The COVID-19 pandemic, a globally disruptive event, triggered numerous civil resistance movements worldwide. These protests, often directed against governmental institutions, provide a unique opportunity to study the emergence of civil unrest under extreme societal pressure. Existing research on civil resistance highlights the roles of grievances, political opportunity, and mobilization capacity as key factors. However, the rapid and multifaceted impact of the pandemic necessitates a more nuanced understanding of how these factors interact and influence protest emergence across diverse national contexts. This study addresses this need by developing a comprehensive model that incorporates various societal conditions, both fast-changing (e.g., infection rates, containment measures) and slow-changing (e.g., socioeconomic inequalities, political freedoms), to analyze the emergence and intensity of protests related to the COVID-19 pandemic. The study uses data from multiple sources to provide a holistic view and to test specific hypotheses related to the influence of both fast and slow-changing conditions.

The existing literature offers several perspectives on the emergence of civil resistance movements. Grievance approaches emphasize the role of underlying societal conditions that fuel discontent and motivate collective action. These grievances may include poverty, state-led discrimination, lack of civil liberties, and economic instability. Political opportunity approaches focus on the contextual factors that create openings for mobilization, such as weakened government legitimacy, political transitions, and the capacity of the government to effectively adapt to crises. Finally, mobilization capacity emphasizes the organizational and logistical aspects of creating and sustaining a protest movement. This research integrates these perspectives, acknowledging the complex interplay of factors and the influence of different timescales involved.

This research employs a mixed-methods approach, combining qualitative literature review with quantitative and computational modeling. First, a causal loop diagram (CLD) was constructed to visually represent the relationships between various societal conditions and the emergence of protests. This diagram incorporated factors related to grievances, political opportunity, mobilization capacity, and government interventions in response to the pandemic. Second, a dataset was constructed from diverse sources (OECD, World Bank, UN, ACLED, Freedom House, Oxford COVID-19 Government Response Tracker, Google Mobility Reports, The Economist) to quantitatively analyze the relationship between these factors and protest events across 27 countries during 2020. Third, statistical analysis using zero-inflated negative binomial (ZINB) and negative binomial (NBM) regression models were used to assess the impact of different variables (fast-changing and slow-changing) on the likelihood and intensity of protests. Finally, a system dynamics model (SDM) was developed, based on the CLD and statistical results, to simulate the temporal dynamics of protest emergence by explicitly modeling latent processes like societal tension and mobilization. This model incorporated two theoretical frameworks: one focusing on societal tension and its triggers, and another on the dynamics of mobilization and disengagement. The model was calibrated using data from four countries (Italy, Spain, USA, and the Netherlands) to assess its explanatory power.

The statistical analysis revealed a complex interplay between societal conditions and protest emergence. The zero-inflated negative binomial regression models showed that fast-changing variables (e.g., containment measures, unemployment, economic support) significantly influenced the likelihood of protests, while slow-changing variables (e.g., group grievances, political rights) were more strongly associated with the intensity of protest events. The systems dynamics model, while showing good fits for some countries (Spain and the Netherlands), struggled to accurately represent the protest dynamics in others (USA and Italy), potentially due to the complexities of differing national contexts, variations in the nature of protests, and data limitations. This highlights the limitations of generalizing models across diverse national contexts. Specifically, lower containment measures were correlated with an absence of COVID-19 related protests. Unemployment increased the likelihood of protests, while economic support measures had the opposite effect. Repression showed less impact than expected by some previous literature. The analysis underscored that phases without protests were often characterized by a lack of specific social dynamics, suggesting the critical role of societal tension as a trigger for protest activity. Slow variables (structural factors) showed a stronger relationship to the intensity of protests than fast variables. This suggests that slow-developing societal conditions may provide the context in which fast-changing events trigger widespread protests.

The findings provide valuable insights into the multifaceted nature of protest emergence during the COVID-19 pandemic. The integration of fast and slow variables in both statistical and computational models enhances our understanding of the timing and severity of protests. While the statistical models provided modest predictive power, the systems dynamics model, though not universally applicable, offered a dynamic representation of the interplay between societal tension, mobilization, and government responses. The discrepancies in the SDM's explanatory power across different countries highlight the need for more nuanced, context-specific models, potentially incorporating factors such as the specific motivations behind protests and the influence of news media and social media. The study confirms the hypothesis that both fast-changing events and slow-developing structural conditions are critical factors in determining both the likelihood and intensity of protests.

This research presents a holistic approach to understanding the emergence of civil resistance during the COVID-19 pandemic by combining qualitative analysis, statistical modeling, and system dynamics. The findings underscore the importance of considering both rapid societal changes and long-term structural factors. Future research could benefit from more granular data, including information on protest participants, motivations, and media influence, to improve the accuracy and generalizability of models. Agent-based modeling could provide a more fine-grained understanding of individual behaviors and social contagion, while incorporating sentiment analysis from social media could offer richer insights into public perception and evolving grievances.

The study faces limitations related to data availability and model generalizability. The reliance on existing datasets restricts the analysis to 2020 and primarily covers WEIRD societies. The inclusion of the Black Lives Matter movement as a potentially confounding factor was acknowledged, but its specific impact could not be fully quantified. Event data biases are inherent to datasets such as ACLED, potentially affecting model accuracy. Furthermore, the SDM showed limited explanatory power for some countries, underscoring the complexities of translating theoretical models to specific national contexts. Improvements in data granularity, including more detailed information on protest motivations and media coverage, are needed for more robust and generalized models.