Why Fifteen Scandals Persuade Less Than One

The article explains the paradox of why many scandals strengthen rather than weaken a politician, drawing on nine psychological mechanisms – the dilution effect, cognitive overload, the conjunction fallacy, narrative transportation, and psychic numbing. It shows that the human brain processes a flood of accusations heuristically and intuitively, which erodes their persuasiveness, whereas a single concrete story with an identifiable victim persuades more effectively than fifteen factual points.

When dozens of serious, fact-backed accusations pile up against a politician, the public paradoxically stops believing — not the politician, but the accusations themselves. This seemingly perverse phenomenon is not a sign of voter stupidity nor of society's moral failure. It is the consequence of deep cognitive mechanisms that evolutionary psychology has documented for decades. Understanding these mechanisms is essential for anyone who wants to communicate effectively about corruption and the abuse of power in a democratic society — from investigative journalists through analysts to civil society.

The example is within arm's reach in any democratic country. Imagine a politician who, over a single decade, faces accusations of subsidy fraud, money laundering through foreign tax structures, collaboration with the secret service of the former regime, a massive conflict of interest, and the takeover of private media. And with every new revelation, his electoral results rise. Then imagine another politician, caught with a briefcase full of cash — a single affair that ends his career instantly. One scandal destroys a career. Fifteen scandals create a prime minister.

This article maps nine interconnected psychological phenomena that explain this paradox, and offers concrete strategies for countering it. Every claim is backed by original studies, meta-analyses, and experimental evidence — including an honest assessment of what the replication crisis has left standing of the original findings and what it has called into question.

The most direct explanation of the accumulation paradox is offered by the dilution effect, described by Richard Nisbett, Henri Zukier, and Ronald Lemley in 1981 in the journal Cognitive Psychology. In a series of experiments, participants were given either only diagnostic information (data that genuinely predicted the outcome) or a combination of diagnostic and non-diagnostic information (data with no predictive value). The result was unambiguous: adding meaningless information to meaningful information systematically weakened participants' predictions. People who knew only a student's grade-point average predicted his academic success more strongly than people who knew the grade-point average plus his favorite color.

The mechanism lies in the representativeness heuristic — people judge probability by comparing a target with a category prototype. Non-diagnostic information reduces the perceived similarity between the target and the category, and thus "dilutes" the impression.

Crucially, this effect is not confined to meaningless data. Philip Tetlock and Richard Boettger demonstrated in 1989 in the Journal of Personality and Social Psychology that a sense of accountability amplifies the dilution effect rather than weakening it. People who knew they would have to defend their judgment tried to incorporate all available information — but were unable to distinguish the diagnostic from the non-diagnostic. The more conscientiously they approached the evaluation, the more they were swayed by ballast. The paradox of the responsible citizen: the effort to be thorough leads to worse judgment.

The dilution effect has been replicated across dozens of domains — from consumer decision-making (Simonson et al., 1993) through forensic auditing (Hoffman and Patton, 1997; Hackenbrack, 1992) and negotiation (Wiltermuth and Neale, 2011) to juror decision-making (Smith, Stasson, and Hawkes, 1998) and the evaluation of moral character (McHugh et al., 2025). It is among the most robust findings in the psychology of judgment.

Important boundaries of the effect were identified by Ellen Peters and Myron Rothbart (2000) in the Personality and Social Psychology Bulletin: what matters is the typicality of the non-diagnostic information. If the non-diagnostic data increase similarity to the diagnostic category, the dilution effect disappears or even reverses. And LaBella and Koehler (2004) showed that when people revise an existing judgment (rather than forming a new one), dilution can turn into a confirmation effect. The dilution effect is therefore not universal — but in a scenario where the media present a mixture of strong and weak accusations simultaneously, the conditions for its operation are nearly ideal.

