Investigators recruited 4,000 participants in total using nonprobability quota sampling—2,000 residents from Tokyo and 2,000 from Osaka. The survey was retrospective and administered online in April 2024. Respondents answered questions about two distinct time windows: one period when an emergency declaration was in effect and one period when no emergency declaration applied. Multiple regression analyses were performed and standardized partial regression coefficients were calculated by the study team; the abstract in the source did not report the numeric coefficient values or p-values. The authors framed their analysis around behaviors that increase transmission risk—commonly referred to as 3Cs (closed spaces, crowded places, and close-contact settings)—and anxiety about those settings.
The cohort responses during the non-emergency period showed several specific relationships between information trust and behavior. The study team observed a negative association between frequenting 3Cs and trust in family or relatives—participants who reported greater trust in their family or relatives tended to report less frequenting of 3Cs. At the same time, frequenting 3Cs was positively associated with daily conversation (for example, routine person-to-person chat), according to the authors. In parallel, anxiety about visiting 3Cs was positively associated with trust in three information sources: medical professionals, family/relatives, and television. In other words, trust in those sources correlated with greater self-reported anxiety about environments that increase infection risk, and that anxiety tracked with reported avoidance behaviors in the non-emergency context.
When the same respondents considered the period under an emergency declaration, the pattern of associations shifted. The investigators reported a negative association between frequenting 3Cs and trust (the abstract describes this as a general negative association between frequenting 3Cs and trust in unspecified sources). Meanwhile, 3Cs anxiety remained positively associated with trust in medical professionals, family/relatives, and television—consistent with the non-emergency period findings for anxiety. A divergent signal emerged: the study team observed a positive association between frequenting 3Cs and trust in celebrities, weekly magazines, and social networking services during the emergency period. That implies participants who relied more on those popular media or personalities reported higher frequency of exposure to closed, crowded, or close-contact settings while an emergency declaration was active.
Authors performed a mediation analysis to probe causal pathways implied by the regression models. The mediation analysis indicated that trust in medical professionals, family/relatives, or television increased 3Cs anxiety, and that this anxiety was associated with decreased frequenting of 3Cs during states of emergency. The study team therefore reported a pathway in which trust in certain authoritative or traditional information sources elevated anxiety about risky settings, and that elevated anxiety was associated with greater avoidance of 3Cs when an emergency was declared. The source abstract did not include detailed mediation statistics, nor did it provide the numerical mediation effect sizes in the text available here.
The authors used a retrospective, online survey and nonprobability quota sampling. These design elements were explicit in the abstract and limit representativeness and causal inference. The abstract did not provide the regression coefficients, confidence intervals, p-values, or detailed demographic breakdowns in the version of the source text provided here; those specifics may be available in the full text but were not included in the abstract. The study did not report which exact trust items comprised the composite labeled broadly as ‘trust’ in some emergency-period associations, and the abstract did not disclose how ‘daily conversation’ was operationalized. These omissions leave open clinical and public-health questions: which precise messages from medical professionals or television drive anxiety; how differences in demographic subgroups modify these associations; and whether interventions that change trusted information sources would alter behavior in prospective designs. The study team reported associations and mediation signals, but randomized or longitudinal designs would be needed to establish causality.
The article ends on empirical limitations and open questions rather than policy recommendations. The full text (linked from the PubMed entry) would need to be consulted for detailed statistics, measurement instruments, and subgroup analyses that could clarify the signals the investigators reported in the abstract.