Convergence and diversity: how collective risk perception shapes public compliance behaviour – a case study of China’s Covid-19 response

Abstract

This study investigates the influence of collective risk perception on public compliance behaviour, aiming to elucidate why cities exhibit significant disparities in pandemic response effectiveness and compliance levels despite implementing similar containment policies. Drawing on the perspectives of power and knowledge, we first construct an explanatory framework for collective risk perception, and employ the most similar systems design (MSSD) to examine the contrasting Covid-19 responses in Wuhan and Shanghai, China. The findings indicate that during the early stages of the Wuhan outbreak, collective risk perception exhibited a high degree of convergence, shaped by unknow knowledge, stringent government interventions, pessimistic expert evaluations, negative societal opinions and the alarming global pandemic trajectory, thereby fostering strong public compliance. Conversely, during the Shanghai outbreak, collective risk perception became highly diversified owing to the proliferation of scientific knowledge, divergence in expert opinions, dramatic shifts in government containment strategies, polarization of social opinion and the availability of alternative international response models, collectively undermining public consensus on governmental measures and attenuating compliance. On the basis of these insights, this study develops a theoretical framework delineating the interaction between collective risk perception and public compliance behaviour, positing that the degree of convergence or diversity in collective risk perception serves as a critical determinant of compliance levels. A convergent risk perception fosters high compliance, whereas a diversified perception engenders behavioural inconsistencies and lower adherence. By advancing the understanding of risk perception dynamics in public crisis governance, this study offers policy implications for future public health emergencies, emphasizing the necessity for governments to strategically construct collective risk perception through continuous and transparent crisis communication, coordinating expert consensus and guiding social discourse to enhance public adherence and optimize crisis governance outcomes.

Introduction

Since late 2019, the unprecedented coronavirus disease 2019 (Covid-19) pandemic has made the response to the public health crisis a national and local priority. As a novel crisis with “very little experience to draw on”, the Covid-19 pandemic is evidently transboundary in nature, spreading on a large scale, with social, political and economic consequences, and has immense geographical dissemination, causing enormous loss of human life and economic and social disruptions [47]. Consequently, the United Nations [66] highlights that the Covid-19 pandemic has imperilled global sustainable development and will rapidly turn from a health crisis into a human and socio-economic turmoil in the next decade. This threat has challenged global governments’ practices of responding to public health risks and called for transformative action to achieve recovery better and prepare all aspects of society for stable coexistence with Covid-19.

In the global response to the Covid-19 pandemic, a key lesson learned underscores the importance of public compliance with containment measures in managing Covid-19 [3]. Citizens’ predisposition to support and comply with government policies plays a crucial role in shaping the country’s pandemic performance, determining whether authorities can effectively control such events [25, 26]. Identifying the factors determining public compliance with containment measures is therefore vital. It influences how governmental authorities facilitate the implementation of these measures to shape public compliance and ensure the effectiveness of crisis response.

Emerging literature specifically addresses issues related to compliance and has identified psychological factors as the most prominent determinants of public compliance behaviour [17, 23]. Among these factors, one consistently identified crucial consideration is the perception of risk. In the Covid-19 practice, it has been demonstrated to play a vital role in fostering public compliance awareness and yield effective outcomes in a range of compliance behaviours, including moral condemnation of physical distancing violations [12], comprehensive adherence to Covid-19 public health guidelines [84], maintaining physical distance [42] and practising hygiene behaviours [48], compliance with self-isolation and lockdown measures [55] and the voluntary uptake of Covid-19 vaccination [7].

However, the mechanism through which public risk perception influences compliance behaviour has remained in a relatively ambiguous domain. Most studies merely report a correlation between risk perception and compliance behaviour and often focus solely on compliance with limited preventive measures (such as social distancing and hand washing) [1, 18]. Additionally, situations may arise where citizens, despite the willingness to comply with measures, face economic constraints that hinder their ability to adhere [2]. Consequently, an individual-level examination of risk perception is insufficient to explain the emergence and variation of collective compliance behaviour. To address this gap, this study adopts a group-level analytical perspective to investigate the interplay between risk perception and compliance behaviour. Such an approach provides valuable insights for policy-makers seeking to enhance risk mitigation and crisis response effectiveness, free from the constraints imposed by individual-level heterogeneity.

The Covid-19 practices in China have opened up a space for our exploration. Despite operating under the same pandemic crisis and implementing similar containment measures, cities in China exhibited significant variation in public compliance behaviours, with increasing compliance fatigue becoming evident. This discrepancy raises a fundamental research question: what underlying conditions account for the observed differences in public compliance during crisis management? We posit that, when controlling for objective contextual factors (i.e., an identical crisis and uniform response strategies), collective risk perception plays a pivotal role in shaping public adherence to crisis response measures. A highly convergent collective risk perception fosters greater compliance, whereas a more diverse and heterogeneous risk perception leads to diminished adherence.

To substantiate our argument, we employ the most similar systems design (MSSD) to empirically test our hypothesis. We selected the core cities of the two waves of the pandemic in China, Wuhan and Shanghai, as comparative cases. We observed differences in the collective risk perception of the public in these two cities and explored how collective risk perception leads to different compliance behaviours, thereby influencing variations in crisis response outcomes. We hope these findings can provide policy-makers with compelling evidence to emphasize the crucial role of psychological determinants, particularly collective risk perception, in crisis response and incorporate it into future policy agendas.

