust in testimony, children thereby acquire a large share of their knowledge in science as well as religion, and this trust is exercised selectively depending on the informant. A theoretical text disseminated by the Cognition and Culture research website likewise points out that beliefs concerning realities not directly observed rely largely on the testimony of others. Neither one tests the repetition of a claim.1920
The suggestibility of young children has been studied separately. According to psychologist Stephen J. Ceci of Cornell University and his colleagues, in a study on the cognitive and social factors of suggestibility in preschool children, more interviews and a longer interval produced the strongest suggestibility, and some children appeared to have internalized false suggestions to the point of resisting debriefing. This study reports partial funding from the National Science Foundation, a US federal research agency. It concerns a related but distinct phenomenon from the illusory truth effect, and does not make it possible to isolate the specific role of repetition.21
Source monitoring appears to be one avenue of protection. According to researchers Jessica W. Giles, Alison Gopnik and Gail D. Heyman of the University of California, in a study published in 2002 in the scientific journal Psychological Science, training preschool children to identify the source of their memories reduces their suggestibility. The authors of a review published in the scientific journal Current Psychology on source monitoring theory and training children reach the same conclusion: the ability to link a memory to its source is one avenue for reducing suggestibility.2223
Over time, the most comprehensive longitudinal measure remains that of Henderson, Simons and Barr: an effect present immediately, after one day, one week and one month, decreasing as the interval lengthens. No comparable follow-up has been published for children.824
A related result concerns memory, not perceived truth. In a theoretical review published by the scientific journal PNAS on gist-based communication, it is reported that 53 of 55 experiments observed, from childhood to adulthood, an increase in gist-based memory errors. These are false memories, not a direct measure of the illusory truth effect.25
Research in science education documents another phenomenon: students’ misconceptions. The authors of an article published in the scientific journal Science & Education, those of the meta-analysis Effectiveness of conceptual change strategies in science education, published by the scientific publisher Wiley, and the educational website The Science Teacher converge on one point: these conceptions can coexist with correct knowledge and resist instruction.262728
These conceptions are not therefore fixed. According to the authors of the meta-analysis published by Wiley, joined on this point by those of the Science & Education article, structured conceptual change strategies improve science learning, with highly heterogeneous effects from one study to another. This research concerns the persistence of conceptions and ways of changing them, not the illusory truth effect: it does not make it possible to attribute students’ misconceptions to the repetition of a statement.2726
Belief perseverance refers to something else again: the effects of initial information may persist after it has been corrected or discredited. According to C.A. Elizabeth Luus, in her 1991 doctoral dissertation in psychology on eyewitness confidence, supervised by Gary L. Wells, correcting erroneous social information generally did not eliminate its effect on confidence.29
In another university dissertation, devoted to implicit weight-related bias, belief perseverance was reported immediately and two days after the initial information was discredited. This second study concerns adults, and neither study relies on the repetition of statements: they describe a phenomenon related to the illusory truth effect, observed in distinct protocols.3029
In children: what is established and what is not
- Established
- Established
- ReportedMore interviews and a longer interval increase preschool children’s suggestibility.21
Result reported by Ceci and his colleagues, with neither sample size nor effect size published in their abstract, and without isolating the specific role of repetition.
- Reported
- Established
- Reported
- Unknown
- Unknown
Established Shown by verifiable documents or data.
Reported Stated by an identified source, not independently confirmed.
Unknown No available source can establish it.
