A small global village: the effects of collectivist, tight and Confucian cultures on the spread of COVID-19

The study addresses why COVID-19 spread and outcomes varied drastically across countries and regions, focusing on cultural determinants rather than only demographic, economic, or policy factors. The central research questions are: (1) whether countries and regions with collectivistic, Confucian-oriented, or tight (restrictive) cultures exhibit lower growth rates of COVID-19 cases than those with individualistic, non-Confucian, or loose cultures; and (2) whether these cultural effects become stronger during lockdowns. Motivated by the role of culture in shaping behavior, social norms, and policy compliance, the authors argue that collectivistic norms emphasizing the common good, as well as tight cultural norms with low tolerance for deviance, may foster greater adherence to preventive measures (e.g., mask-wearing, social distancing) and support for stringent interventions, thereby reducing spread. The study aims to quantify these cultural associations with COVID-19 spread dynamics using a cross-country analysis and to evaluate how lockdown contexts moderate these effects.

Prior research underscores multiple non-cultural factors affecting COVID-19 spread and control, such as travel restrictions, age structure, GDP per capita, prior epidemic exposure, smoking prevalence, and public health strategies. Cultural influences have been discussed conceptually, including higher mask usage in collectivistic societies and the role of shared norms in responses to collective crises (e.g., climate change). Empirical evidence on collectivism and COVID-19 has been mixed: Webster et al. (2021) found no robust negative association (and even positive associations within the U.S.), whereas Jiang et al. (2022) and Maaravi et al. (2021) reported fewer cases in collectivistic contexts. This study differentiates itself by focusing on case growth rates rather than cumulative case counts, enabling analysis of temporal dynamics and the varying salience of culture, particularly during lockdown periods. The work also incorporates Confucian cultural classification and tightness-looseness as complementary cultural dimensions to triangulate cultural effects.

Data and period: Daily COVID-19 data were sourced from Our World in Data for 190 countries/regions from 01/22/2020 (WHO emergency committee meeting) to 12/31/2020 (pre-widespread vaccination), yielding 55,184 country-day observations. Hofstede individualism scores were used to construct collectivism, and additional cultural measures included a Confucian cultural-circle indicator and tightness-looseness scores for a subset of countries. Continuous variables were winsorized at 1% and 99%. Key variables: - Dependent variable (spread): Weekly case growth rate CASE_GROWTH_it = ln(1+COVID_CASES_it) − ln(1+COVID_CASES_i,t−7), mitigating weekday–weekend reporting biases. A parallel weekly death growth DEATH_GROWTH was defined similarly for deaths. - Main cultural predictors: • COLLECTIVISTIC = 1 / [ln(INDIVIDUALISM)], with higher values indicating more collectivistic orientation (Hofstede). • CONFUCIAN: Dummy = 1 for China, Hong Kong, Indonesia, Japan, South Korea, Malaysia, the Philippines, Singapore, Thailand, Vietnam (Confucian cultural circle), 0 otherwise. • LN(TIGHTNESS): Natural log of cultural tightness scores from Gelfand et al. (2011) for 31 countries (merged sample N≈10,095 observations). Controls: LN(POPULATION), AGE65 (percent ≥65), LN(GDP) per capita, SARS (prior SARS experience in 2003), SMOKE (smoking prevalence), religion shares (PROTESTANT, CATHOLIC, MUSLIM), and LN(LENIENT_ENFORCEMENT) indicating more lenient (weaker) policy enforcement. Continent and month fixed effects were included. Empirical strategy: Ordinary least squares (OLS) regressions of CASE_GROWTH (and separately DEATH_GROWTH) on cultural measures with controls and fixed effects. Cultural effects were also examined under different policy contexts by splitting the sample into lockdown vs. non-lockdown periods based on lockdown start/end dates compiled from Wikipedia, then estimating models separately and statistically testing coefficient differences (χ² tests). Additional corroboration used tightness-looseness regressions in the reduced 31-country sample. The study also examined cultural effects on weekly death growth to test consistency across outcomes.

