Contributions
- Demonstrates negative association between conspiracy beliefs and health-protective behaviors (wearing masks, hand-washing, social distancing) across three independent samples
- Establishes positive association between conspiracy beliefs and checking social media for COVID-19 information
- Shows social media use predicts reduced health-protective behaviors independent of conspiracy belief
- Documents platform-specific effects, with YouTube showing the strongest negative association with protective behaviors
- Provides evidence that broadcast media (TV/radio) show opposite effect—positive association with protective behaviors
Method
Three online questionnaire surveys of UK adults:
Study 1: Self-selecting panel (N=949); measured conspiracy beliefs, general media preference (legacy vs. social), health-protective behaviors Study 2: Stratified random sample (N=2250); focused on frequency of checking social media for COVID-19 news and specific protective behaviors Study 3: Stratified random sample (N=2254); detailed measurement of knowledge sources including four named platforms (YouTube, Facebook, WhatsApp, Twitter) vs. broadcast/print media
Statistical tests used Mann–Whitney–Wilcoxon U tests for media usage relationships, Fisher's exact test for conspiracy belief associations. Study 3 included linear probability models controlling for demographics, conspiracy belief, and media usage simultaneously.
Results
All three studies supported H.3 (negative relationship between conspiracy beliefs and health-protective behaviors), with strongest effects for beliefs about non-existence of coronavirus (CB.3.3) and exaggeration of deaths (CB.3.4).
Studies 2 and 3 supported H.2.1/H.3.2: Conspiracy belief positively associated with checking social media for COVID-19 information (Study 2: U-test p<0.001; Study 3: 0.62–0.67 effect size).
Study 3 media effects (H.3.1–H.3.4):
- Broadcast media: small but significant negative correlation with conspiracy beliefs (p<0.001) and positive correlation with protective behaviors (p=0.046)
- Social media platforms: strong positive correlation with conspiracy beliefs (U=424,640, p<0.001, CI: 0.62–0.67) and strong negative correlation with protective behaviors (U=342,191, p<0.001, CI: 0.35–0.41)
- YouTube strongest platform effect, followed by Facebook
- Friends/family as source: weaker effects in same direction as social media
Linear probability model (Table 5): When controlling for demographics, conspiracy belief emerges as stronger predictor of non-engagement than age/gender. Media usage—especially social media—emerges as the most powerful predictor overall (β = -0.39, p<0.001).
Notes
Strengths: Rigorous methodology combining self-selecting and representative samples, multiple independent surveys reduce generalization risk. Detailed media measurement in Study 3 (platform-specific rather than aggregate) reveals YouTube as primary problematic vector. Regression analysis controls for age/gender confounds; shows media effects persist and are substantive. Public health relevance high.
Limitations: All studies rely on self-reported media use and behavior compliance, introducing social-desirability and recall bias. Study 1 used self-selecting sample (younger, more digitally engaged). Correlational design limits causal inference.
Research gaps: Limited investigation of why social media use and conspiracy beliefs co-occur (homophily, algorithmic amplification, lower baseline skepticism?). No mechanism testing of what specific features of social media (algorithmic curation, peer networks, affordances for unverified claims) drive the effects.