At school: textbooks, teachers, curricula
School materials themselves can transmit errors. According to an institutional post by the British assessment organization Cambridge Assessment, an analysis of 138 textbooks found that many explained the concept of effect size in a simplified or incomplete manner. According to the scientific news website ScienceDaily, the same error was repeated for nearly one hundred years in chemistry textbooks. According to the authors of a 2011 study on physics textbooks, all the textbooks examined continued to attribute the measurement of the gravitational constant to Cavendish, without the number of textbooks or their selection method being specified in this report. The psychology news website PsyPost and a website devoted to the teaching of psychology, for their part, report that introductory psychology textbooks still distort famous experiments and controversial debates.3132333435
The best-documented case is that of Ernst Haeckel’s embryo drawings. A doctoral dissertation on these drawings, evolution and secondary-school biology textbooks searched for embryo grids in 240 textbooks published between 1907 and 2010; the published record does not provide results by decade. Kenneth R. Miller, a coauthor of biology textbooks, acknowledges on his website the use of these representations in two of his textbooks before a correction in 1998. These two accounts agree on the material persistence of representations derived from Haeckel’s drawings until the end of the 20th century. The authors of an article in the scientific journal Theory in Biosciences, for their part, show that embryology remained present in biology textbooks in the German Democratic Republic from 1951 to 1988 despite several reforms; the continuity of the topic does not prove the persistence of any particular error.36373839
The accuracy of the drawings and the scope of Haeckel’s biogenetic law are two distinct questions. According to biologist Michael K. Richardson and his colleagues, in a letter published by the scientific journal Science, comparison with actual embryos revealed numerous incorrect details in the drawings. According to the authors of an article in the scientific journal Biological Reviews of the Cambridge Philosophical Society, a limited version of the biogenetic law would be supported by several recent studies, provided that it is applied to individual traits. A restricted version of the principle can therefore be defended without validating the drawings or strict recapitulation.404142
The characterization as fraud remains debated. The historian of science Robert J. Richards, in an article in the scientific journal Biology & Philosophy, argues that fraud has not been proven. Regarding the presence of the drawings in current textbooks, positions diverge: according to the National Center for Science Education, a US organization advocating the teaching of evolution, contemporary textbooks and developmental biologists no longer rely on Haeckel’s work, whereas, according to the Discovery Institute, an organization promoting intelligent design, these drawings persist in a large number, or even a majority, of modern textbooks. These two texts take part in a controversy over evolution and do not necessarily use the same criteria: reproducing a derived image does not mean basing the scientific explanation on the biogenetic law. The analysis of the 240 textbooks confirms material persistence, without assessing the caption or conceptual use of each image.43444536
Teachers also adhere to false ideas about the brain, known as neuromyths. Researchers Sanne Dekker, Nikki C. Lee, Paul A. Howard-Jones and Jelle Jolles of Vrije Universiteit Amsterdam and the University of Bristol surveyed British and Dutch teachers in 2012 for the scientific journal Frontiers in Psychology: average adherence reached 49% of the myths presented. According to the same study, greater general knowledge of the brain predicted greater adherence to neuromyths, and reading popular media was associated both with greater neuroscience literacy and with more false ideas; these are associations observed in a voluntary sample of teachers interested in neuroscience, with no established causality. A Spanish study accompanied by an international comparison, published in the same journal, finds that they are highly prevalent in several countries.46474849
Two open-access reviews concern, respectively, the persistence of these myths and interventions intended to dispel them. According to the authors of a review of 24 articles, the weighted prevalence of neuromyths reached 52.1% among educators, with no decrease observed between 2012 and 2020; belief in visual, auditory and kinaesthetic learning styles reached a weighted average of 89.1% across 37 samples published between 2009 and 2020, with no decline detected. This comparison is based on observational studies with varying questionnaires and populations; it does not demonstrate a causal effect of repetition.5051
Scientific or neuroscience training is associated with lower belief in neuromyths, without eliminating them, according to Psychology Today magazine and Forbes. According to The Conversation, in an Australian survey, only 7% of the educators surveyed correctly rejected the claim about the effectiveness of learning styles. According to the specialized publication Education Week, an international study indicates that most teachers had encountered neuromyths in initial or continuing training rather than in the media; the corresponding primary publication is not identified. Finally, the authors of a multilingual neuromyth diagnostic tool report that linguistic adaptations and structural checks of earlier questionnaires are rarely carried out or reported, which weakens some international comparisons.5253545556
In Quebec, the Ministry of Education regulates textbooks through an approval procedure. According to the ministry, instructional packages are assessed according to ministerial criteria, may be subject to corrections requested from the publisher, and are entered on an official list after approval. However, some resources used in schools do not have this status: the ministry and websites of educational advisers at school service centres indicate that material may be used without being an approved instructional package, even when it claims to comply with the curriculum.57585960
These findings describe errors that persist in textbooks and beliefs that persist among teachers. No published study directly measures the effect on students’ beliefs of the same false claim being repeated in class for years; the link between these forms of persistence and the illusory truth effect therefore remains a hypothesis, not a finding.1151
At school: what is established and what is not
- Established
- Disputed
- Disputed
- Established
- ReportedThe prevalence of neuromyths did not decline between 2012 and 2020.51
Finding from a review of 24 articles, based on observational studies using varying methods.