Theoretical basis

Risk perception and public compliance behaviours

The public’s risk perception towards the current health threat has been identified as a critical factor in the public’s voluntary compliance behaviours [62]. The protective motivation theory holds that the public’s compliance with preventive measures depends on the level of their risk perception [26, 46]. Many studies have further demonstrated that the public with higher levels of perceived threat and perceived effectiveness of measures had stronger compliance with containment measures [27, 60, 61]. Moreover, if people perceive that containment is effective, they will be willing to tolerate a certain amount of discomfort caused by strict restrictions [18]. International studies on the Covid-19 pandemic have also underscored the necessity for governments and experts to acknowledge the multidimensional nature of risk perception which is essential for fostering community engagement in recommended behaviours and ensuring adherence to non-pharmaceutical interventions (NPIs) – a cornerstone strategic approach for effective infection control [23]. The importance of public risk perception during a crisis is more important than we realize, as it can improve public compliance and promote the success of government strategies [20].

Measuring risk perception from the power-knowledge perspective

Numerous variables affecting risk perception have been widely discussed and verified. Among these, the individual characteristics (such as demographic traits, psychological attributes, value orientations and domain-specific knowledge and understanding) of the “perceiver” have received the most scholarly attention [50], but they exhibit significant variation and instability across individuals. To mitigate research errors stemming from individual differences, this study examines collective risk perception at the group level, which possesses both temporal and structural stability.

The shaping of collective risk perception is closely linked to the extent to which groups engage with external environmental information [41]. Groups rely on experts, government sources, media and community members as potential credible information sources to receive, interpret and disseminate information, collectively forming a shared understanding of risk [19]. In crisis situations, governments, as entities wielding national power, serve as the primary response actors. They utilize well-established power structures and resource management systems to shape, guide and influence public crisis responses. However, the dynamic, sudden and complex nature of modern crises poses new challenges to rigid and inflexible bureaucratic systems. Consequently, professionals with expertize – who possess the attribute of “knowledge” – have increasingly become recognized as crucial actors in crisis management. Particularly in the context of emerging infectious pandemics, experts with “knowledge” attribute can employ “rapid” heuristic matching to respond to limited information, identifying similar emergencies on the basis of their specialized knowledge and providing critical support for decision-making [35]. As policy issues increasingly encompass emerging phenomena and unknown risks, integrating knowledge management can assist decision-makers in identifying potential hazards, enhancing predictive capabilities, strengthening the robustness of problem recognition and improving the reliability of decision-making [9, 75].

Against this backdrop, this study examines the factors influencing collective risk perception from both power and knowledge perspectives. From the power perspective, government measures and public opinion are identified as the primary factors, while from the knowledge perspective, scientific knowledge and expert assessments serve as key determinants. Notably, on the basis of empirical observations, we introduce international experience as a moderating variable that influences collective risk perception (Fig. 1).

Fig. 1

The influence logic of collective risk perception from the power-knowledge perspective

Government measures and public opinion from the perspective of power

As a central institution of state power, the government exerts a significant influence on public risk perception across diverse crisis contexts, with its impact closely mediated by the level of public trust in governmental authority. Multiple studies of various risks in different cultural contexts have consistently a strong demonstrated correlation between trust in government and the perception of risks [11, 45]. The public with higher trust in the government will perceive a strong alignment between their own interests and authorities,they are thus intended to support government intervention and have a higher risk perception influenced by strict mitigation measures [22, 25], whereas lower trust in the government leads to the opposite effect.

China exercises stricter control over public opinion compared with Western countries, with the government implementing extensive and diverse regulatory measures across social media platforms to reinforce legitimacy and governance capacity [29]. This regulatory environment necessitates the inclusion of public opinion as a dimension within the power perspective of risk perception analysis. Previous research has found that media and the social opinion formed in them are the primary sources of shaping the public’s risk perception in an environment where risk issues are complex and highly uncertain, such as avian influenza, H1N1 influenza, and Covid-19 [54, 64]. In the digital era, social media has emerged as a central real-time information conduit during crises [77], facilitating the exchange of opinions, beliefs and information while influencing collective interpretations of risk events [64]. However, online information is often more fragmented than traditional news, leading to the accumulation of negative perceptions through a snowball effect [33, 38]. And social media can become a conduit for misinformation and rumours, undermining institutional and expert credibility [68]. Consequently, the media’s choice of risk issues and sentiment of policy interpretation plays a crucial role in crisis management, as it will differentiate the social opinion and strongly affect the public’s risk perception of the event [30].