- Collectivism and spread (Table 2): COLLECTIVISTIC is negatively associated with CASE_GROWTH. In the fully controlled model, β = -25.094 (t = -5.662, p < 0.001). Given σ(COLLECTIVISTIC) = 0.055, a one standard deviation increase corresponds to a 1.38% reduction in weekly case growth (0.055 × 25.094% = 1.38%). Model (1) R² = 0.573 with only COLLECTIVISTIC; Model (2) R² = 0.608 with controls. - Confucian cultural circle and spread (Table 3): CONFUCIAN is negative and significant across specifications. With controls, β = -7.598 (t = -12.665, p < 0.001), implying Confucian-circle countries/regions have on average 7.598% lower weekly case growth than others. R² ≈ 0.594 with controls. - Lockdown moderation (Table 4): Cultural effects strengthen during lockdowns. For COLLECTIVISTIC, β_lockdown = -73.454 (t = -4.201, p < 0.001) vs β_non-lockdown = -22.512 (t = -4.658, p < 0.001). The coefficient difference is significant (χ² p = 0.008). Interpreting magnitudes: a one SD increase in COLLECTIVISTIC reduces weekly case growth by 4.04% during lockdown (0.055 × 73.454%) vs 1.24% during non-lockdown (0.055 × 22.512%). - Lockdown moderation for Confucian (Table 5): CONFUCIAN β_lockdown = -26.169 (t = -10.290, p < 0.001) vs β_non-lockdown = -7.841 (t = -11.429, p < 0.001); difference = -18.328 (p < 0.001), confirming stronger cultural effects during lockdowns. - Tightness-looseness (Table 6): LN(TIGHTNESS) is negatively associated with CASE_GROWTH (β = -4.300 without full controls; β = -6.864 with controls; both p < 0.001) in the 31-country subsample (R² up to 0.679), corroborating that tighter cultures have lower spread. - Death growth (Table 7): Cultural measures are negatively related to weekly death growth. COLLECTIVISTIC: β = -28.309 (t = -8.433, p < 0.001); CONFUCIAN: β = -11.596 (t = -23.172, p < 0.001); LN(TIGHTNESS): β = -3.806 (t = -5.505, p < 0.001). Results control for the same covariates and fixed effects. - Controls: Lower spread associated with prior SARS experience (e.g., β ≈ -2.309 in Table 2), higher GDP per capita (e.g., β = -1.980), higher age 65+ share (e.g., β = -0.446). Higher spread associated with larger population (β ≈ 1.1–1.7), greater smoking prevalence (positive coefficients in most models), higher Muslim or Catholic shares (positive in several models), and more lenient enforcement (positive coefficients). Protestant share is often negatively related to spread. Overall, countries/regions with collectivistic, Confucian, and tight cultures have lower COVID-19 case and death growth rates, with effects amplified during lockdowns.

The findings directly support the hypotheses that collectivistic, Confucian-oriented, and tight cultures are associated with lower COVID-19 spread and that these cultural effects intensify under lockdown conditions. Mechanistically, collectivistic norms emphasize the common good and compliance with social norms (e.g., mask-wearing, distancing), while tight cultures enforce rules and have low tolerance for deviance, facilitating coordinated responses to threats. The amplification during lockdowns suggests that stringent policies are more effective where cultural norms support collective action and rule adherence. These results highlight culture as a key contextual factor conditioning the effectiveness of public health policies, helping explain cross-regional differences in outcomes (e.g., stronger effects in East Asia relative to Western Europe and North America). The consistent negative associations with death growth further reinforce the robustness of cultural effects beyond case dynamics.

This study contributes by: (1) analyzing growth rates of COVID-19 cases (and deaths) rather than static counts, revealing dynamic cultural impacts; (2) demonstrating that collectivistic, Confucian, and tight cultures are associated with lower spread; and (3) showing that cultural effects strengthen during lockdowns, indicating culture conditions policy effectiveness. Practical implications include crafting public health messaging and policies that emphasize common interests, personal responsibility, strong norms, and community solidarity, especially where individualistic norms prevail. International collaboration is recommended given global interdependence. Future research should extend beyond the examined cultural dimensions, consider dynamic cultural changes over time, and incorporate richer epidemiological and behavioral structures (e.g., higher-order interactions) to better capture real-world spread processes.

- Modeling approach: OLS regressions may not capture real-world epidemic processes or higher-order network interactions that can drive superspreading and complex dynamics. Future work should model higher-order interactions and variant competition dynamics explicitly. - Cultural dynamics: Culture was treated as fixed over the study period, though cultural attitudes and compliance may evolve (e.g., fatigue with prolonged lockdowns), suggesting potential time-varying effects. - Cultural scope: Only collectivism–individualism, Confucian classification, and tightness–looseness were examined; other cultural dimensions may also matter. - Data limitations: For death growth analyses, key health covariates such as comorbidities were not included due to data constraints, which may affect interpretation. Tightness analyses were limited to countries with available scores (31), reducing generalizability. - Period coverage: Analysis ends 12/31/2020 and does not account for widespread vaccination effects or later variants, which may alter cultural impacts.