In the language of political scandals: when an investigative report contains three devastating accusations and seven weaker or merely contextual ones, the weaker accusations do not reinforce but rather weaken the impression made by the strong ones. The brain of the average recipient processes them as a whole, and the resulting impression is less extreme than if he had heard only the single strongest affair.

Herbert Simon, the later Nobel laureate in economics (1978), formulated in 1955 in the Quarterly Journal of Economics the concept of bounded rationality: the human mind does not have the capacity for objectively rational behavior in a complex environment, and therefore it "satisfices" — it chooses the first acceptable solution rather than the optimal one. The famous scissors metaphor: one blade represents cognitive limitations, the other the structure of the environment. Rationality depends on the fit between the two.

Daniel Kahneman developed this idea into a dual-system model in his book Thinking, Fast and Slow (2011). System 1 works quickly, automatically, emotionally — and handles the overwhelming majority of cognitive activity. System 2 is slow, analytical, effortful — and has sharply limited capacity. When System 2 is overloaded, tired, or under pressure, System 1 takes control with its full arsenal of heuristics and biases.

How this manifests in the context of persuasion was described by Richard Petty and John Cacioppo in the Elaboration Likelihood Model (ELM, 1986). The model distinguishes the central route (careful evaluation of arguments) and the peripheral route (reliance on simple cues — who is saying it, how many arguments there are, how it looks). The critical insight: when the ability or motivation to process information declines, people switch from the central to the peripheral route. They stop evaluating the quality of arguments and begin reacting to their number, to the credibility of the source, to the emotional tone.

Kiesler and Mathog (1968) directly demonstrated that the source-credibility effect was significant only under conditions of high distraction — under low cognitive load, people genuinely processed the arguments and source credibility ceased to matter. Complementarily, Shelly Chaiken (1980), in the heuristic-systematic model, showed that under conditions of low processing, people use simple rules such as "experts are right" or "length means strength."

The practical consequence for the accumulation of scandals is devastating. Fifteen accusations at once overload the average citizen's System 2. The brain switches to peripheral processing. Instead of analytically evaluating each accusation, it begins reacting to cues: Who is saying it? (The opposition — untrustworthy.) Is it complicated? (Yes — probably propaganda.) Is it too much? (Yes — probably exaggerated.) It is precisely at this moment that the key defense mechanism is activated.

It should be added that while the ELM framework has extensive empirical support, the exact three-way interaction pattern from the original study by Petty, Cacioppo, and Goldman (1981) proved difficult to replicate (Chung and Park, 2018, Annals of the International Communication Association). The general principle, however — that cognitive overload shifts processing toward heuristics — is supported by converging evidence from many research traditions.

When the brain, switched into System 1, detects the pattern "too many accusations against one person," it may automatically activate a pre-set cognitive schema — persecution, conspiracy, witch hunt. This process is related to several well-documented mechanisms.

Proportionality bias: Jan-Willem van Prooijen and Eric van Dijk (2014), in five studies published in the Journal of Experimental Social Psychology, demonstrated that people have a basic heuristic — a big consequence must have a big cause. LeBoeuf and Norton (2012) found that participants were more willing to believe in a conspiracy behind an assassination if they were told that its consequences were more serious (75% versus 64%). In the context of political scandals: a large number of simultaneous accusations "feels" disproportionate — there must be something big, coordinated, and deliberate behind it. Not evidence of corruption, but an organized campaign.

Hyperactive pattern detection: Van Prooijen and Van Vugt (2018, Perspectives on Psychological Science) summarized that belief in conspiracies correlates with the tendency to perceive patterns in random stimuli (r = 0.22–0.45) and with the over-detection of intentional agency (r = 0.16–0.42). The strongest predictor of belief in one conspiracy theory is belief in others — even mutually contradictory ones (Wood, Douglas, and Sutton, 2012; positive correlation r = 0.14–0.26). Conspiratorial thinking is not a specific reaction to particular evidence but a cognitive style.