Scientific knowledge and expert assessment from the perspective of knowledge

Knowledge has proven to correlate strongly with perceived risk across risk perception research. It is currently in vogue among the scientific community to claim that there is a clear positive correlation between knowledge and risk perception, because understanding related knowledge reduces the uncertainty caused by viruses and diseases [28, 81, 83]. An “unknown” risk is more likely to be observed to develop feelings of “fear” because it is a consequence of involuntary exposure, can lead to significant death and is, therefore, psychologically alienating [15, 50]. On the contrary, some scholars have observed a negative correlation between knowledge and risk perception [6, 81]. One possible explanation for the discrepancy could be the “sense of threat’ that knowledge brings. Owing to differences in virus nature, severe acute respiratory syndrome (SARS) and Covid-19 have relatively mild symptoms of infection compared with Middle East Respiratory syndrome (MERS), so people who know more about them may have a stronger sense of control over the outbreak and thus feel less threatened and have weaker risk perception. [81]

The difference in knowledge highlights a relatively sharp disconnect between laymen and experts in risk perception [37, 52], meaning that experts and their opinions on risk can affect public risk perception [69]. In the context of risks, if knowledge is missing, then people need to rely on experts who assess the risk to interpret the information [53]. Moreover, the public will assess risks and adopt adaptive behaviour on the basis of advice from trustworthy experts because of society’s epistemic trust and universal belief in scientists and the scientific community [51]. As the Covid-19 pandemic has shown, experts can not only provide information and data to the public, but also play an essential role in the development of science-led policies, thus influencing the public’s perception of risk [5].

Selection of the moderating variable: international experience

International experience serves as an essential reference for authorities and the public in the face of crisis. Extensive literature has focused on examining and discussing learning lessons from national creeping mega-crisis responses to date [63, 71, 78], such as following China’s example of imposing some type of lockdown, as well as measures agreed upon by countries, such as social distancing, using face masks and closing social gathering points. However, few studies have looked at the impact that international experience has on public risk perception. The experience of other countries in dealing with crises can provide the public with alternative practices. When the public finds the impact and consequences of the Covid-19 epidemic in other countries acceptable, the perceived risk to the public reduces, and vice versa.

Methodology

Case selection: why compare Wuhan and Shanghai?

Wuhan and Shanghai were the central cities in China’s two major waves of the Covid-19 pandemic, both experiencing severe outbreaks, the highest death tolls and profound impacts on national pandemic control policies. Under the unified directives of the Chinese central government, both cities implemented similar containment measures, including strict lockdowns, mass testing and centralized treatment, ensuring a high degree of policy consistency at the institutional level. However, despite these similarities in pandemic response strategies, the two cities exhibited significant differences in the final reported cases and the level of policy compliance during implementation.

Research method: most similar systems design (MSSD)

This study employs the most similar systems design (MSSD), using Wuhan and Shanghai as case studies to investigate why significant differences in pandemic control outcomes emerged despite similar policy contexts and containment frameworks. The core logic of MSSD is that when systems share a high degree of similarity, the key variables influencing the dependent variable – pandemic control effectiveness and public compliance – are likely to stem from a limited set of differentiating factors.

To ensure comparability, we confirm the similarities in case selection. Wuhan and Shanghai exhibit a high degree of similarity in the following aspects:

• a. Policy strategies: both cities implemented strict lockdowns, extensive nucleic acid testing, and centralized treatment measures, all under the central government’s directive. They also enforced similar health code management and quarantine policies.

• b. Epidemic context: both cities served as critical nodes in the nationwide transmission of Covid-19 and experienced peak viral spread during their respective lockdown periods.

• c. Socioeconomic characteristics: as China’s most developed metropolitan areas, Wuhan and Shanghai share large population sizes, advanced healthcare systems and robust governmental administrative capacities.

• d. Policy implementation actors: pandemic response in both cities was executed through municipal governments, grassroots community organizations and public security systems, with professional guidance provided by the national Centers for Disease Control and Prevention (CDC).

Dependent variable: differences in pandemic control effectiveness and public compliance

Wuhan and Shanghai exhibited significant differences in pandemic control effectiveness. Wuhan’s rigorous enforcement was remarkable, with its confirmed cases significantly reduced and the social order and economic development gradually restored (Fig. 2). In contrast, Shanghai’s confirmed cases fluctuated wildly after the peak and even reached another peak. Until the lockdown was eased, over 600,000 people were diagnosed, ten times that of Wuhan (Fig. 3).

Fig. 2

Cumulative confirmed cases and new daily cases in Wuhan under lockdown

Fig. 3

Cumulative confirmed cases and new daily cases in Shanghai under lockdown

The substantial disparity in confirmed cases also reflects differences in public compliance behaviour. Compliance behaviour in the context of this study refers to the extent to which the public adheres to the government’s crisis management measures, including both proactive adherence (that is, voluntarily following preventive measures, such as social distancing and mask-wearing) and passive compliance (that is, acceptance of and compliance with restrictions owing to government enforcement, such as lockdowns or quarantines). This study utilizes data from the Global Database of Events, Language, and Tone (GDELT) to measure variations in non-compliance during the Covid-19 pandemic by analyzing the volume of news reports on public defiance. GDELT is the most comprehensive and widely covered open-access database, containing trillions of news event data points since 1 January 1979, with a global news coverage rate of 98.4%, ensuring high representativeness and reliability. The rationale for this method lies in the premise that the frequency of news reports can serve as an indicator of actual public behaviour under specific policy contexts, facilitating cross-temporal and cross-regional comparisons. GDELT data indicate that during Wuhan’s lockdown, the volume intensity of news reports mentioning public defiance or “protest” was 0.0192, whereas in Shanghai, it was 0.0516 – suggesting a significantly lower level of public compliance in Shanghai compared with Wuhan (Table 1).

