The strength of conspiracy beliefs versus scientific information: the case of COVID-19 preventive behaviours

The study investigates whether distinct conspiracy beliefs about the origin and management of COVID-19 more strongly predict individuals’ intentions to adopt preventive behaviours (vaccination, app-based contact tracing, and physical distancing) than objective official information about COVID-19. Grounded in the Implicit Theories Model, the authors posit that people hold organized implicit theories (beliefs) about COVID-19’s origin that shape interpretations, mistrust in institutions, and expectations about the pandemic’s consequences. They test whether these beliefs directly affect preventive intentions and whether mistrust and expectations mediate these relationships. The work addresses the public health importance of understanding drivers of compliance with preventive measures amid widespread misinformation and uncertainty.

The paper reviews evidence that societal crises heighten uncertainty and control needs, fostering conspiracy beliefs (van Prooijen & Douglas, 2017). COVID-19 spawned numerous conspiracies (e.g., hoax/exaggeration, lab creation, bioweapon, 5G causation), which can have harmful consequences: fueling discrimination, reducing trust in institutions and experts, political disengagement, and lower adherence to health guidelines, testing, and vaccination (European Commission, 2020; Freeman et al., 2020; Imhoff & Lamberty, 2020; Karić & Međedović, 2021; Romer & Jamieson, 2020). Different conspiracy themes may imply different behaviors: hoax beliefs relate to reduced containment practices; purposeful creation beliefs to self-centered prepping (Imhoff & Lamberty, 2020). Exposure to conspiracy rhetoric can reduce intentions to follow public health guidelines, even when juxtaposed with scientific information (Bolsen et al., 2020). The review also notes how reliance on institutions typically rises with unfamiliar topics (Shepherd & Kay, 2012), but conspiracy beliefs may disrupt this tendency, creating a false sense of being informed from non-official sources (Stempel et al., 2007; Klein et al., 2019). The authors highlight gaps regarding which specific beliefs are most influential, their link with physical distancing from infected people, and the role of expectations about COVID-19’s future consequences.

Pre-study: Two phases identified and structured implicit theories (ITs) about COVID-19’s origin.

• Phase 1 (Exploratory): The team collected 74 statements (April 2020) from Spanish-language conspiracy-oriented forums, social media, and YouTube. Five expert academics classified items by similarity into five cultural categories; 44 ambiguous/redundant items were removed, yielding 30 items across five categories.
• Phase 2 (ITs as knowledge): Five questionnaires (one per theory) were created using the 30 items. N=110 Spanish-speaking participants (19–65 years) each completed one questionnaire, responding as the story’s characters on 5-point Likert scales. Typicity Index identified representative items (≥4). Ten items were removed, leaving 20 items across four theories. EFA (principal components) supported four factors; KMO=0.85; Bartlett χ²=1377.934, df=190, p<0.001.

Study 1: Tested whether the four mental representations function as independent conspiracy belief factors and their relations with mistrust, expectations, and preventive intentions. Compared two SEM models: M1 (direct effects from beliefs to intentions, mistrust, expectations) vs M2 (partial mediation by mistrust and expectations).

• Sample: N=265 (89 male, 176 female), mean age=45.39 (SD=12.86; range 18–78); 67.2% with university degree.
• Procedure: Online survey via social media.
• Measures:
  • Conspiracy belief theories (COVID-19 ITs questionnaire from Pre-study, items reworded to self-beliefs). EFA (KMO=0.899) yielded four factors explaining 63.61% variance: Alien CT (5 items; eigen=7.71; top loading r=0.90), Economy CT (4 items; eigen=2.06; top r=0.82), Earth CT (6 items; eigen=1.77; top r²=0.78), Freedom restriction T (5 items; eigen=1.19; top r=0.78).
  • Mistrust in institutional information: 4 items (5-point Likert), higher=more mistrust.
  • Expectations of pandemic consequences: 7 items (4 positive, 3 negative reversed), higher=more positive expectations.
  • Intentions: App use (5 items), Vaccination (2 items; one reversed), Physical distancing from infected people (4 adapted items from CAMI-S; higher=greater willingness).
• Analysis: SEM (AMOS 26) comparing M1 vs M2.

Study 2: Replication with added independent variable: official COVID-19 information (M3). Compared M1 vs M3.

• Sample: Initial N=148; final N=142 (37 male, 105 female), mean age=43.50 (SD=11.33; range 18–82); 71.8% with university degree.
• Procedure: Online survey two months after Study 1; distribution via social media.
• Measures: Same as Study 1 plus Official COVID-19 Information (20-item questionnaire based on Spanish Ministry of Health guidance; 9 false items; 5-point agreement scale). Right-answer rates per item: 31–97.2%; mean score 79.4% (SD=14.3; range 10–100%). COVID-19 ITs EFA replicated four factors (KMO=0.822); variance explained 60.23%: Alien CT (eigen=6.93), Economy CT (2.25), Earth CT (1.74), Freedom restriction T (1.30).
• Analysis: SEM (AMOS 26) comparing M1 vs M3.

Study 3: Pooled sample (Studies 1+2) to confirm M1.