Political extremism amplifies the effect: Van Prooijen, Krouwel, and Pollet (2015, Social Psychological and Personality Science) found a quadratic relationship between the intensity of political ideology and conspiracy beliefs — both the extreme left and the extreme right believe in conspiracies more than the center. This means that it is precisely the most loyal voters of the accused politician who are most prone to interpret the accumulated accusations as an organized conspiracy.

The mechanism can be summarized: accumulation of accusations → proportionality mismatch ("so many at once cannot be natural") → pattern detection → activation of the "coordinated attack" schema → intensified search for intent ("who is behind this?") → rejection of the accusations as a whole. The entire chain runs in System 1, automatically and without conscious effort.

Amos Tversky and Daniel Kahneman, in 1983 in Psychological Review, published the famous "Linda" experiment: a majority of participants rated the conjunction of two events as more probable than one of them alone — a violation of the basic rule of probability theory, P(A∧B) ≤ P(A). This conjunction fallacy persisted even among statistically educated participants and was replicated in geopolitical forecasting, medicine, law, and gambling.

The classic conjunction fallacy shows that people mistake representativeness for probability — a detailed, coherent story seems more plausible even when it is mathematically less probable. But its symmetric corollary is equally important: a description of events that requires many claims to be simultaneously true becomes intuitively ever less probable, even if each claim is independently acceptable.

Consider: "The politician embezzled subsidies, laundered money through foreign structures, collaborated with the secret service, exploited a conflict of interest, and manipulated the media." If the subjective probability of each claim is 70%, the conjunction of all five — assuming independence — is a mere 16.8%. The brain does not perform this calculation explicitly, but at an intuitive level it registers that "for all of this to be true, too many things would have to come together at once." The result: a paradoxical feeling that the more complex accusation is less credible than the simpler one.

The conjunction fallacy can be significantly weakened by frequency framing (from 85% down to roughly 20% error rate) and by financial incentives (from 58% to 33%; Charness, Karni, and Levin, 2010). But in ordinary media consumption these conditions do not hold — citizens judge political descriptions of events intuitively, not on the basis of Bayesian calculation.

The term "Gish Gallop" was coined in 1994 by the physical anthropologist Eugenie Scott, founder of the National Center for Science Education in the United States, after the creationist debater Duane Gish, who in debates spewed out dozens of arguments so quickly that his opponent had no chance to refute them. The technique exploits the asymmetry captured in Brandolini's law: "The amount of energy needed to refute nonsense is an order of magnitude greater than the energy needed to produce it."

But what happens when the arguments are valid? Research shows that the form of the message — a flood — triggers a defensive reaction independent of the quality of the content. Christopher Paul and Miriam Matthews of the RAND Corporation (2016) described the "firehose of falsehood" model with four features: high volume, speed and continuity, a lack of commitment to reality, and a lack of commitment to consistency. The key psychological finding: when audience interest is low, a message's persuasiveness depends more on the number of arguments than on their quality — the peripheral route of the ELM in action.

But paradoxically it also works in reverse. A study published in Social Cognition (2022) demonstrated that as the number of generated arguments rises, perceived quality declines — participants who came up with twelve arguments rated their average quality lower than those who came up with three. And only quality, not quantity, predicted attitudes. Pelham, Sumarta, and Myaskovsky (1994) identified the "numerosity heuristic" — the tendency to equate "a lot" with "very important" — but it operates only under conditions of low elaboration. Under careful scrutiny, adding weak arguments reduces the persuasiveness of the whole.

There is also a remarkable asymmetry between positive and negative persuasion. Research shows that in positive persuasion, more claims lead to stronger conviction, but in negative persuasion the number of claims has no significant effect — negative messages "saturate" more quickly. This directly implies that the accumulation of accusatory (negative) messages hits a persuasion ceiling sooner than positive communication does.