Table 1 The volume intensity of public resistance news during lockdown periods in Wuhan and Shanghai

In summary, despite facing the same crisis and implementing similar pandemic control policies, Wuhan and Shanghai exhibited substantial differences in containment effectiveness and public compliance. To further explore the underlying causes of these disparities, we conduct a comparative analysis using multiple secondary data sources, including news reports, social media discussions and existing literature. Through this approach, we aim to provide an innovative perspective on public risk perception and compliance behaviour, offering valuable insights for global policy-makers in adequately addressing such crises in the future.

Sample

The primary research sample includes all Covid-19-related news articles published online by three major official Chinese news agencies (Xinhua News Agency, People’s Daily Online and China Central Television (CCTV)) during the Wuhan lockdown (23 January 2020 to 8 April 2020; 76 days) and the Shanghai lockdown (28 March 2022 to 1 June 1 2022; 65 days) (N = 248). Additionally, the study includes all Covid-19-related news articles published online by the Shanghai Release WeChat public account (the Shanghai Municipal Government’s official media channel) from January 2020 to June 2022 (N = 1,364). Content analysis was conducted using a combination of manual and computational coding methods, with coding categories inferred from prior research. Owing to data availability constraints, the study also incorporates previously published research data, including two epidemiological studies on Covid-19 and six sentiment analysis studies examining pandemic-related discussions on social media during the Wuhan and Shanghai lockdowns (Details in Appendix 2 and 3).

Content analysis

Three trained coders manually collected and coded all news articles using Microsoft Excel. One coder was an author, while the other two were trained and compensated research assistants. The first coder was responsible for collecting and coding all news articles published during the Wuhan lockdown (23 January 2020 to 8 April 2020), the second coder handled articles from the Shanghai lockdown (28 March 2022 to 1 June 2022), and the third coder coded all Covid-19-related articles published by Shanghai Release from January 2020 to June 2022.

To refine the coding framework, the two authors independently reviewed 320 news articles (20% of the sample). Subsequently, 20% (n = 320) of the news articles underwent reliability testing through triple coding. The four primary variables coded were: (1) scientific knowledge, (2) expert assessment, (3) government measures and (4) public opinion. Intercoder reliability was assessed using Krippendorff’s alpha, with randomly selected data from the full news sample. The intercoder reliability estimates, measured using Krippendorff’s alpha, ranged from 0.82 to 0.95.

For the expert assessment variable, we employed Gooseeker software to conduct keyword extraction, word frequency analysis, and sentiment analysis, enabling a comparative examination of expert evaluations during the Wuhan and Shanghai lockdowns. Each news article was assigned a single sentiment score – positive, negative, or neutral – ranging from −1 to + 1. Additionally, Covid-19 knowledge underwent secondary coding, which included four subcategories: (1) virus transmissibility, (2) medications, (3)vaccines and (4) viral virulence.

Analysis of risk perception in two cities

We assess the differences in collective risk perception between Wuhan and Shanghai based on five key variables identified within the power-knowledge framework: scientific knowledge, government measures, public opinion, expert assessment and international experience. We expect to understand how much differences emerge and the extent to which they influence public compliance behaviour. In doing so, we pay particular attention to trends in collective risk perception and compare them between the two cities, adding further insights.

Scientific knowledge

Wuhan was one of the earliest cities affected by the Covid-19 outbreak, during which both the scientific community and the public had extremely limited knowledge of the virus. From a scientific perspective, the earliest global scientific insights on Covid-19 originated from two articles published in China’s leading medical journals [57, 59]. Word frequency analysis indicates that the predominant epidemiological characteristics discussed included terms such as “coronavirus pneumonia”, “case fatality rate”, “asymptomatic carriers”, “silent infections”, “animal hosts”, “droplet transmission” and “universal susceptibility”. This suggests that scientific understanding of the virus remained constrained, with critical knowledge gaps regarding pandemic containment, including antiviral treatments, therapeutic strategies and long-term health effects.

From the public’s perspective, knowledge acquisition lagged behind scientific discoveries. A network analysis of the scientific knowledge dimension reveals that during Wuhan’s lockdown, official messaging to the public primarily emphasized five key points: (1) Covid-19 was “highly infectious”, with confirmed “human-to-human transmission” and the presence of “super-spreaders”, (2) many aspects of the virus remained “unknown”, (3) there was “no specific treatment” available, (4) “vaccine development” and (5) “strict prevention” were necessary (Fig. 4).

Fig. 4

A network analysis of the scientific knowledge dimension during Wuhan’s lockdown

Unlike Wuhan, which faced an unknown virus, by the time of the Shanghai lockdown, global understanding of Covid-19 had significantly advanced. From a scientific perspective, the virus had been identified as the Omicron variant, with its key characteristics – transmissibility, virulence, immune evasion and cross-immunity interactions – well-documented [24]. The reduction in scientific uncertainty tends to diversify individual behavioural choices [70, 73]. On one hand, official discourse emphasized that Omicron was “highly transmissible, spreads rapidly and has a highly concealed transmission process (0201114)”, and thus reinforced the “zero-COVID” strategy. High-frequency terms such as “zero-COVID”, “screening” “nucleic acid testing” and “epidemiological investigation” repeatedly appeared in official media narratives, reinforcing public risk perception. On the other hand, official messaging also acknowledged the “weakened virulence” of the virus, the rising proportion of “asymptomatic infections” and the “effectiveness of vaccination”, signaling a reduced individual threat from the virus.

