The COVID-19 pandemic led to the implementation of various strategies to limit the virus's spread globally. However, significant public resistance to these measures arose, fueled by mistrust in government and the spread of misinformation. This study addresses the question of why citizens support (or oppose) protective behaviors. Existing research suggests that support stems from both self-interest (fear of infection) and prosocial motives (concern for others). However, a crucial gap exists in understanding how these motives interact with sociocultural contexts, specifically the level of trust in the government. This understanding is vital for developing effective strategies to mobilize support for public health initiatives during crises. The study aims to examine how individual and country-level trust in government moderates the relationship between self-oriented fear and other-oriented empathic concern regarding COVID-19 and support for preventive behaviors. Mobilizing people for large-scale cooperation is challenging due to uncertain incentives and the difficulty in tracking individual contributions. Cooperation can be driven by both altruistic and selfish motives, and the balance between these motives may vary across different contexts, with trust in the government playing a key role. While some evidence suggests a link between trust in government and cooperation, the relationship is not straightforward. This study proposes that trust in government acts as a boundary condition, moderating the relationship between support for COVID-19 containment behaviors and both self-oriented and other-oriented motivations. It argues that under conditions of low trust, unpredictability increases, making individuals doubt the effectiveness of their sacrifices. This should hinder empathy-driven cooperation, while increasing the reliance on self-protective behaviors driven by fear. Conversely, high trust fosters confidence in the collective benefit of individual actions, strengthening the impact of empathic concern. The study uses a large international sample to test these hypotheses, accounting for various socio-cultural and pandemic-related factors. The diverse sample includes both high and low-income countries and those severely and less affected by the pandemic, allowing for a comprehensive examination of the proposed moderating effect of trust in government.

The literature review explores previous research on the factors influencing compliance with COVID-19 preventive measures. Studies have shown that self-interest, particularly fear of infection, is a key driver of such behaviors. Other research highlights the role of prosocial motives, such as empathy and altruism, in motivating individuals to engage in protective behaviors for the benefit of others. However, there’s a lack of consistent findings across different cultural contexts and levels of societal trust. Some studies found other-oriented concerns more influential than self-oriented fear, while others found the reverse. The existing research emphasizes the importance of trust in motivating large-scale cooperation, particularly in the context of uncertainty. Trust, especially in authorities like government officials, is crucial in promoting collective action. However, evidence regarding the specific relationship between trust in government and COVID-19 compliance is mixed. Some studies found no relationship between trust and adherence to guidelines, while others observed weak associations. This inconsistent evidence highlights the need for further investigation into the moderating role of trust in shaping the relationship between motives and behavior. The literature also discusses the challenges of mobilizing large-scale cooperation, particularly when uncertain incentives and the lack of traceability of individual contributions are involved. Previous studies have demonstrated that the level of predictability and perceived effectiveness of actions influence an individual's willingness to engage in cooperative behaviors.

This study employed a multi-level regression analysis using a large, cross-national sample of 12,758 adults from 34 countries. The data were collected through online surveys over approximately one year (February 2021 to December 2021), using a convenience sampling method. Participants were predominantly young adults (mean age 26.9 years), with a higher proportion of women (67%). The data underwent several cleaning steps: removal of respondents under 18 or without age data, listwise exclusion of cases with missing values, and exclusion of countries with less than 150 complete responses. The final dataset included 12,758 participants from 34 countries. Several measures were used. Trust in government was assessed using four items adapted from Kerr et al. (1999) for the individual level and using data from the World Values Survey (2017-2020) for the country level. Empathic prosocial concern was measured using a three-item scale from Pfatteicher et al. (2020), and fear of COVID-19 was assessed using a seven-item scale (Ahorsu et al., 2022). Support for COVID-19 containment behaviors was measured using an extended version of the Tepe and Karakulak (2021) scale, comprising 13 items. Data analysis involved a four-step multi-level regression analysis. Step 1: Null model. Step 2: Fixed-effects model with random intercepts and fixed slopes. Step 3: Random slopes model allowing slopes of predictors to vary across clusters. Step 4: Addition of interaction terms. Covariate effects for age, gender, and various country-level variables (HDI, hospital beds, data collection month, government stringency level, daily COVID-19 cases and deaths) were included in Steps 2-4. The study also included an exploratory analysis to examine the robustness of the findings when generalized trust was added as a covariate and using generalized trust instead of governmental trust as a moderator. Finally, a fixed effects regression model with cluster-robust standard errors was used to check robustness. A power analysis was conducted a priori to determine the necessary sample size per country.

The analysis revealed significant main effects across all models. At the individual level, higher trust in government, greater empathic concern, and stronger fear of COVID-19 were all significantly associated with greater support for COVID-19 containment behaviors. The random slopes model confirmed that these relationships varied across countries. Importantly, significant interaction effects were found at the individual level. Trust in government moderated the relationship between both empathic concern and fear of COVID-19, and support for containment behaviors. The association between empathic concern and support was strongest when trust was high, while the association between fear of COVID-19 and support was strongest when trust was low. These findings were graphically represented (Figs. 1, 2). At the country level, analyses based on World Values Survey data revealed a different pattern. The main effects of empathic concern and fear of COVID-19 on support for containment behaviors remained significant, but there was no significant association between country-level trust in government and support for containment behaviors. Furthermore, while the interaction between fear of COVID-19 and country-level trust was significant, supporting the hypothesis, the interaction between empathic concern and country-level trust was not. Results are graphically represented (Fig. 3). The exploratory analysis adding individual-level generalized trust did not significantly alter the findings. Likewise, using generalized trust as a moderator in place of trust in government did not provide additional evidence for significant interactions. Results using a fixed effects model with cluster-robust standard errors supported the key findings.

The findings show that trust in government plays a significant moderating role in the relationship between individuals' motivations and support for COVID-19 containment behaviors. At the individual level, high trust strengthens the positive association between empathy and support for preventive measures, whereas low trust enhances the influence of fear on behavior. At the country level, the effect of trust in government on behavior is less prominent, with the effect of fear being more significant, although empathy still showed a substantial influence. The relatively similar effects of empathy on compliance across high and low trust contexts suggests empathy may be a consistent motivator, regardless of the level of trust. The greater effect of fear under low trust contexts aligns with the idea that under conditions of uncertainty, individuals prioritize self-protection. The exploratory analyses confirmed the relative robustness of findings with the use of generalized trust. The overall significance of both empathy and fear suggests both motivations are relevant across different trust contexts, although their relative importance varies depending on the level of trust in the government. These results highlight the importance of both individual and contextual factors in promoting public health compliance during crises. The unique role of trust in government underscores that effectively promoting collective action requires more than just relying on citizen goodwill.

This study highlights the importance of trust in government as a critical moderator in the relationship between individual motivations (empathy and fear) and support for public health measures during a pandemic. The individual level findings demonstrate that trust enhances the impact of empathy, while fear is more influential when trust is low. Country-level analysis revealed similar trends, especially regarding the significant role of fear in low-trust settings. The robustness of the findings across various analyses underscores the need to consider both individual-level and societal-level factors in public health interventions. Future research should explore causal relationships and test these findings in different crisis situations.

The study's correlational nature limits conclusions about causality. Although the large, diverse sample enhances generalizability, potential sampling biases related to age, risk level, and internet access may limit the representation of at-risk populations. The reliance on self-reported support for behaviors, instead of observed behavior, may also affect the interpretation of results. The small magnitude of some interaction effects should be considered. Future research should focus on addressing these limitations by employing experimental designs and collecting data from more representative samples.