Health literacy, religiosity, and political identification as predictors of vaccination conspiracy beliefs: a test of the deficit and contextual models

Conspiracy theories are beliefs that events or actions are orchestrated by covert actors with hidden agendas. In the vaccination context, such beliefs target the pharmaceutical industry, alleged adverse effects, and concealment by institutions. During COVID-19, new narratives (e.g., virus origin, depopulation, disinfectants as cures) proliferated and have been linked to harmful health decisions, reduced trust in healthcare, and lower support for public health measures. The deficit model of public understanding of science posits that higher scientific literacy yields more positive science attitudes, but evidence suggests only small to moderate correlations and domain variability. The contextual approach argues that social variables (e.g., politics, religion) can have direct and moderating effects on science attitudes, potentially shaping or buffering the impact of literacy. This study tests both approaches in Croatia during the COVID-19 pandemic, asking whether higher health literacy relates to lower vaccination conspiracy beliefs and whether religiosity and political identification are associated with, and moderate, this relationship. Hypotheses: H1 health literacy negatively correlates with vaccination conspiracy beliefs; H2 religiosity positively correlates; H2a religiosity moderates the literacy–beliefs link; H3 right-wing political identification positively correlates; H3a political identification moderates the literacy–beliefs link.

Scientific literacy encompasses knowledge and ways of thinking needed to navigate modern scientific issues; attitudes vary by domain and ethical content. Health literacy overlaps with scientific literacy and includes accessing, understanding, appraising, and applying health information. Prior work links health literacy to vaccine hesitancy and lower acceptance of COVID-19 misinformation; higher digital health literacy relates to fewer misperceptions. Evidence on direct links between scientific/health literacy and COVID-19 conspiracy beliefs is mixed, with some domain-specific effects. Contextual factors are crucial: political orientation has related to vaccine conspiracism and hesitancy in some countries (e.g., U.S. conservatives show more hesitancy; partisan patterns vary by context), but findings are inconsistent elsewhere (e.g., France shows divisions by party closeness rather than left–right). Trust in public health institutions can overshadow partisanship. Religiosity and spirituality have been associated with conspiracy beliefs and hesitancy in several European contexts, though not uniformly. Moderation by political identification and religiosity has been found in other science domains (e.g., stem cell research, climate change, nanotechnology), with knowledge effects sometimes stronger among conservatives and religious individuals. The present study extends this literature to vaccination conspiracy beliefs, testing both main and moderating effects of religiosity and political identification in Croatia.

Design and measures: Cross-sectional survey of university students. Health literacy was measured via the HLS-EU-Q6 (six items; examples include judging reliability of health risk information in media and using doctor-provided information for decisions). The HLS-EU-Q6 correlates strongly with the longer HLS-EU-Q, has been validated internationally, and showed Cronbach’s alpha = 0.76 in this study. Vaccination conspiracy beliefs were measured with the 7-item Vaccine Conspiracy Beliefs Scale (Shapiro et al., 2016), scored on a 7-point Likert scale (e.g., beliefs about cover-ups linking vaccines to autism, fabrication of safety data, and pharmaceutical companies hiding dangers). The Croatian version was previously validated; Cronbach’s alpha = 0.95. Religiosity and political identification were each measured on 1–10 scales (lower values denote lower religiosity and left-wing identification). Controls: parental educational index (1–5), place of residence (urban/rural), gender (female/male), and field of study (STEM vs social sciences, arts and humanities). Sample and procedure: N = 729 university students from eight faculties/departments at Josip Juraj Strossmayer University of Osijek, Croatia. Study programs were randomly selected proportionate to fields; the sample closely matched the university distribution by field. Females were slightly overrepresented. Surveys were administered during class (∼10 minutes). Data collection occurred September–October 2021. Ethical approval was granted; informed consent was obtained; anonymity and voluntary participation were ensured. Missing data were minimal (0–1.23%) and imputed via regression multiple imputation. Analysis: Structural equation modeling (SEM) with maximum likelihood estimation was used to assess direct and interaction effects. Model fit benchmarks: CMIN/df < 3, RMSEA < 0.05, CFI/TLI > 0.90. Two models were tested: Model 1 included main effects of health literacy, religiosity, and political identification on vaccination conspiracy beliefs with controls; Model 2 added interaction terms (health literacy × religiosity; health literacy × political identification). Fit indices: Model 1 CMIN/df = 2.28, CFI = 0.975, TLI = 0.967, RMSEA = 0.041 (90% CI 0.034–0.047); Model 2 CMIN/df = 2.40, CFI = 0.943, TLI = 0.930, RMSEA = 0.044 (90% CI 0.041–0.047). Descriptives: vaccination conspiracy beliefs mean 22.98 (SD 11.46; range 7–49); health literacy mean 15.94 (SD 3.37; range 6–24); political identification mean 5.81 (SD 2.98); religiosity mean 5.36 (SD 2.33).