Expert assessment

Owing to the lack of adequate knowledge and misunderstanding of the new crisis, the public will be willing to trust experts with common knowledge and collective wisdom when an unknown crisis arises [43]. An empirical study in China has found that the majority of the public has a high degree of trust in the high-level expert group of the National Health Commission, which was the most authoritative expert team in China during the outbreak response. The guidelines of pandemic prevention they delivered are regarded as “the Bible”, and people even subconsciously question the pandemic’s views that they do not express [72].

During the Wuhan lockdown, sentiment analysis of expert evaluations yielded an average score of −0.2, indicating a predominantly negative tone (Table 2). Experts initially confirmed that the outbreak was in a “rising phase” and warned of the virus’s strong interpersonal transmissibility. High-frequency terms in expert assessments included “human-to-human transmission” and “super-spreaders”, emphasizing the virus’s rapid spread. Additionally, owing to the absence of “targeted treatments” (010130), expert recommendations were limited to “early detection and isolation”, “mask-wearing”, ‘increased vigilance” and ”reduced mobility” and other basic self-protection measures. As confirmed cases surged, experts reported “chaotic conditions” in Wuhan, acknowledging severe shortages of medical resources and overwhelming hospital burdens [76]. These assessments conveyed a high-risk evaluation of the outbreak, exacerbating public anxiety, depression, stress and heightened risk perception related to Covid-19 [82].

Table 2 Sentiment analysis of expert assessments during the Wuhan and Shanghai lockdowns

During the Shanghai lockdown, expert assessments of the pandemic displayed significant divergence. Sentiment analysis revealed an almost equal distribution among positive (n = 18, 33.9%), negative (n = 17, 32%) and neutral (n = 18, 33.9%) evaluations (Table 2). On one hand, some experts confidently asserted that “preparedness for the new variant had been achieved” [44]. Keywords such as “treatment protocols”, “vaccination”, “three-party coordination” and “three-drug therapy” frequently appeared in expert discourse, emphasizing systematic medical interventions. On the other hand, some experts still warned that the Omicron variant could still cause severe illness and mortality, particularly among high-risk groups such as “the elderly and individuals with chronic underlying conditions”(0201145). Moreover, Omicron was described as exhibiting “increasing immune evasion capabilities” (0201159), with experts cautioning that “even if disease severity remains unchanged, a rising case count could still have a significant impact on public health” (0201155). This polarization of expert assessments may have contributed to variations in public risk perception and self-protective behaviour during the Shanghai outbreak.

Government measures

When people have limited time, knowledge, and cognitive capacity to evaluate risk deliberately, they must rely on trusted institutions to assess the risk of hazards [14, 69]. In China, the public exhibits a high level of trust in the government [79], which in turn influences their risk perception of Covid-19 and compliance behaviour in response to governmental actions.

Wuhan initiated its 76-day lockdown on 23 January 2020, with the emergency release of six Covid-19 control notices, implementing citywide lockdowns, stringent health screenings, travel restrictions and mandatory home isolation. These measures extended beyond the conventional definitions of localized containment, lockdowns and quarantine, conveying two clear messages to the public: (1) the outbreak had entered its most critical phase (010171) and (2) emergency medical supplies were running low (010172), highlighting the severity and complexity of the Covid-19 crisis in Wuhan. Between 24 January and 17 March 2020, Wuhan issued Notices No. 7 to No. 20 to address both the worsening internal outbreak and the external threat of imported cases. This series of control measures reflected the government’s firm stance on containment, fostering a climate of social tension and uncertainty that reinforced high-risk perception with pandemic control policies.

Before implementing citywide lockdown measures, Shanghai adopted a precision-based epidemic control strategy that relied on efficient contact tracing and large-scale nucleic acid testing. This approach not only contributed to achieving the zero-COVID objective to some extent but also balanced pandemic control efficiency with socioeconomic needs. In pre-lockdown contents from the Shanghai Release, high-frequency terms were “precision prevention and control”, “resumption of work and production”, “contact tracing”, “livelihood”, “nucleic acid testing” and “dynamic zero-COVID”, reflecting the city’s policy orientation of minimizing social costs while maintaining effective containment measures. However, with the emergence of the Omicron variant, Shanghai abruptly shifted to citywide static management, mandating that all residents “remain indoors at all times” (0201400) and warning that “those who leave home without authorization or refuse to comply will be dealt with according to regulations” (0201473). This sudden shift in governmental measures placed the public in a state of tension, caught between two contrasting policies: on one hand, a previously relaxed precision-based approach, and on the other, an immediate and absolute mobility restriction. This policy inconsistency and instability easily intensified public risk perception differences and social conflicts, contributing to collective emotional volatility and variations in compliance behaviour.