Melanie Green and Timothy Brock published in 2000 in the Journal of Personality and Social Psychology the concept of narrative transportation — absorption in a story involving imagery, emotion, and concentration. In four experiments they demonstrated that greater absorption increases conviction in line with the story, improves the evaluation of protagonists, and — critically — reduces critical thinking. Transported readers found fewer flaws in the story than non-transported ones.

The meta-analysis by Braddock and Dillard (2016, Communication Monographs; 37 studies, N = 7,376) quantified the effect: the correlation of narrative persuasion with beliefs r = 0.17, with attitudes r = 0.19, with intentions r = 0.17, and with behavior r = 0.23. The effect is "small to moderate, but meaningful" — and stable across media and research procedures.

Why one story is more effective than fifteen facts is explained by the identifiable victim effect and psychic numbing, the work of Paul Slovic. Deborah Small, George Loewenstein, and Paul Slovic, in a key experiment (2007, Organizational Behavior and Human Decision Processes), demonstrated that donations to a seven-year-old girl from Mali averaged $2.38, whereas donations to statistically described millions of starving people averaged a mere $1.14. And, even more importantly: adding statistics to the girl's story reduced donations from $2.38 to $1.43. The statistics did not strengthen the emotional appeal — they undermined it. Analytical processing suppressed empathy.

It must be noted, however, that a pre-registered replication (Maier, Wong, and Feldman, 2023, Collabra: Psychology) did not reproduce the identifiable victim effect (ηp² = 0.000). The authors pointed to possible publication bias in the earlier literature. The general principle — that emotions respond more strongly to concrete individuals than to abstract statistics — has broader support, but the exact size of the effect remains a matter of debate.

Slovic's formula is nevertheless intuitively compelling: the emotional response is maximal at N = 1, begins to decline at N = 2, and collapses at higher values, which become "mere statistics."

In the context of scandals: a single story about a concrete victim — an entrepreneur ruined by the unfair competition of the politician's conglomerate, a municipality that lost subsidies because of fraud — persuades more than a fifteen-point list of affairs. The list activates analytical processing, which suppresses the empathic reaction. The individual story enables narrative transportation, which reduces critical thinking and increases persuasiveness.

Brendan Nyhan and Jason Reifler published in 2010 in Political Behavior a study in which participants exposed to a factual correction strengthened their original mistaken belief — instead of correcting it. The "backfire effect" became one of the most cited concepts in the discussion of "post-truth" politics (over 9,000 citations).

But the science has moved on. Thomas Wood and Ethan Porter (2019, Political Behavior), in five experiments with more than 10,100 participants, tested 52 topics — and did not find a single case in which a correction produced a backfire effect. Even on the originally problematic topic the effect disappeared once the correction was worded more simply.

And in 2021 Nyhan himself published in the Proceedings of the National Academy of Sciences an article in which he acknowledged that the scientific literature — including his own subsequent research — does not support the interpretation that backfire effects explain the spread and durability of mistaken beliefs. Numerous replication attempts failed (Guess and Coppock, 2018; Ecker et al., 2021; Haglin, 2017; Swire et al., 2017; Wood and Porter, 2019).

The current expert consensus: the backfire effect is real, but extremely rare and fragile — not a general property of human cognition. People usually update their factual beliefs when corrected, even on politically charged topics. What corrections often do not change, however, are attitudes and political preferences — a person may accept the facts but not change their voting behavior.

What is, by contrast, highly robust is belief perseverance. Lee Ross, Mark Lepper, and Michael Hubbard (1975, Journal of Personality and Social Psychology) demonstrated that participants who were given fabricated feedback about their performance and subsequently fully informed of the deception continued to rate themselves in line with the original false feedback. The beliefs had taken on "a life of their own" — participants had spontaneously generated causal explanations, and even after the data were discredited, the explanations persisted.

