Effects of media on preventive behaviour during the COVID-19 pandemic

The study addresses how media use influences voluntary nonpharmaceutical preventive behaviours during the COVID-19 pandemic, particularly attitudes toward refraining from going out. Given that pharmaceutical measures emerged relatively late and voluntary behaviour is critical for long-term disease control, understanding media effects is important. Prior work suggested SNS browsing promotes preventive behaviour, but most evidence came from single time points early in the pandemic. Because the pandemic was prolonged and conditions changed over time, the authors aimed to investigate how media effects on preventive behaviour evolved, using a time-series (panel) perspective in Japan across 2020 and 2021.

Media effects theory distinguishes traditional mass media (e.g., TV, newspapers) with generally higher reliability and uniform messaging from social media (e.g., SNS) characterized by user-generated content, algorithmic curation, selective exposure, and potentially lower reliability. Social media can foster echo chambers, confirmation bias, and polarization, which may influence attitudes toward preventive behaviour, especially amid uncertainty. During COVID-19, diverse information—including misinformation and personal opinions about policies—circulated widely on social platforms. While fake news propagation may be less prevalent in Japan than in some other countries due to higher trust in mass media, social media could still affect preventive behaviour. Prior studies have linked social media exposure to heightened perceptions of severity and negative emotions, potentially increasing fear of going out and promoting self-restraint. However, much of this evidence comes from early pandemic snapshots, leaving the persistence and evolution of these effects over time unclear.

Design: Two-wave panel survey conducted in Japan in 2020 (first wave) and 2021 (second wave) to explore temporal changes in media effects on preventive behaviour. Measures: Multiple constructs measured identically across waves (with knowledge items updated to current conditions), including: - Stay-home intentions and preventive norms; frequency of going out for various purposes; personal and perceived public opinions on refraining from going out; blame attribution for unnecessary outings; predictions of others’ stay-home behaviour and risk perception; perceived influence of mass media, social media, and friends/family on one’s own and general citizens’ opinions about stay-home orders (Table 1). Independent variables (Table 2): - Frequency of mass media use (newspapers, TV, news sites of newspapers/TV stations, Yahoo! News). - Frequency of social media use (Facebook, Twitter, LINE, YouTube, other news sites, summary sites/blogs). - Media literacy (subjective ability to judge accuracy of mass media information). - Media suspicion (skepticism toward mass media information). - Political interest; selective exposure; Twitter browsing; Twitter posting. Analytic approach: - Cluster analysis: Hierarchical clustering (Ward’s method) on attitudes/behaviours related to refraining from going out, yielding three clusters each year: self-restraint, moderate, and going out. - Model selection: Stepwise logistic regression (bidirectional, AIC-based) predicting cluster membership at each time point from media-related variables. For 2021 cluster outcomes, both 2020 and 2021 independent variables were candidates to account for changes in media use over time. Additional models examined predictors of cluster shifts between 2020 and 2021. Sample: Panel totaling 987 observations across waves (Table 3 cross-tab of 2020 vs 2021 cluster memberships).

- Three clusters emerged in both years: self-restraint, moderate, and going out. Refraining from going out was lower overall in 2021 than in 2020. - Distribution changes: In 2020, over half of respondents were in the self-restraint group and 19% in the going out group. By 2021, the self-restraint group had dropped to about 29%, indicating a relative increase in going out. - SNS browsing and self-restraint: SNS browsing consistently predicted higher likelihood of belonging to the self-restraint group in both 2020 and 2021. - SNS and going out: SNS browsing was associated with reduced going out during the early pandemic (2020) but this effect disappeared by 2021. - Twitter use modes: Tweet browsing positively predicted self-restraint; tweet posting negatively predicted self-restraint (robust across time). - Demographics: The going out group was consistently younger and more likely male. - Political interest: In 2020, higher political interest aligned with self-restraint (and low interest with going out), but this pattern was not observed in 2021, possibly due to shifts from hard to soft news coverage over time. - 2021 cognitive/media factors: Higher media suspicion and greater selective exposure were associated with higher likelihood of being in the going out group in 2021. - Transitions: Individuals shifting from self-restraint to going out over one year were less influenced by media type and more by cognitive factors (e.g., selective exposure, media suspicion).

The study demonstrates that media effects on preventive behaviour are time-sensitive. While SNS browsing reliably aligns with self-restraint attitudes, its early-pandemic association with reduced going out did not persist a year later, suggesting that as information environments and public familiarity with COVID-19 evolved, media influences changed. Demographic factors (age, sex) and cognitive orientations (selective exposure, media suspicion) also play distinct roles, particularly in later stages. These findings address the research question by showing that the relationship between media use and preventive behaviour is not static and must be understood in a longitudinal, context-dependent framework. The results are relevant for public communication strategies, indicating that reliance on SNS to promote sustained self-restraint may be effective for some aspects (e.g., reinforcing self-restraint identity) but insufficient to curb going out behaviours as contexts change; cognitive and demographic heterogeneity must be considered.

This study provides longitudinal evidence that the effects of media use—especially social media—on preventive behaviours during COVID-19 vary over time. SNS browsing consistently aligns with self-restraint, but its early negative association with going out dissipated after one year. Individuals transitioning from self-restraint to going out appear more influenced by cognitive factors than by media type. The findings underscore the need for time-series analyses when designing and evaluating media-based public health communications and for differentiating effects on self-restraint versus going out behaviours. Future work should further examine evolving content and user selective exposure over time and how cognitive orientations interact with media use to shape sustained preventive behaviours.