Social opinion

During the Wuhan lockdown, public sentiment in social opinion leaned toward positivity. In the early stages, discussions were dominated by the unknown nature of the virus, with primary concerns centred on topics such as “the closure of Huanan Seafood Market” and “unexplained pneumonia cases in Wuhan” (010206). While uncertainty generated some negative emotions, neutral sentiments remained predominant (010205). As the outbreak spread nationwide in the mid-stage, public attention shifted to infection and mortality-related topics, leading to a rise in negative emotions such as “fear”, “sadness” and “anger” (010226). However, owing to strong state-led control measures and the fact that only a limited number of regions were severely affected at the time, public confidence in overcoming the pandemic remained intact, with comments on related posts frequently including words such as “stay strong”, “stay safe” and “hope” (010217). In the later stages, as the epidemic situation gradually improved, negative sentiment declined significantly, while neutral sentiment increased markedly, and positive sentiment also showed an upward trend (010205).

During the Shanghai lockdown, negative sentiment dominated public discourse. In the early stages, high-frequency terms in the Shanghai Release centred on the rapidly increasing numbers of “new cases”, “confirmed cases” and “asymptomatic infections”, directly presenting the most severe aspects of the outbreak to the public [49]. And public reactions largely reflected “fear” over the stringent containment measures (020210), with negative emotions prevailing. As the lockdown progressed into the mid-stage, frustration and sadness over the prolonged home isolation and daily inconveniences surpassed fear, becoming the dominant negative emotions. Widespread shortages of essential supplies and soaring prices further intensified public dissatisfaction, making “commodity prices” and “livelihood security” the most salient topics of negative sentiment (020218). This period saw growing criticism and distrust toward government management and crisis response [31]. The prolonged lockdown’s economic and livelihood risks triggered a concentrated outbreak of collective negative emotions. The most frequently discussed topic among the public became how to strategize against local community and grassroots government authorities for self-preservation, significantly increasing the likelihood of conflicts induced by adverse events (020201).

International experience

According to the Baidu index provided by the most popular Internet search engine in China, the UK and the USA are the two countries that have attracted the most attention from the Chinese public since 2020 [4], which make them a representative country for our measurement of international experience.

As Wuhan entered lockdown, both the USA and the UK were also experiencing the onset of their respective Covid-19 outbreaks. Owing to delays in governmental response, the number of confirmed cases in the US surged sharply by mid-March. Although the country quickly declared a state of emergency, its national pandemic response remained disorganized, with widespread shortages of testing supplies, personal protective equipment (PPE), critical medical resources and healthcare personnel. As a result, the number of confirmed cases and deaths in the US far exceeded those in Wuhan (Fig. 5), making Covid-19 the leading cause of death in the country at the time [58] The UK, by contrast, initially pursued a herd immunity strategy to prevent excessive strain on the National Health Service (NHS). However, as confirmed cases rose rapidly, the government ultimately imposed a nationwide lockdown. By the later stages of Wuhan’s lockdown, the UK’s case count had already far surpassed that of Wuhan, and reports of severe PPE shortages became widespread [39].

Fig. 5

Confirmed cases and deaths in Wuhan, the USA and the UK from 23 January to 7 April 2020

In short, the Covid-19 situation in the UK and the USA was discouraging during the Wuhan pandemic. They did not implement the same strict restrictions as China and faced more catastrophic and intractable consequences, which were narrated to the Chinese public by the Chinese government. The scenario of the whole world facing the same dilemma heightened public risk awareness of the Covid-19 pandemic, while the comparatively more effective response of the Chinese government further strengthened public compliance with preventive measures.

During the Shanghai lockdown, the USA adopted a minimal-disruption approach to pandemic management, refraining from implementing any lockdown measures. Instead, it shifted its response focus to prevention, protection and treatment, providing Americans with the necessary tools to safeguard themselves, including free testing, vaccinations, expanded medical resources, epidemiological surveillance and safer work environments. The effectiveness of this strategy was evident, as the USA became the first major economy to surpass pre-pandemic levels of economic output. Similarly, the UK’s response strategy was encapsulated by the principle of “living with Covid-19”. It prioritized protecting the most vulnerable populations, replacing mandatory interventions with public health guidance and surveillance for the general public. The preventive interventions of both the USA and the UK, which emphasized vaccination, diagnostics, and treatment, offered the pandemic-fatigued Chinese public an alternative vision of returning to normal social life. However, the higher infection and mortality rates under this approach also served as a stark warning, creating a paradoxical perception of Covid-19 risks among the Chinese public.

Discussion

Why do we observe starkly different response outcomes and public compliance behaviours under the same crisis and similar containment measures? In this study, we explores the critical role of collective risk perception in shaping public compliance by comparing the Covid-19 responses in Wuhan and Shanghai, China. This may help explain Wuhan’s faster control of the Covid-19 impact despite various constraints in the early stages of a new virus outbreak.

In retrospect of the pandemic responses in Wuhan and Shanghai, distinct trends in collective risk perception are discernible between the two cities. The public in Wuhan exhibited a pronounced convergence in risk perception, whereas in Shanghai, collective risk perception manifested a diverse and even adversarial trend. This divergence directly resulted in disparities in public compliance behaviour, consequently influencing the respective outcomes of their responses to the Covid-19 crisis. We will further discuss these observations in the subsequent analysis.