- Established
- Unknown
Established Shown by verifiable documents or data.
Disputed Contradicted by other documented evidence.
Reported Stated by an identified source, not independently confirmed.
Unknown No available source can establish it.
Even the implausible? A boundary still under debate
According to Hassan and Barber, earlier experiments observed the effect for highly implausible claims, whereas the authors of the 2026 meta-analysis conclude that the increase is not reliable for extremely implausible claims. The categories and protocols are not identical. No general threshold makes it possible to determine at what degree of implausibility repetition ceases to have an effect.27
A measured increase does not necessarily mean a shift to belief. According to the authors of a preprint deposited on the OSF platform and devoted to highly implausible statements, repetition could raise their evaluations, with weaker or uncertain effects at the extreme levels; in an exploratory analysis, about half of the participants showed no effect or an opposite effect. This remains the result of a preprint and a particular protocol.61
Plausibility matters
- Disputed
Disputed Contradicted by other documented evidence.
Bernays: repetition as an old sales method
“The conscious and intelligent manipulation of the organized habits and opinions of the masses is an important element in democratic society. Those who manipulate this unseen mechanism of society constitute an invisible government which is the true ruling power of our country.”62 These are the words with which Edward Bernays opens his book Propaganda, published in 1928 by Horace Liveright.
French translation: the conscious and intelligent manipulation of the organized habits and opinions of the masses is an important element of democratic society. Those who manipulate this invisible mechanism of society constitute an invisible government, which is the true ruling power of our country.62
In the same book, Bernays describes a doctrine according to which repeating an idea would be enough to create a conviction: “It was one of the doctrines of the reaction psychology that a certain stimulus often repeated would create a habit, or that the mere reiteration of an idea would create a conviction.”62
French translation: it was one of the doctrines of reaction psychology that a certain stimulus repeated often would create a habit, or that the mere reiteration of an idea would create a conviction. Bernays links this doctrine to the old sales method, based on advertising reiteration, rather than presenting it as his own principal method.62
The advertiser sought by means of reiteration and emphasis directed upon the individual, to break down or penetrate sales resistance.62
French translation: the advertiser sought, through reiteration and emphasis directed at the individual, to break down or overcome sales resistance.62
Bernays contrasts this approach with action directed at groups and their trusted figures. In his example involving bacon, he describes an approach consisting of bringing doctors into dietary recommendations. He also invokes Trotter, Le Bon and Walter Lippmann in connection with groups, as well as the psychologists of Freud’s school in connection with unconscious desires. These proposals belong to his historical argument; they are not experimental findings on the illusory truth effect.62
Advertising, politics and media: what the examples show
For advertising, the authors of two experiments published in the scientific journal Cognitive Research: Principles and Implications directly test product claims. According to their study, repetition increases perceived truth after a single reading, in contexts containing 20% and 60% advertising content. The effect disappears when participants assess accuracy upon first exposure. This is a test of belief in claims, not merely of brand memorization.63
In political communication, the result may be less linear. According to the authors of the experiment published in the scientific journal International Journal of Communication, repetition of negative political messages may reduce their credibility while worsening attitudes toward the issue presented. Believing a message and reacting negatively to it are therefore distinct measures.64
For the media, repeated broadcasting is not sufficient to establish that several newsrooms repeated an identical formulation. The BBC World Service lists several broadcasts, from July 4 to 6, 2020, of its programme devoted to the false theory linking 5G and coronavirus. However, its presentation page does not publish comparative transcripts that would make it possible to identify the same formulation repeated by independent media outlets.65
The wording matters, because a claim does not always recur in the same terms. Repetition word for word produces the classic effect, and it is the form that the literature has tested most extensively.662711
Consistency of meaning across exposures, by contrast, is necessary: repeating only the theme of a claim or some of its referents is generally insufficient, and repetition whose meaning is incompatible with the initial statement may decrease perceived truth.66767