This is complemented by motivated reasoning (Ziva Kunda, 1990, Psychological Bulletin; 9,000+ citations): people arrive at the conclusions they want to arrive at, using a biased selection of cognitive strategies to do so. Kunda showed that motivation does not operate outside of reason — but manipulates which evidence it summons from memory, how it constructs that evidence, and how it evaluates it. An important condition is that the desired conclusion must be defensible — a person accepts it only if they can construct a seemingly reasonable justification.

In the context of accumulated accusations, motivated reasoning works like this: the loyal voter wants to believe that their candidate is innocent. The more accusations they receive, the more material they have at their disposal for selective criticism — in one accusation they find weak methodology, in another a dubious source, in a third suspicious timing. The accumulation of accusations paradoxically gives the motivated critic more ammunition for rejecting the whole.

The phenomenon of scandal fatigue gained empirical support in a study by Staffan Kumlin and Peter Esaiasson (2012, British Journal of Political Science), who analyzed data from Western European elections over the period 1977–2007. They found that elections accompanied by scandal were exceptional at the end of the 1970s, but today nearly half are affected by them. And as the frequency of scandals rose, their negative effect on satisfaction with democracy faded. The key finding: it is not the citizens who are to blame for the fatigue, but the character of the scandals — only "highly generalizable" scandals (involving multiple politicians and parties) retain their impact.

Operating in parallel are habituation — the basic psychological principle of reduced response to a repeated stimulus — and psychic numbing in Slovic's sense. With each additional scandal, the emotional intensity of the response declines. Outrage fatigue is documented in the context of social media (Zheng and Lee, 2022, New Media & Society) and correlates with reduced willingness to take political action.

Diane Vaughan, in her book The Challenger Launch Decision (1996), described the normalization of deviance: the process by which a deviation from correct behavior becomes culturally normalized because it does not immediately lead to catastrophe. In politics, this means that repeated exposure to scandals shifts the boundary of what is perceived as scandalous. The Brennan Center for Justice (2025) describes how brazenness distorts the perception of which behavior is corrupt and reinforces the public's belief that corruption is a normal part of politics.

The illusory truth effect also plays a role (Hasher, Goldstein, and Toppino, 1977; 2025 meta-analysis: g = 0.37 across 182 studies and N = 31,184): repeated claims are perceived as truer. But this effect works in both directions — it increases the perceived truthfulness of both the accusation and the defense. And in a situation where the politician repeats his defense ("it's a political campaign") more often than the media repeat the accusation, the illusory truth effect paradoxically serves the defense more.

The "everyone steals" syndrome is the logical culmination. Sociological surveys in the Czech Republic repeatedly show that around three quarters of citizens believe that "almost all or most public officials" are involved in corruption (survey by Masaryk University, 2012–2013). Corruption is understood as a natural property of the entire political class without distinction — this "hollowing out" of the anti-corruption concept complicates a genuine anti-corruption agenda. Cynicism becomes a breeding ground for authoritarianism: if everyone steals and lies, all that remains is to call for a strong leader who will impose order.

CVVM data for 2024 confirm the depth of the problem: the government was trusted by 17–24% of Czechs depending on the survey period, the Chamber of Deputies by 18–23%, and 10–13% of citizens were satisfied with the political situation. Meanwhile, local politicians (mayors) were trusted by 66–68%. The gulf between local and national trust is a direct symptom of the "everyone steals" syndrome at the national level.

The credibility of a message depends not only on its content but on the perception of its source. Carl Hovland and Walter Weiss (1951, Public Opinion Quarterly), as part of the Yale persuasion research program, demonstrated that highly credible sources persuade significantly more than low-credibility ones — even with identical message content. More recent research shows that this effect is strongest precisely under conditions of low cognitive processing — that is, exactly in a situation of information overload.

The hostile media effect (Vallone, Ross, and Lepper, 1985, Journal of Personality and Social Psychology) showed that both pro-Israeli and pro-Arab students rated identical television news coverage as biased against their own side. Both groups saw the same report — and both perceived it as hostile. The effect has been reliably replicated over thirty years across topics and cultures (Perloff, 2015).