Wuhan: convergence of public collective risk perception amid the chaos

In the Wuhan pandemic, the public collective risk perception demonstrate a consistent and unified trend across all five measurement dimensions. The lack of knowledge in the scientific community and the government about Covid-19 and its suppression has sent a message of an “extremely dangerous pandemic” to the public, making them full of fear of the unknown and take the strictest precautions to protect themselves from harm. The government measures have also been unprecedentedly stringent and coercive, and limit all outdoor activities of residents and infected people as much as possible to ensure the fastest control of the infection scope and trend. Experts, as another authoritative subject, issued a gloomy assessment of the Covid-19 threat and declared that there was no targeted and effective treatment for the disease, which further heightened public nervousness. At the same time, stringent containment measures and the predominantly positive tone of public discourse reinforced public confidence in ending the pandemic, leading to a willingness to comply with preventive measures. International experience has not helped and upsetting. The UK and the USA both have seen a surge in confirmed cases, far surpassing Wuhan in the late stages of the pandemic. Additionally, both face severe shortages of medical resources and overstretched hospitals, which also bring warning signs to the public.

Therefore, we can readily observe a convergence trend of collective risk perception among the public in Wuhan when Covid-19 knowledge is unknown, government measures are unprecedentedly draconian, expert assessments are pessimistic, social opinions instil confidence and, even worse, international experience does not help us. These circumstances have led to an unparalleled, unified and reinforced collective perception of risk, pointing towards a common direction, which is to shape public compliance and gain public active cooperation through amplifying the severity of the crisis while constructing an optimistic vision for resolution, ultimately shaping public compliance and securing collective cooperation, facilitating a swift response to overcoming the crisis (Fig. 6).

Fig. 6

Convergence of public collective risk perception in Wuhan

Shanghai: diversity of public collective risk perception in risk evolution

On the contrary, in the process of addressing Covid-19 outbreak in Shanghai, we can observe a trend of diversity in public collective risk perception across five measurement dimensions. The first comes from the munificence of Covid-19 knowledge. Public understanding of the virus was relatively comprehensive and in-depth, facilitated by scientific advancements and widespread information dissemination. Adequate knowledge generated varied perspectives among the public, enabling them to make personalized behavioural choices on the basis of their own judgments. The ongoing Covid-19 pandemic has given rise to an increasing number of experts expressing, bifurcate and even opposing perspectives. Some contend that the pandemic remained severe and required continued vigilance, while others offered a more optimistic assessment, arguing that existing preparations were sufficient. Shifts in government intervention further contributed to the diversification of collective risk perception, particularly as the initial “minimal impact” policy was rapidly replaced by a “strict containment” strategy. Nevertheless, the enduring conflict between prolonged stringent restrictions and the social needs of residents has led to opposing mainstream viewpoints in social opinion: official consensus emphasizes the severity of the pandemic and advocates for strict control, while voices on social media call for a relaxation of restrictions, arguing that the situation is not as dire. Furthermore, international experiences offered alternative approaches to pandemic response. Both the USA and the UK adopted a “living with Covid-19” strategy, achieving certain economic gains and meeting societal needs, despite a rise in confirmed cases. This has, to some extent, differentiated the shared collective risk perceptions of the public, who are gradually experiencing lockdown fatigue, resulting in decreased compliance.

Therefore, it is apparent that there is a diverse trend of the collective risk perception among the public in Shanghai when Covid-19 knowledge is known, international experience provides alternative choices, government measures are self-contradictory, expert assessments are highly differentiated, and social opinion from official consensus and social media has different demands. This directly results in a lack of unified adherence to government policies, with distinct behavioural choices emerging within the public. Moreover, the public begins to anticipate and advocate for alternative measures rather than mandatory restrictions. Consequently, government control is no longer as effective as witnessed in Wuhan (Fig. 7). 

Fig. 7

Diversity of public collective risk perception in Shanghai

The important role of collective risk perception

The comparative analysis of the responses to the Coovid-19 pandemic in Wuhan and Shanghai indicates that, in the face of a public health crisis, the public’s compliance with policies is influenced by their collective risk perception. A substantial and rapidly growing body of literature has partially validated this relationship, suggesting that psychological determinants, particularly risk perception, can influence compliance [40, 62]. Indeed, unless something poses a risk that can be perceived as a threat, why would one make efforts to mitigate it, especially when required to engage in burdensome activities (such as self-isolation) and restrictive measures (such as maintaining physical distance) [3]. However, a limitation of existing evidence is that it often conceptualizes risk perception at the individual level by self-report measurements. While the term “collective risk perception” is occasionally employed, the term “collective” refers more to individuals considering risk perception from a collective or social identity (aligned with the social identity corresponding to the various groups we belong to). When societal identity is salient, people tend to act in accordance with the norms, values, and interests of their affiliated group rather than their personal interests [65]. However, in practice, this still does not fully transcend the individual-level perspective of risk perception.

Therefore, when considering the overall compliance differences among the public in the two cities, we expect to conceptualize the public as a collective entity, examining how the collective risk perception of this group influences its compliance behaviour. Correspondingly, as we observe risk perception from a collective perspective, we find that it exhibits trends of both convergence and diversity rather than a simple criteria of strength. Convergent collective risk perception tends to result in high public compliance, while divergent collective risk perception leads to lower compliance, thereby impacting the effectiveness of crisis response. On the basis of this, we have constructed an analytical framework of public compliance behaviour, specifically focusing on how collective risk perception influences public compliance and policy outcomes in crisis.

