Legacy media as inhibitors and drivers of public reservations against science: global survey evidence on the link between media use and anti-science attitudes

The paper addresses growing public reservations toward science—such as preference for faith over expertise, moral resentment toward scientific influence, and disinterest in scientific knowledge—that can undermine decision-making, crisis response, and democratic processes. Prior survey research suggests anti-science attitudes cluster among older, lower-educated, more religious, and politically conservative individuals, and may be lower among frequent legacy media users. However, evidence is limited by a focus on single or WEIRD countries, emphasis on pro-science rather than anti-science orientations, lack of media-use measures, and limited attention to national contexts that might moderate media effects. This study uses cross-national data to assess the prevalence of anti-science attitudes worldwide and test whether legacy media use relates to these attitudes across diverse contexts. It poses: RQ1—global prevalence; H1—negative association between legacy media use and anti-science attitudes; RQ2—variation of this association across countries; RQ3—moderation by press freedom, prevalence of populist rhetoric, and uncertainty avoidance.

Prior work has linked anti-science attitudes to sociodemographic and ideological factors and suggested legacy media can foster trust in science via positive portrayals, framing, and cultivation processes. Yet most studies are country-specific, WEIRD-centric, and focus on trust or pro-science attitudes rather than anti-science orientations, which may be distinct constructs. Cross-cultural research on anti-science attitudes has typically omitted media-use variables, and existing multilevel studies have emphasized socioeconomic context over media-system factors. Theory and comparative media research indicate media effects can depend on individual differences and country-level contexts (media freedom, political discourse, cultural traits such as uncertainty avoidance). Populist rhetoric in particular may challenge scientific and academic elites, potentially shaping the media–attitude link.

Data: Secondary analysis of World Values Survey Wave 7 (Jan 2017–Aug 2020), face-to-face interviews in 49 countries (N = 70,867; gender: 47.4% male; age: M = 42.7, SD = 16.3) for descriptive analyses. Inferential analyses use 38 countries with complete measures (N = 54,658; gender: 47.2% male; age: M = 43.4, SD = 16.8). Known WVS cross-national caveats acknowledged (sampling, timing, modes, translation, measurement invariance). Country-level indicators: (1) Freedom of the Press Index (Freedom House; 0–100; higher = more freedom; sample M = 57.6, SD = 27.2). (2) Prevalence of populist rhetoric (Global Party Survey 2019), constructed by averaging expert ratings (0–10) on populist vs. pluralist rhetoric and its importance across parties within each country (higher = more populist rhetoric; M = 6.10, SD = 0.90). (3) Uncertainty Avoidance Index (Hofstede; −150 to 230; higher = more avoidance; M = 69.8, SD = 21.7). Measures: Anti-science attitudes measured by mean of three 10-point items: dependence on faith over science, science undermining morality, and disinterest in science (M = 5.23, SD = 2.12). Reliability: α = 0.54; multilevel ESEM indicated a single factor (χ² = 728.374, df = 96, p < 0.001; CFI = 0.953; TLI = 0.929; RMSEA = 0.079; SRMR = 0.036). Pro-science items were not reverse-coded into the scale due to conceptual and psychometric issues; additional factor analyses showed distinct factors with weak correlation (r = 0.01). Legacy media use measured as mean frequency of daily newspapers, TV news, and radio news (0 = never to 4 = daily; α = 0.51; M = 2.23, SD = 1.08). Controls: gender (binary), age (years), education (9 ISCED categories), income (1–10), town size (1–10), religiosity (1–10 importance of God), political orientation (1 left – 10 right). Analysis: Bayesian modeling via brms with weakly informative priors; results reported as means and 89% HDIs. RQ1: 49 country-specific Bayesian null models (intercepts only) for anti-science prevalence. H1/RQ2: Bayesian multilevel regression (Model 1) with anti-science as outcome, legacy media use as predictor, varying intercepts and slopes by country, controlling covariates. RQ3: Model 2 adds cross-level interactions between legacy media use and each country-level indicator. Frequentist robustness checks yielded similar results.