• Sample: N=407 (126 male, 281 female), mean age=44.73 (SD=12.37; range 18–82); 68.8% with college degree.
• Procedure and Measures: Same as prior studies. EFA confirmed four factors (KMO=0.912); variance explained ~60.23%; eigenvalues: Alien CT=7.54; Economy CT=2.07; Earth CT=1.71; Freedom restriction T=1.10.
• Analysis: SEM (AMOS 26) testing M1 on combined data.

Pre-study: Participants’ knowledge about COVID-19 origins organized into four prototypical implicit theories (20 items; 4 factors supported by EFA; KMO=0.85; Bartlett p<0.001).

Study 1 (N=265):

• Low average endorsement of conspiracy beliefs but meaningful associations with outcomes (see Table 1 for descriptive stats, e.g., Alien CT M=1.28, SD=0.59; Economy CT M=2.09, SD=1.11; Earth CT M=2.24, SD=0.89; Freedom restriction T M=2.75, SD=1.02).
• SEM results: M1 (direct effects) fit well: χ²=4.475; df=7; p=0.724; RMSEA=0.000; NFI=0.995; CFI=1.000. M2 (partial mediation) also fit (χ²=0.521; df=1; p=0.470; RMSEA=0.000; CFI=1.000) but was inferior by parsimony/fit indices (AIC M1=80.475 vs AIC M2=88.521). Conclusion: Conspiracy beliefs directly related to lower preventive intentions and higher mistrust; expectations not supported as mediators.

Study 2 (N=142):

• Official information mean score: 79.4% correct (SD=14.3).
• SEM: M1 fit: χ²=7.446; df=7; p=0.384; RMSEA=0.021; NFI=0.981; CFI=0.999. M3 (with official information) also fit: χ²=7.110; df=7; p=0.418; RMSEA=0.011; NFI=0.983; CFI=1.000, but official information had no significant paths to mistrust, expectations, or intentions. Parsimony favored M1 (AIC M1=83.446 vs AIC M3=103.110). Conclusion: Objective official information did not explain additional variance nor buffer conspiracy effects.

Study 3 (N=407):

• Pooled analysis confirmed M1: χ²=4.771; df=7; p=0.688; RMSEA=0.000; NFI=0.996; RFI=0.980; CFI=1.000. Conspiracy beliefs had direct effects on lower willingness to engage in preventive behaviors, increased mistrust, and shaped expectations.

Pattern-level insights:

• Beliefs about restrictions of individual freedom most strongly undermined acceptance of app-based tracking and physical distancing; they were associated with greater mistrust. Interestingly, they were positively related to vaccination intention in some analyses (interpreted as a personally controlled, optional protective action).
• Beliefs attributing origins to dark external forces (economic powers or aliens) were linked to reduced proactive behaviors (e.g., lower vaccination intention for alien-origin beliefs) and higher mistrust.
• Earth/Mother Earth beliefs related to more positive expectations and relatively more acceptance of institutional measures like apps, with less negative impact on behaviors than other conspiracies.
• Across studies, expectations about consequences showed weaker or inconsistent mediating roles compared with direct belief effects.

The research question asked whether conspiracy beliefs versus objective official information better predict adherence to COVID-19 preventive behaviors. Across three studies, conspiracy beliefs consistently operated as organized, independent factors directly influencing lower willingness to engage in preventive actions and higher mistrust in institutions. Official information did not significantly affect mistrust, expectations, or intentions and did not weaken the effects of conspiracy beliefs. These findings suggest that conspiracy beliefs provide an interpretive framework that substitutes for institutional guidance, fostering a sense of understanding and control that undermines reliance on official sources. Different conspiracy themes have distinct behavioral implications: freedom-restriction beliefs particularly impair acceptance of measures perceived as intrusive (apps, distancing), while some beliefs may co-occur with positive expectations or selective acceptance (e.g., vaccination as a voluntary measure). The results extend prior work by showing that, even with low overall endorsement levels early in the pandemic, these beliefs exert measurable, direct impacts on public health behaviors, highlighting the challenge of countering misinformation solely via factual information campaigns.

The study demonstrates that organized conspiracy beliefs about COVID-19’s origin exert direct, robust effects on mistrust and on reduced intentions to adopt preventive behaviors, whereas objective official information neither explains additional variance nor buffers these effects. Not all conspiracy beliefs operate identically: restrictions-of-freedom narratives most strongly weaken acceptance of app tracking and distancing, while some beliefs are associated with more positive expectations or selective compliance (e.g., vaccination). Contributions include: (1) empirical derivation and validation of four implicit conspiracy belief factors; (2) evidence favoring direct-effects models over mediation by mistrust/expectations; (3) null influence of official information in explaining behaviors once beliefs are accounted for. Future research should expand the range of conspiracy narratives, use longitudinal designs to track belief-behavior dynamics, and conduct cross-cultural studies to assess cultural moderators and the evolution of beliefs over pandemic phases.

Key limitations include: (1) sampling during the first pandemic wave (2020), when conspiracy narratives were less developed; participants showed low overall endorsement, potentially attenuating effects; (2) non-probabilistic, online samples from Spain limit generalizability; (3) cross-sectional designs preclude causal inference and temporal dynamics; (4) need for larger, culturally diverse samples and balanced representation of believers vs non-believers; (5) measurement scope could be broadened to include additional emerging conspiracy themes and contextual moderators (e.g., political polarization, cultural values).