Do Real and Invented Scandals Fuel Beliefs in Conspiracy Theories?

Over the past 20 years, conspiracy theories have come to invade the public arena, notably the internet and social media. Whereas a Google search with the keywords “conspiracy theory” achieved 10 million hits in 2013 (Brotherton, 2013), in 2019 it reached about 15 million (Wagner-Egger, 2022), and in 2022, after the pandemic, more than 16 million. Following the health concerns of COVID-19, the often so-called pandemic of misinformation has been overwhelming (e.g., van Mulukom et al., 2020) and worried both scientific institutions and governments (e.g., Burton-Jeangros et al., 2021).

Psychologists have investigated conspiracy theory beliefs extensively since the 2010s. According to a classical definition (Keeley, 1999), conspiracy theories explain specific historical events as the significant causal agency of a relatively small group of persons acting secretly with malevolent intentions. However, Wagner-Egger (2022) observed that conspiracy theories are not simple alternative explanations to the official version; they are severe accusations of conspiracy without sufficient proof. These insufficient proofs are errant data or apparent anomalies of the official version, usually found on unverified alternative websites. This religion of conspiracy (i.e., belief without sufficient proof) radically differs from what may, by analogy, be called the science of conspiracy (Wagner-Egger, 2022). Real conspiracies such as the Watergate affair or the tobacco industry were not discovered by conspiracy theorists online on alternative websites but by professional investigators (e.g., investigative journalists and prosecutors) who collected direct evidence of the conspiracy (e.g., confessions, official documents, or emails) validated in court (Wagner-Egger et al., 2019).

Scholars have attempted to summarize the predictors of conspiracy theory beliefs. Wagner-Egger (2022) identified five categories of psychological and social variables in the literature which could predict beliefs in conspiracy theories. These five categories can be summarized in three families: sociopolitical factors (i.e., with societal-political and social subdimensions), psychological factors (i.e., with personal and cognitive subdimensions), and communicational factors. The societal-political factors refer to the fact that people from backgrounds characterized by lower SES, income, education, higher insecurity, who experience feelings of anomie (i.e., distrust of authorities, loss of political control, etc.) or suffer from higher objective and perceived economic inequalities adhere more strongly to conspiracy theory beliefs. The social dimension refers to the possible intergroup use of conspiracy theories. Every group, party, or country may accuse their opponents, outgroups, minorities, etc., of conspiration. Because conspiracy theories consist of severe accusations of conspiracy (i.e., secret malevolent action by a small group of people) without sufficient proof (Wagner-Egger, 2022), they are instrumental in creating hostile intergroup relations. Psychological factors include personality dimensions (such as paranoia, anxiety, narcissism, etc.) and cognitive dimensions (such as cognitive biases, intuitive thinking, paranormal beliefs, pseudoscience acceptance, etc.). Communicational factors include the narrative attractiveness of conspiracy theory beliefs and the massive spreading of rumors allowed by the internet and social media.

However, to our knowledge, a crucial cause has ever been examined. Conspiracy believers themselves often claim they believe in conspiracy theories because real scandals and conspiracies have occurred in history, such as the Watergate affair or the Tuskeegee experiments (Butter & Knight, 2020). This support of beliefs based on true affairs is also claimed by scholars who defend conspiracy theories as being neither irrational nor pathological (e.g., Harambam, 2020), despite the empirical evidence summarized above. Ironically, the social criticism of psychological research is itself conspiratorial: Claiming that psychological research “pathologizes” conspiracy theory beliefs (e.g., Basham & Dentith, 2016; Harambam, 2020) – although this is the result of dozens of independent empirical studies that could have delivered the opposite results – means attributing a negative human intention where there is none (Douglas et al., 2015).