- Global prevalence (RQ1): Anti-science attitudes are moderately prevalent worldwide (M = 5.23 on 1–10; 89% HDI: 5.22–5.24) with substantial cross-national variation. Higher scores in many Latin American countries (e.g., Nicaragua, Bolivia, Colombia, Mexico); lower scores in several Anglo-American and East Asian countries (e.g., Australia, Japan, US, China). Some non-WEIRD countries (e.g., Indonesia, Ethiopia) exhibit low anti-science attitudes. About 14% of variance is due to country differences. - Association with legacy media use (H1/RQ2): Globally, no evidence that legacy media use reduces anti-science attitudes (b = 0.008, 89% HDI: −0.032–0.048). However, effects vary by country: more frequent use is associated with lower anti-science attitudes in Thailand, Ethiopia, Nigeria, and the US; but associated with higher anti-science attitudes in Turkey, Bangladesh, Cyprus, Bolivia, Serbia, and Peru. - Covariates (Model 1): Male gender (b = −0.109, 89% HDI: −0.130–−0.077) and higher education (b = −0.090, 89% HDI: −0.098–−0.081) relate to lower anti-science attitudes; age (b = 0.004, 89% HDI: 0.003–0.005), religiosity (b = 0.118, 89% HDI: 0.111–0.124), and right-leaning political orientation (b = 0.048, 89% HDI: 0.042–0.054) relate to higher anti-science attitudes. Income shows a small positive association (b = 0.015, 89% HDI: 0.007–0.023). Town size is near zero. - Moderation by country context (RQ3, Model 2): The prevalence of populist rhetoric moderates the media–attitude link: legacy media use increases anti-science attitudes where populist rhetoric is high and decreases them where it is low (interaction b = 0.068, 89% HDI: 0.013–0.119). Interactions with press freedom and uncertainty avoidance are near zero. Model fit statistics (e.g., WAIC) were similar across models.

Findings indicate that anti-science attitudes are unevenly distributed globally, with higher levels in parts of Latin America and lower levels in many Anglo-American and East Asian countries, though notable exceptions exist. Individual factors (education, religiosity, ideology, gender, age) explain much of the variance, while country-level differences account for a smaller share. Contrary to a general expectation that legacy media foster pro-science orientations, their use does not uniformly reduce anti-science attitudes; instead, effects are context-dependent. In countries like the US, Nigeria, and Thailand, greater legacy media use is associated with weaker anti-science attitudes, consistent with cultivation of positive views of science. In others (e.g., Turkey, Bangladesh, Bolivia, Cyprus, Peru), more use aligns with stronger anti-science attitudes, potentially due to critical or populist-inflected media discourse. The moderating role of populist rhetoric suggests that in contexts where political discourse challenges academic elites, legacy media are less able—or may even be more likely—to convey anti-scientific sentiment, either by amplifying populist narratives or losing influence among populist-leaning audiences. These results refine understanding of when and where legacy media may mitigate or exacerbate public reservations toward science.

This study provides cross-national evidence on the prevalence of anti-science attitudes and shows that the relationship between legacy media use and such attitudes is highly context-dependent. It highlights the moderating role of populist rhetoric in shaping whether media use inhibits or drives anti-science sentiment. Contributions include: (1) focusing on anti-science attitudes as distinct from pro-science views; (2) incorporating legacy media use into cross-country analyses; and (3) identifying a key contextual moderator. Future research should: (a) examine digital and social media use and its interactions with populism and press freedom; (b) measure specific legacy media channels and science-specific media exposure; (c) incorporate broader media-system and cultural context indicators beyond press freedom, and refine measurements of uncertainty avoidance; (d) employ causal designs (experiments, panels) to test directionality; and (e) use qualitative approaches to probe how publics interpret “science” and related constructs.

- Data and design: Cross-sectional survey data limit causal inference and cannot disentangle whether media use shapes attitudes or vice versa. The WVS, while extensive, faces cross-national comparability challenges (sampling procedures, fieldwork timing, modes, translations), potential measurement invariance issues, and external validity concerns. - Measurement: Anti-science attitudes are captured with three items (α = 0.54), reflecting moderate internal consistency typical for broad constructs; the aggregate legacy media use measure (newspapers, TV, radio) may mask channel-specific effects. Pro- and anti-science constructs are distinct, complicating comparisons. - Context indicators: The Freedom of the Press Index is broad; uncertainty avoidance relies largely on dated Hofstede indices with debated validity; other relevant media-system factors (ownership, regulation, internet diffusion) were not included. - Coverage: Inferential analyses include 38 countries due to missing items in some nations, potentially affecting generalizability.