People’s responses to the COVID-19 pandemic during its early stages and factors affecting those responses

The study addresses why the rapid spread of COVID-19 could not be prevented in its early stages and what behavioral responses occurred among individuals. Context is provided by global escalation after WHO’s pandemic declaration (March 11, 2020) and comparisons to prior outbreaks (e.g., 1918 influenza, SARS, H1N1, Zika, Ebola). The paper argues that historical lessons and prior-virus evidence may not fully apply to COVID-19 and that there was a lack of rapid, appropriate actions early on. The purpose is to provide policymakers with immediate, science-based evidence on public responses to inform communication and intervention strategies. Focusing on individuals as the core of collective action, the study uses a life-oriented approach and a nationwide retrospective survey (Japan, end of March 2020) to capture changes in daily activities, travel, and related psychological factors during the early outbreak.

The paper references historical and recent public health crises (Spanish Flu per Soper, 1919; Ebola, H1N1, Zika, SARS) and contemporary COVID-19 social/behavioral science (e.g., Bavel et al., 2020; West et al., 2021). It highlights that non-pharmaceutical interventions and behavioral changes can suppress spread (e.g., Cowling et al., 2020’s Hong Kong survey) but notes gaps: many studies rely on repeated cross-sectional averages without modeling inter-factor relationships, leaving side influences and causal structures unclear. The study positions itself to fill this gap by modeling interdependencies among information reliability, risk perceptions, attitudes, social-network triggers, and changes in life behaviors within a life-oriented framework.

Design: Nationwide retrospective online panel survey grounded in a life-oriented approach capturing interdependent changes across multiple life domains. Timing and context: Survey conducted March 23–30, 2020, covering behavior changes between December 2019–February 2020 (pre-change baseline) and March 2020 (post-change). Early-stage Japan context: key governmental announcements (e.g., Feb 28 school closures effective March 2) and evolving global situation. Sample: 1052 residents across Japan with distributions of age, gender, and region controlled to approximate the national population. Education: 45% university degree or above. Occupations varied (private sector 41.2%, public/education 6.2%, housewives/house-husbands 15.9%, part-time 13.8%, students 9.2%, no job including pensioners 10.6%). Marital status: 50.9% married, 42.5% single. Mean household size 2.56; 44.3% owned homes. Income distribution provided in ranges. Sample size constrained by budget. Ethics approval obtained (HUIDC-2020-009); informed consent collected. Measures (see Table 1 overview): - Bookings: cancellations/postponements/no changes and timing for domestic/overseas tourism and business trips, domestic flights, intercity rail/other modes; activities such as concerts, sports, movies, eating out, gatherings. - Daily life changes and timing: work/school schedules (flexible hours, off-peak commuting, telework/study), shopping (online vs in-store), family life (conversation, housework), travel behavior (public transport, car, walking/cycling; domestic/overseas trips), recreation/hobbies/exercise, social networking (SNS, face-to-face), avoidance of crowded places, sleep, expenditures (energy use, household spending, medical expenses), panic buying (masks, antiseptics, emergency food, daily necessities), detailed time-use (in-home/out-of-home) before vs after. - Triggers: self-understanding; requests from government; requests from workplace/school; recommendations from family; recommendations from colleagues/acquaintances/friends. - Information: knowledge items (basic transmission, symptoms, disinfectant efficacy, awareness of school closure announcement); sources and daily time spent (TV, SNS, radio, face-to-face, paper media); perceived reliability by source (central/local governments, experts, domestic/overseas news, medical institutes, workplace/school). - Risk perceptions: perceived increasing risk at national, prefecture, city/town, frequently visited places, community; perceived dangerousness; personal comparative risk; perceptions of others’ risk perceptions across spatial scales. - Attitudes toward policymaking: perceived expertise, fairness, enthusiasm, readiness/management systems, trust, accountability, sufficiency of information disclosure; perceived policy effects; views on intensity/necessity of measures; willingness to provide information and cooperate. Analytical approach: Aggregate descriptive analyses and temporal transitions; Structural Equation Modeling (SEM) with latent variables estimated using bootstrap to quantify direct/indirect effects among latent constructs: changes in daily life (dependent), information reliability, risk perceptions, attitudes, and triggers of behavioral changes. A complementary data mining approach was also applied to identify predictors for specific behavioral changes (e.g., decreased public transport use, daily trips, crowded-place avoidance, domestic travel, daily interaction).