Inferring new conspiracy theories from past real conspiracies is not sound reasoning. For example, the fact that some past pharmaceutical scandals occurred does not automatically raise the risk of a new pharmaceutical scandal (perhaps the firms have become more responsible, increased their controls, etc.). And even if it were the case, one must evaluate this risk by considering the numerous valuable drugs the pharmaceutical companies sell (the base rate). Thus, conspiracies may not be “thought” or “suspected” from past cases, but professionally investigated. Yet it is probably true that real scandals fuel a lack of confidence in firms and authorities, and then irrational (i.e., uninvestigated but only believed) conspiracy theories.

This idea that real scandals fuel conspiracy theories has never been empirically tested. Oliver and Wood (2014) observed that, in 3 out of 5 conspiracy theories, political knowledge negatively correlates with beliefs. Béna and colleagues (2019) observed that, understandably, people believe more conspiracy theories they have more heard about (the same result may be found in the data of Marques et al., 2021, in which awareness of conspiracy theories correlates with beliefs in conspiracy theories). Whereas these studies investigate close factors (proper political knowledge, understanding of conspiracy theories), none have investigated the specific effect of real (vs. invented) scandals on conspiracy theory beliefs. That is the purpose of this research.

To this end, we asked two independent groups of participants to say whether they did or did not know about, and comment briefly on, either real political-scientific-economic scandals (i.e., having really happened, such as the fake scientific studies from the tobacco industry or invented political-scientific-economic scandals (such as a Swiss bank that is supposed to have invested sustainable funds in fossil fuel. After this task, we measured conspiracy theory beliefs. In two other independent groups, we first measured conspiracy theory beliefs and then asked participants to say whether they did or did not know about those scandals. We then asked them to comment briefly on the same real or invented political-scientific-economic scandals. We hypothesized that real (H1a) and invented (H1b) scandals would increase conspiracy theory beliefs. We also predicted that the more participants adhere to conspiracy theory beliefs, the more they report knowing about real (H2a) and invented (H2b) scandals.

Method

Results

We first tested hypotheses that real (H1a) and invented (H1b) scandals would increase conspiracy theory beliefs by running a 2 (Order: Scandals-CTs vs. CTs-Scandals) × 2 (Type of Scandal: Real vs. Invented) between-subject ANOVA. Contrary to our hypothesis, we observed no main effects of order, F(1, 479) = 0.14, p = .710, η_p² < .001. Conspiracy Theory Beliefs were not higher after scandals (M = 3.20, SE = 0.08) than before (M = 3.24, SE = 0.07). There were no main effects of Type of Scandal, F(1, 479) = 2.97, p = .086, η_p² = .006, nor interaction effects between Order and Type of Scandal, F(1, 479) = 0.25, p = .619, η_p² = .001.

To test the hypotheses that the more participants report knowing real (H2a) and invented (H2b) scandals the more they adhere to conspiracy theories, we computed Pearson correlation coefficients between Conspiracy Theory Beliefs and the number of Recognized Scandals (the two orders and types of scandals were merged). There was no overall correlation between Conspiracy Theory Beliefs and the number of Recognized Scandals (real and invented), r(481) = −.02, p = .590. This correlation was as expected by H2b, significantly positive for invented scandals, r(253) = .13, p = .033, but contrary to H2a negative (but nonsignificant) for real scandals, r(226) = −.09, p = .184.

To verify these results, we computed a moderation analysis testing the effects of the Type of Scandal, Conspiracy Theory Beliefs, and their interaction with the number of Recognized Scandals. Conspiracy theory beliefs did not significantly predict the overall number of scandals, β = −.08, t = −1.45, p = .149. The Type of Scandal significantly and strongly predicted the number of Recognized Scandals, β = −.56, t = −14.66, p < .001, meaning that overall more than twice the number of Real Scandals (M = 2.23, SE = 0.07) were recognized than Invented Scandals (M = 0.95, SE = 0.06). The significant interaction effect, β = .14, t = 2.39, p = .017, confirmed that, while Conspiracy Theory Beliefs were related positively to the recognition of Invented Scandals, they were related negatively to the recognition of Real Scandals (see Figure 1). Adding the interaction effect to the model explained significantly more variance, ΔR² = .01, ΔF(1, 479) = 5.72, p = .017. The full model explained nearly one-third of the total variance of our sample, R² = .32, F(3, 479) = 74.05, p < .001.