Bivariate correlations: Vaccination conspiracy beliefs correlated negatively with health literacy (r = -0.25, p < 0.01) and parental education (r = -0.19), and positively with religiosity (r = 0.27, p < 0.01) and right-wing political identification (r = 0.15, p < 0.01). Place of residence showed a small negative correlation (r = -0.13, p < 0.01). Religiosity and political identification were strongly correlated (r = 0.43, p < 0.01). Model 1 (main effects): Health literacy and religiosity were significant predictors of vaccination conspiracy beliefs. Lower health literacy predicted higher conspiracy beliefs (standardized β ≈ -0.21, p < 0.01), and higher religiosity predicted higher conspiracy beliefs (β ≈ 0.24, p < 0.01). Political identification was not a significant predictor when other variables were controlled. Among controls, higher parental education and male gender predicted lower conspiracy beliefs; place of residence also showed a small effect. Model 2 (interactions): The interaction health literacy × religiosity was significant and negative (standardized β ≈ -0.10, p < 0.05), indicating that the protective effect of health literacy against conspiracy beliefs was stronger at higher levels of religiosity. Probing the interaction showed that at one SD below mean religiosity, the unstandardized literacy effect was -0.30 (ns), while at one SD above mean religiosity it was -0.75 (p < 0.05). The interaction health literacy × political identification was not significant. Overall, hypotheses supported: H1 supported; H2 supported; H2a supported; H3 not supported in multivariate models; H3a not supported. Both SEM models demonstrated acceptable fit to the data.

Findings support both the deficit model and aspects of the contextual model. Higher health literacy relates to lower endorsement of vaccination conspiracy beliefs, indicating that knowledge remains relevant. Religiosity exerts a direct positive association with conspiracy beliefs and moderates the literacy effect, such that literacy matters more among the more religious—opposite to some prior domains where knowledge effects were stronger among the less religious—potentially due to ceiling effects among already pro-science groups or because literacy may operate partly by increasing trust in science. Political identification showed only a small bivariate correlation and no unique or moderating effect in multivariate models, which the authors attribute to Croatia’s specific socio-political context: science topics (including vaccination) were not deeply aligned with left–right ideological divisions, and elite partisan cues were mixed, especially during COVID-19. The Catholic Church’s messaging was supportive of public health measures but equivocal on vaccination mandates, potentially shaping the religiosity–beliefs link via institutional rather than doctrinal mechanisms. The results suggest contextual, institutional, and ad hoc factors can shape vaccine-related conspiracism beyond ideological content alone. Future work should examine trust in science as a mediator and disentangle ideological versus institutional pathways.

Health literacy is inversely associated with vaccination conspiracy beliefs, but knowledge alone is insufficient to explain such beliefs. Religiosity has both direct and moderating effects, indicating that values and institutional contexts shape how literacy translates into attitudes. Political identification did not uniquely predict or moderate conspiracy beliefs in this Croatian student sample, likely reflecting context-specific institutional dynamics rather than ideological polarization. Overall, the study provides partial support for a stronger contextual model (moderation by religiosity) alongside support for the deficit model (main effect of literacy). Practically, vaccination campaigns should not rely solely on improving literacy; they should also address value-laden concerns, trust, and institutional influences. Future research should test mediation by trust in science, assess specific COVID-19 vaccination conspiracy beliefs, and compare across populations and countries to identify context-dependent patterns.

- Student sample from a single Croatian university limits internal and external validity; students were somewhat less religious and less right-wing than the general population. - Health literacy was self-perceived (HLS-EU-Q6), which may be confounded with constructs like trust in the healthcare system. - The vaccine conspiracy scale measured general vaccination conspiracies; the assumption that responses reflected COVID-19-related conspiracies was not directly tested. - The study did not test mediation models (e.g., trust in science) and did not measure deeper political or socio-psychological variables that might clarify mechanisms. - Findings are context-specific; replication in other populations and countries is needed for generalizability.