Behavioral changes and bookings: - By late March 2020, cancellation/postponement rates surged after Feb 28 school-closure announcement. Watching sports cancellations/postponements peaked at 94%, concerts 88%, overseas business trips 87%, gatherings 85%, overseas tourism 81%, domestic business trips 79%, movies 76%. Domestic tourism trips and flights increased cancellations more slowly; many booking cancellations plateaued after March 20. - Daily life changes (largest shares): 75% avoided crowded places; 60% reduced daily trips; 53% reduced eating out; 47% reduced face-to-face with friends/acquaintances; 40% reduced shopping at stores. - Travel behavior: 36% decreased public transport use; 29% increased car travel; 27% increased walking/cycling. Only 14% shifted commuting/schooling to off-peak; 18% had more flexible work/school hours. Non-daily travel: 27% reduced domestic travel; 12% reduced overseas travel. - Activity participation/time use: 37% increased family conversations; 30% more time on hobbies; 29% more personal housework; 25% increased SNS at home; 20% increased sleep; 10% increased physical exercise. Telework/study at home increased for 14%. Online shopping/tele-shopping increased (21%); online/telephone meal orders (9%). 26% reduced stop-by behavior on commute; 20% reported decreased amount of work. 23% reduced exercise (time/frequency) despite 10% reporting increases. - Expenditures and purchases: 32% reported increased household energy consumption, while 50% reported a decrease; 35% tried to reduce overall household expenditure; 8% reported increased medical expenses. Panic-buying tendencies: 9% tried to purchase masks; 7% antiseptics; 11% emergency food/daily necessities. - Wellbeing and prosocial behavior: 37% became mentally/physically tired after longer stays at home; 16% tried to encourage others’ behavioral changes. Life values: 18% agreed their view of life had changed; 41% disagreed. Information, risk, attitudes, trust: - Knowledge: 98% knew droplet/contact transmission; 93% knew key symptoms; 80% knew 70% alcohol disinfectant efficacy; 97% knew of national school closure announcement. - Information exposure: up to 64 minutes/day via TV and 22 minutes via social media; up to 15 minutes face-to-face; <10 minutes via radio/paper media. Reliability: health/medical bodies rated reliable by 57.1%; central government information was next; workplace/school information least reliable. - Risk perceptions: 79% perceived increasing national infection risk; lower at finer scales (e.g., prefecture 61%, city/town 43%, community 33%). 58% viewed COVID-19 as dangerous, but only 32% felt more likely than others to be infected. - Trust and governance: Only ~18–20% agreed central/local governments had sufficient knowledge/fair view; 27% agreed health/medical organizations had sufficient expertise. About 25% felt information disclosure was enough; ~20% felt management systems/measures were in place. Trust in central/local governments about preventing spread: ~24%; trust in health/medical bodies: 41%. Many wanted stronger accountability (e.g., 62% said central government should be more accountable) and more thorough measures (56% central, 48% municipalities). 65% disagreed that unaffected localities need no measures. Modeling insights (SEM and data mining): - SEM standardized total effects show the strongest driver of changes in daily life is triggers of behavioral changes (total effect ≈ 0.402), followed by risk perceptions. Direct and total effects of information reliability and attitudes on changes in life were not significant; however, information reliability and risk perceptions significantly influenced triggers. - Within the triggers latent construct, recommendations from family (loading ≈ 0.79) and from colleagues/acquaintances/friends (≈ 0.78) were most salient, underscoring the role of close social networks. - Perceived risks at local scales (prefecture, city/town, neighborhood) were more influential on changes in life than risks at broader scales. - Data mining identified differentiated predictors for specific behaviors: e.g., decreased public transport use associated with increased perceived risk at often-visited places and others’ national risk perceptions; reduced daily trips linked to self-understanding trigger, reliability of medical information, perceived dangerousness, and others’ risk perceptions; avoidance of crowded places linked to willingness for self-restraint, perceived dangerousness, government-request trigger, and reliability of workplace/school information; decreased domestic travel linked to government requests and perceptions of local risks/policy effects; reduced daily interaction linked to others’ local risk perceptions and colleague/friend recommendations. Implications: - Poor/insufficient communication and low mutual trust between government and public likely contributed to spread by limiting voluntary behavioral change. Effective, targeted risk communication leveraging close social networks is crucial. The proposed LAST (Life-oriented Activity-Space-Time) framing emphasizes aligning needs, activities, places (with sufficient space), duration, and timing to mitigate transmission.

The findings directly address why early containment was difficult: despite knowledge of the virus and some measures, public trust in central/local governments was low, perceived information reliability varied, and the share of voluntary self-restraint and coordinated behavioral changes was insufficient. The SEM results clarify pathways: reliable information and risk perceptions shape behavioral triggers, especially via recommendations from close social networks, which then drive life changes. Thus, improving communication quality, transparency, and trust, and engaging social networks can amplify effective behavioral responses. Differentiated communication strategies are necessary because different behaviors (e.g., public transport reduction, crowded-place avoidance, domestic travel reduction) respond to distinct combinations of triggers, perceptions, and information sources. The LAST approach operationalizes how individuals can redesign activities across needs, spaces, and times to practice physical distancing while maintaining essential functions.

This study provides an early-stage, comprehensive picture of Japanese residents’ behavioral and attitudinal responses to COVID-19 (March 2020) using a life-oriented framework and nationwide survey. It quantifies the central role of social-network-based triggers and locally perceived risks in driving changes in daily life, highlights low trust in government and concerns about information disclosure and policy readiness, and documents broad shifts in mobility, activity participation, and consumption. Policy contributions include emphasizing: (1) improved, transparent, and timely risk communication; (2) leveraging close social networks to activate behavioral triggers; (3) differentiated messaging tailored to specific behaviors; and (4) a LAST (Life-oriented Activity-Space-Time) approach to guide safe activity scheduling and location choices. Future research should further develop behavioral components within epidemic models, refine targeted communication strategies across stakeholders, assess compensation/support mechanisms when essential activities are curtailed, and examine how long higher-order social relationships can be adjusted without undue harm.

The survey captures Japan’s early-pandemic period (end of March 2020) when confirmed cases were relatively low, which may limit generalizability to later phases or other countries. The sample size (1052) was determined by budget constraints. Measures rely on retrospective self-reports of behaviors and timings, which may introduce recall bias. While SEM and data-mining approaches were applied, detailed specifications and external validations are limited in the text. Results reflect perceptions and reported behaviors rather than observed compliance or epidemiological outcomes.