The recognition of real scandals was significantly and negatively correlated with a right-wing political orientation, r(226) = −.17, p = .009, which was not the case for invented scandals, r(253) = −.01, p = .828. We thus followed the same analytical strategy as above and computed a moderation analysis testing the effects of the Type of Scandal, Conspiracy Theory Beliefs, and their interaction on the number of Recognized Scandals. Political orientation significantly and negatively predicted the overall number of scandals, β = −.15, t = −2.82, p = .005. The Type of Scandal significantly and strongly predicted the number of Recognized Scandals, β = −.56, t = −14.76, p < .001, meaning that overall more Real Scandals were recognized than Invented Scandals. Adding the interaction effect to the model explained more variance only near significantly, ΔR² = .01, ΔF(1, 479) = 3.59, p = .059. The full model explained more than one-third of the total variance, R² = .32, F(3, 479) = 75.02, p < .001. The near-significant interaction effect, β = .10, t = 1.89, p = .059, revealed that, while right-wing Political Orientation was not related to the recognition of Invented Scandals, it was related negatively to the recognition of Real Scandals (see Figure 2).

Regarding the moderation analyses, the achieved powers are (1) over 99% for the linear model with the interaction between CT beliefs and type of scandal with numerator and denominator degrees of freedom of 3 and 479, respectively, and an f²’s effect size of .46 – that is, the ratio of _adjR² to (1 − _adjR²); and (2) over 99% for the linear model with the interaction between political orientation and type of scandal with the numerator and denominator degrees of freedom of 3 and 479, respectively, and an f²’s effect size of .47.

Finally, we computed Pearson correlation coefficients (see Table 1). Beliefs in conspiracy theories were related to the political right and the intake of fewer vaccine doses. Women adhered to conspiracy theories more than men. Age did not significantly correlate negatively with Conspiracy Theory Beliefs.

Table 1 Means, standard deviations, and Pearson correlations

Variable	M	SD	1	2	3	4	5
Note. M and SD are used to represent mean and standard deviation, respectively. Gender was coded as 0 = Women and 1 = Men. *p < .05/15 (alpha corrected for multiple comparisons).
1. Conspiracy theory beliefs	3.22	1.16
2. Recognized scandals	1.56	1.15	−.02
3. Age	20.46	4.58	−.03	.07
4. Gender	1.25	0.44	−.25*	.01	.02
5. Political orientation	3.11	1.27	.15*	−.07	−.04	.05
6. Vaccine doses	2.58	0.75	−.21*	.04	−.11	.01	−.12

Table 1 Means, standard deviations, and Pearson correlations

View as image HTML

Regarding the significant Pearson correlations, the achieved powers are (1) over 99% for the correlation of −.25 between CT beliefs and gender with 476 observations; (2) 91% for the correlation of .15 between CT beliefs and political orientation with 483 observations; (3) over 99% for the correlation of −.21 between CT beliefs and vaccine doses with 470 observations.

Discussion

This study did not confirm the hypothesis that exposure to real or invented scandals increases conspiracy theory beliefs. The absence of this effect may result from (1) our population of university students who are less at risk of espousing conspiracy theory beliefs, given their high level of education (the median score of conspiracy theory beliefs was 3 on the 1–7 scale); (2) the fact that conspiracy theory beliefs are not so easily malleable (with only four scandals briefly presented); (3) the fact that exposure to political-scientific-economic scandals does not affect conspiracy theory beliefs in the general audience. Of course, this preliminary study does not allow us to reach a firm conclusion about these possibilities, something future studies may attempt to evaluate.