Leslie John, Hayley Blunden, and Heidi Liu (2019, Journal of Experimental Psychology: General), in eleven experiments, documented "shooting the messenger": people rate the bearer of bad news as less likeable, even when the messenger bears no responsibility for the news. The effect is stronger when the bad news is unexpected or difficult to explain.

In the context of political scandals, the combination of these effects is especially toxic. When the accusation is brought by media that supporters of the accused politician consider hostile, the hostile media effect and shooting the messenger are activated simultaneously. The content of the message is secondary; the reaction to the source is primary. As Cruz and Carpenter (2024) showed, the perception of a source's credibility depends on ideological closeness, not on actual expertise — a local radio host may be perceived as more competent than a respected professor if he shares the recipient's ideology.

Understanding the mechanisms is the first step. The second is overcoming them. Research offers several empirically grounded strategies that investigative journalists, analysts, and civil society can apply.

William McGuire (1961, 1964) developed inoculation theory by analogy with vaccination — a weak dose of counterarguments creates "mental antibodies." The meta-analysis by Banas and Rains (2010, Communication Monographs; 54 tests) confirmed that inoculation messages are superior both to supportive messages and to control conditions. Inoculation provides protection even against new, unforeseen arguments (cross-protection). The effect persists over the medium term but requires renewal after roughly two weeks.

Sander van der Linden and colleagues (2017, Global Challenges) successfully applied inoculation against climate disinformation. Jon Roozenbeek and van der Linden (2022, Science Advances), in a study with N = 6,464 in the lab and N = 22,632 on YouTube, demonstrated that ninety-second inoculation videos improve the ability to recognize manipulation, increase self-confidence, and improve sharing decisions. Google/Jigsaw deployed these videos as YouTube ads for millions of users.

In practical terms for journalists, this means: before publishing an investigation, preemptively refute predictable defensive strategies. Explain to readers: "Politicians accused of corruption typically respond by deflecting attention, attacking the journalists, and claiming an organized campaign." This technique activates cognitive immunity — when the politician later actually uses these tactics, the reader recognizes them as predicted manipulations rather than legitimate defense.

The Panama Papers model — a coordinated release of a strong opening package followed by the gradual release of further revelations over weeks and months — maintains narrative momentum without overloading the audience. The Watergate model — the gradual peeling back of the layers of a scandal over two years — allowed each revelation to build on the previous one. By contrast, the "everything at once" model activates the dilution effect, cognitive overload, and the conspiracy schema.

Allen and Preiss (1997), in a meta-analysis of 15 studies, found that the combination of narrative and statistical evidence is more persuasive than either alone. Zebregs et al. (2015) refined this: statistics act more strongly on beliefs and attitudes (the cognitive level), stories on behavioral intentions (the emotional level). In practice: begin with a concrete story of a concrete victim (narrative transportation activates empathy), then anchor it with systemic data (scope, sums, number of cases).

Sam Wineburg and Sarah McGrew (2017, 2019) compared how professional fact-checkers, historians with PhDs, and Stanford students assessed the credibility of websites. The historians and students read vertically — examining the URL, the logos, the "About us" section. The fact-checkers read laterally — they immediately left the page and searched for what trustworthy outside sources said about it. The fact-checkers reached more accurate conclusions in a fraction of the time. Journalists can model lateral reading by transparently showing how they verified their sources.

Lord, Lepper, and Preston (1984, Journal of Personality and Social Psychology) demonstrated that the explicit instruction to "consider the opposite" reduces the biased assimilation of evidence — and is more effective than the general instruction to "be objective." The strategy also works against anchoring (Mussweiler, Strack, and Pfeiffer, 2000), hindsight bias (Arkes et al., 1988), and overconfidence (Koriat et al., 1980). For journalists and analysts: build explicit prompts into the text for readers — "Before you reject these accusations, try to imagine: what would you think if the same evidence were directed against a politician you dislike?"