The scope of a rewording remains debated. According to the authors of the 2024 study published by the scientific publisher Taylor & Francis, repeating the general meaning (gist) and repeating the wording word for word both lead to high and relatively similar levels of belief as long as the information is not debunked, whereas the authors of the study The illusory truth effect requires semantic coherence across repetitions, disseminated on the scientific platform ScienceDirect, report that, in their first three experiments, no manipulation other than repetition word for word significantly changed truth judgments. The manipulations are not the same: the first study repeats the general meaning of information, while the second exposes participants in particular to the theme or to certain referents. A faithful rewording of the meaning could therefore have an effect where simple familiarity with the subject is insufficient, although these two studies do not allow a conclusion to be drawn under a constant protocol.6866
According to the authors of the 2024 study, a correction is also less effective after repetitions of the general meaning than after repetitions word for word.68
Word for word or reworded
- Established
- Established
- DisputedNonliteral repetition produces the same effect as repetition word for word.6866
The 2024 study reports similar beliefs after repetition of the general meaning or word for word; the study on semantic coherence finds no significant effect from its other manipulations in its first three experiments. The manipulations being compared differ.
- ReportedA correction is less effective after repetition of the general meaning than after repetition word for word.68
Finding reported by the authors of the 2024 study.
Established Shown by verifiable documents or data.
Disputed Contradicted by other documented evidence.
Reported Stated by an identified source, not independently confirmed.
What these examples do not demonstrate
- UnknownThe BBC programme case establishes word-for-word repetition by several independent media outlets.65
The page documents broadcasts of a programme, without textual comparison between media outlets.
- UnknownA claim repeated by several apparently independent sources produces a stronger effect than one repeated by a single source.67
The study on source credibility and simulated publications does not describe this direct comparison.
Unknown No available source can establish it.
What can reduce the effect
Directing attention to accuracy from the first encounter is an approach supported by several experiments. According to the authors of the study published in 2020 in Cognitive Research: Principles and Implications, two experiments involving a total of 575 participants found an effect after an initial evaluation of the headlines’ interest, but not after an initial evaluation of their truthfulness. The authors of the advertising study found this protection with other material.6963
Not all warnings are equal. According to the authors of the 2026 meta-analysis, a label explicitly indicating that content was false eliminated the effect in the studies reviewed, whereas Henderson, Simons and Barr report its persistence despite certain warnings or prior indications of truthfulness. The authors of a study on individual differences published in Psychonomic Bulletin & Review, for their part, indicate that warnings can mitigate the effect without eliminating it. A general warning, information provided beforehand and a label attached to the claim are different interventions. Their effectiveness cannot be treated as equivalent.7970
Naming a source does not guarantee that the effect will disappear either. According to the authors of the four experiments published in 2020 in Cognitive Research: Principles and Implications, source credibility and prior exposure have additive effects: repetition retains an independent influence. The authors of the 2026 meta-analysis also emphasize that the usefulness of a reliability indication may depend on whether it is remembered at the time of judgment.677
Verification can nevertheless correct beliefs. According to researchers Xingyu Liu, Li Qi, Laurent Wang and Miriam J. Metzger, in their 2023 scientific study published by SAGE, belief decreases on average after verification, and the perceived credibility of the verifier is associated with better correction. This result does not establish that simply displaying the verifier’s name neutralizes repetition.71
Correction does not mean that reminders of the error should always be avoided either. According to the authors of the scientific article Can corrections spread misinformation to new audiences?, three experiments involving a total of 1,718 participants did not find an overall paradoxical reinforcement of false ideas through familiarity.72
Real but conditional protections
- Established
- Disputed
- Unknown
Established Shown by verifiable documents or data.