We confirmed our hypothesis that higher conspiracy theory believers would identify more invented scandals, but not the same hypothesis about real ones. Our findings contradict the epistemological healthy hypothesis about conspiracy theory beliefs (e.g., Basham & Dentith, 2016; Coady, 2006; Harambam, 2020; Pigden, 1995). That hypothesis asserts that some conspiracy theories are helpful and socially necessary because real conspiracies do exist. In contrast, higher conspiracy theory believers were worse than lower believers at identifying real scandals, but slightly “better” at identifying invented scandals. This result aligns with the recent study by Lantian and colleagues (2021), who found that higher conspiracy theory believers underperformed lower believers in critical thinking tasks (argumentation and reasoning). Several additional studies showed that high believers display more intuitive and less analytical reasoning than low believers (for a review, see, e.g., Wagner-Egger, 2022). Thus, contrary to their usual claims, conspiracy theory believers are not more critical and less naive than the general population. Once more, it even seems to be the contrary.

Interestingly, we found a tendency for right-wing people to be less sensitive to – and thus vice versa left-wing people to be more sensitive to – the existence of real scandals. This result echoes findings that right-wing individuals display more system justification (e.g., Jost, 2020) and believe more in a just world (Lambert et al., 1999) than left-wing people do. The ignorance of real scandals fits with the overestimation of the justice of the system and the world.

Finally, correlation analysis mainly supported past literature about conspiracy theory beliefs, which are generally supported more by the political right (Imhoff et al., 2022) and relate to the intake of fewer vaccine doses (e.g., van Mulukom et al., 2020). Women adhere to conspiracy theories more than men, as discovered in some other studies with large samples (Gilles et al., 2022; Imhoff et al., 2022, study 2; Popoli & Longus, 2021; Stanovich et al., 2016; van Mulukom et al., 2020; Wagner-Egger et al., 2022). Note that the reverse result (Freeman & Bentall, 2017; Krouwel et al., 2017; Lobato et al., 2014), and sometimes no differences were found (Darwin et al., 2011; Dyrendal & Hestad, 2021; Goertzel, 1994; Kuhn et al., 2021; Swami et al., 2012). We clearly need a large meta-analysis at this point. Unlike most studies (for a review, see Wagner-Egger, 2022), age did not significantly negatively correlate with conspiracy theory beliefs.

A limitation of our findings is that real and invented scandals were not perfectly similar. Real scandals were split across politics (Sarkozy, Snowden) and health (tobacco, mediator). In contrast, invented scandals regarded health (Novartis, masks), finance (Swiss bank), and political/health (Guatemalan president’s noncompliance during Covid). Hence, further preregistered studies should systematically test more scandals about various issues. A second limitation of this study is that we tested university students, who less likely tend to adhere to conspiracy theories and are younger than the general population. A broader study in the general population may thus find some effects of exposition to real and invented scandals and perhaps increase our correlational findings (as variance will increase).

Conclusion

These results show that reminding people of real and invented scandals did not have a short-term effect on conspiracist beliefs. Moreover, they support the argument that conspiracy theories are useless in fighting real scandals, as higher believers in conspiracy theories were not better than lower believers at recognizing real scandals. Our results suggest that the better way to identify real political, scientific, and economic scandals is not to believe in conspiracy theories and have a left-wing political orientation. These findings align with the negative correlations Oliver and Wood (2014) observed between political knowledge and conspiracy theories beliefs.

More broadly, beliefs in conspiracy theories (i.e., a religion of conspiracy) may all be considered useless – and even counterproductive – tools for social critique (Wagner-Egger, 2022; Wagner-Egger et al., 2019), as no conspiracy theorist has ever discovered a real conspiracy. Professional inquiries in front of courts prove the latter (i.e., “science of conspiracy”) and are based on direct evidence – not errant data. In short, accusing French president Emmanuel Macron of being a satanist pedophile based on insufficient proof is counterproductive (socially, you will probably be wrong and ridiculed). On the contrary, bringing former president Sarkozy in front of a court and condemning him with solid proof is essential in democracies.