The accumulation paradox — the more evidence, the less persuasive — is not an anomaly. It is the natural consequence of a cognitive architecture that evolved for an environment radically different from the modern media ecosystem. The dilution effect weakens strong accusations with an admixture of weaker ones. Cognitive overload switches the brain to heuristics. The conspiracy schema is activated by proportionality bias. The conjunction fallacy intuitively reduces the credibility of complex descriptions of events. Narrative transportation favors one story over fifteen facts. Psychic numbing weakens empathy as the scale grows. Motivated reasoning gives defenders more ammunition with every additional accusation. Scandal fatigue normalizes deviance. And shooting the messenger discredits the source instead of the message.

None of these mechanisms testifies to the public's stupidity — each is a rational adaptation to limited cognitive resources in an information-saturated environment. But their cumulative effect is toxic to democratic accountability.

The key takeaway for investigative practice is unambiguous: less is more. A single, precisely documented affair, communicated through narrative, gradually, and with preemptive refutation of defensive strategies, has a higher chance of breaking through cognitive defenses than fifteen affairs at once. David defeated Goliath with a single stone. Fifteen stones at once would have flown in all directions — and Goliath would simply have laughed.

But there is also an optimistic message from the research. The backfire effect is far rarer than was believed — people are mostly able to update their factual beliefs. Inoculation techniques demonstrably work across the ideological spectrum. And lateral reading can be taught relatively easily. The problem is not that citizens are unpersuadable. The problem is that we present information to them in a way that activates precisely the cognitive mechanisms we want to overcome. The solution does not lie in shouting louder — but in a smarter architecture of the message.

This article draws on original studies and meta-analyses published in peer-reviewed journals (Cognitive Psychology, Psychological Review, Journal of Personality and Social Psychology, Political Behavior, Perspectives on Psychological Science, PNAS, Communication Monographs, Science Advances, and others). For findings affected by the replication crisis (the backfire effect, the exact three-way interaction pattern of the ELM, the identifiable victim effect), the current state of the evidence is explicitly stated. Czech data come from CVVM and surveys by Masaryk University.

Dilution effect — Nisbett, Zukier, and Lemley, 1981 — robust, dozens of replications.

Bounded rationality — Simon, 1955; Kahneman, 2011 — foundation of the field.

Conspiracy schema — Van Prooijen and Van Vugt, 2018 — active research area.

Conjunction fallacy — Tversky and Kahneman, 1983 — robust, moderators refined.

Gish Gallop / firehose effect — Paul and Matthews, 2016 (RAND) — descriptive model.

Narrative transportation — Green and Brock, 2000 — meta-analysis: r = 0.17–0.23.

Backfire effect — Nyhan and Reifler, 2010 — called into question (Wood and Porter, 2019).

Belief perseverance — Ross, Lepper, and Hubbard, 1975 — robust.

Motivated reasoning — Kunda, 1990 — robust, 9,000+ citations.

Scandal fatigue — Kumlin and Esaiasson, 2012 — empirical support.

Normalization of deviance — Vaughan, 1996 — case study, widely cited.

Inoculation — McGuire, 1961; van der Linden, 2022 — meta-analysis and field experiments.

Lateral reading — Wineburg and McGrew, 2017 — active pedagogical research.

Transparency of creation

The conception, structure, and editorial line of the article are the work of the author, who prepared the content outline, established the key theses, and directed the entire creative process. Generative AI (Claude Opus 4.6, Anthropic) was used as a tool for research, fact-checking, and elaborating the author's draft.

The author verified the key findings and approved the final wording. No part of the text was published without conscious authorial control. Factual data were verified against the publicly available sources cited in the text.

This approach complies with the transparency principles of EU Regulation 2024/1689 (AI Act). #poweredByAI