Disputed Contradicted by other documented evidence.
Unknown No available source can establish it.
| Study | Authors and institutions | Declared funding | Declared links or interests |
|---|---|---|---|
| 2026 meta-analysis (Nature Communications) | Steeven Ye, David Attali, Maria Ghazi, Arnaud Cachia, Mathieu Cassotti and Grégoire Borst; Université Paris Cité and CNRS | French National Research Agency (ANR), FakeAd project, ANR-21-CE28-0025; CIFRE grant administered by ANRT, a doctoral arrangement conducted partly within a company, company not named in the article | Declared affiliation with Ipsos (Steeven Ye) and Éditions Nathan, a school textbook publisher (Maria Ghazi); three authors signed a study included in the meta-analysis, which they disclose |
| 2021 longitudinal study (Journal of Cognition) | Emma L. Henderson, Daniel J. Simons and Dale J. Barr; article deposited in the University of Glasgow’s institutional repository | Kingston University doctoral grant; grant from the European Association of Social Psychology (EASP); research funds from Kingston University and Cardiff University | No competing interests declared |
| 2012 survey on neuromyths (Frontiers in Psychology) | Sanne Dekker, Nikki C. Lee, Paul A. Howard-Jones and Jelle Jolles; Vrije Universiteit Amsterdam and University of Bristol | No funding declared | Absence of declared conflict of interest |
| Study on truth judgments across development, 2020 (Psychological Science) | Lisa K. Fazio and Carrie L. Sherry | Not reviewed (paywalled article) | Not reviewed (paywalled article) |
| Foundational 1977 experiment (Journal of Verbal Learning and Verbal Behavior) | Lynn Hasher, David Goldstein and Thomas Toppino; Temple University and Villanova University | No funding mentioned in the article | No mention |
| 2006 article on trust in testimony (Child Development) | Paul L. Harris and Melissa A. Koenig | NICHD F32 HD42860 grant, from the National Institute of Child Health and Human Development, an institute of the US National Institutes of Health, awarded to Melissa A. Koenig | Not reviewed |
| Study on preschool children’s suggestibility | Stephen J. Ceci, Cornell University, and colleagues | Partial funding from the National Science Foundation, the US federal research agency | No mention |
| Working paper on the cognitive processes underlying the truth effect | Stump, Schumacher, Voss and Klauer; University of Freiburg | Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), GRK 2277 program, “Statistical Modeling in Psychology” | Not reviewed |
Declaring funding or an affiliation says nothing, by itself, about the quality or direction of a study. This information is provided so that readers know who produced and supported the research.
The limits of a laboratory effect
The authors of the systematic mapping published in Psychonomic Bulletin & Review emphasize a body of literature largely based on Western convenience samples, word-for-word repetitions and short delays. They also point to a lack of accessible data, preregistrations and external replications. The authors of the 2026 meta-analysis find substantial heterogeneity and detect an underrepresentation of small studies with negative or non-significant results. This meta-analysis includes authors affiliated with Ipsos and Éditions Nathan, a school textbook publisher, as indicated in the table above.117
Context can itself modify the effect. According to researchers Reed Orchinik, David G. Rand and Rahul Bhui of the MIT Sloan School of Management, in a preprint, a preregistered experiment involving 4,947 participants found a markedly reduced effect in an environment where most information was false. This result must be considered in light of the document’s preprint status.74
What repetition proves, and what it does not prove
For readers, the useful distinction remains the one between recognizing a sentence and having evidence. The experiments show that a first impression can influence judgment; they also show that directing attention to accuracy can reduce this influence. They do not make it possible to infer the truth of a frequently heard statement, or the intentions of those who repeat it.36963