In the analytical framework of public compliance behaviours, we propose evaluating the collective risk perception of the public from five dimensions: government measures, scientific knowledge, social opinions, expert assessments and international experiences. We also introduce a creative perspective on the trends in collective risk perception changes, namely, the tendencies towards convergency and diversity, leading to different public compliance behaviours. Notably, our observation perspective consistently remains at the group level, focusing on public groups rather than individual entities to observe changes in risk perception within the collective influenced by the five factors. This approach is advantageous in eliminating individual differences, enabling a more systematic interpretation of changes in collective behaviour under similar policies. Ultimately, we argue that convergent collective risk perception tends to result in high compliance behaviour, whereas diverse collective risk perception leads to low compliance behaviour (Fig. 8).

Fig. 8

The analytical framework of public compliance behaviours

Conclusions

During the Covid-19 pandemic, cities across China implemented similar containment policies to address the same issue, providing an opportunity to explore discussions on the differentiation in public compliance behaviour. In this study, we investigate the reasons behind variations in public compliance among different cities facing the same crisis under containment policies. To this end, we utilized evidence from two mainland Chinese cities, Wuhan and Shanghai, to discover the significant role of collective risk perception. We further explicate the impact of differentiated collective risk perception on public compliance behaviour and the effectiveness of risk response measures across five dimensions: scientific knowledge, expert assessments, government measures, social opinions, and international experiences. Additionally, our findings reveal distinctive trends in collective risk perception between Wuhan and Shanghai. The former demonstrates a convergence trend associated with heightened public compliance, while the latter exhibits a diversity trend correlating with lower levels of public compliance. Ultimately, based on these findings, we have formulated an analytical framework for understanding public compliance behaviour.

In the realm of risk perception and public compliance, many analytical interests have yet to be fully revealed. We identified and analysed the transformation of collective risk perception in these two cities, exploring how this dynamic change affects public compliance behaviour and response outcomes during the crisis. In comparison to previous research, we took a creative approach by discussing the impact of risk perception from a collective level, asserting that the convergence and diversity of collective risk perception are crucial factors influencing the differentiation in compliance behaviours. Through comparative studies, we elucidated how this impact is realized.

Our findings provide compelling evidence for governments worldwide, recommending that they incorporate the development and control of public collective risk perception into their crisis management practices. During a crisis, the effectiveness of government measures largely depends on the public’s willingness to comply, which in turn is shaped by their perception of the crisis. This requires governments to ensure continuous and transparent communication regarding the evolving nature of the crisis and the rationale behind policy decisions to reduce internal societal confusion, scepticism and polarization, and to prevent conflicting narratives from undermining public trust. Additionally, it is crucial to recognize that the role of knowledge authorities, such as experts, is much more significant in crisis contexts than typically assumed. Therefore, governments must coordinate experts to minimize information conflicts and ensure the consistency of risk communication. Furthermore, attention should be given to guiding public discourse on social media to align with government objectives, ensuring that citizens perceive the measures as appropriate, reasonable and collectively necessary. Ultimately, a unified public understanding can significantly enhance the effectiveness of crisis management.

It is important to note that our discussion on collective risk perception at the group level does not imply neglecting individual risk perception. Instead, we approach the public as a whole to understand how changes in psychological motivations within the collective influence group behaviour, leading to different policy outcomes. In fact, discussions on individual risk perception and collective risk perception have long been present. It has been found that the formation of risk perception differs on the basis of whether individuals consider their self-interest or personal group identity. This realisation highlights the divergence between individual risk perception and collective risk perception. However, a potential oversight is whether there are differences between individual risk perception and collective risk perception. Suppose there are no intergroup differences in risk perception, meaning that the crisis presents the same high risk for both young and old individuals rather than differing risks as seen in the case of Covid-19. In that case, individual risk perception and collective risk perception may appear similar, making comparisons challenging. Therefore, we choose to discuss public risk perception at the group level to avoid potential debates arising from this issue.

As with all other studies, this research has some limitations. The first is the heavy reliance on secondary data. While secondary data provides a large amount of reference information and effectively supports our analytical framework, we must also acknowledge that such data may not fully reflect the true situation of pandemic response. For instance, differences in pandemic reporting standards, data collection methods and public health systems across countries may affect the accuracy of the data. This is especially significant when making cross-country comparisons, where data biases may be more pronounced. Therefore, future research that incorporates more primary data may offer a more comprehensive presentation of public reactions under different contexts. Secondly, there are challenges in measuring collective risk perception. Collective risk perception is a complex psychological and social phenomenon with multiple levels and dimensions. We chose to approach this by examining the factors influencing public collective risk perception from the perspective of “power-knowledge” in the objective environment to avoid bias stemming from internal differences within the group. However, from a broader perspective, collective risk perception is also influenced by group characteristics. Therefore, in practice, accurately quantifying these perceptions and determining their relative importance is a problem that warrants further exploration.

In summary, although this study highlights the important relationship between collective risk perception and public compliance behaviour, these limitations indicate that future research still requires richer empirical data and theoretical support to further enhance the understanding and optimization of pandemic crisis response.