Divergent Emotional Responses and Social Connections in China and the United States During the COVID-19 Pandemic

The study examines how public emotions and social connections evolved during the COVID-19 pandemic in China and the United States, and how these patterns differed across cultural contexts. Motivated by evidence that emotions like anxiety, anger, and sadness shape risk perception and behavior during crises, the research investigates both negative (anxiety, anger, sadness) and positive emotions as expressed on social media. It focuses on three types of social connections—family, collective, and country—to assess their roles in buffering negative emotions or fostering positive ones over distinct phases of the pandemic (baseline, warning, isolation, normalization). The importance of a cross-cultural, longitudinal approach is emphasized given different cultural norms (collectivism vs. individualism), policies, and crisis trajectories in China and the U.S.

Prior work shows crises often increase negative and reduce positive emotions on social media, but self-efficacy, meaning-making, and cultural factors can bolster positive emotions. Research highlights that anxiety rises under uncertainty, anger when needs are blocked, and sadness with loss. Cultural paradigms shape emotion expression and regulation; collectivist contexts emphasize harmony and cooperation, potentially altering how emotions are expressed and how social ties function in crises. Earlier studies explored changes in social connections at micro (work networks) and macro (shifts toward family) levels, but rarely differentiate among family, collective, and country ties in relation to emotions across cultures. Cross-cultural risk perception research shows structural similarities but varying salience of factors, warranting refined cross-cultural analyses of emotion during prolonged crises like COVID-19.

Data: Social media posts from China’s Weibo (covering 355 cities) and the U.S. (Reddit and X/Twitter across 18 major cities). Analyses focus on early pandemic periods corresponding to four phases—baseline, warning, isolation, normalization—illustrated by temporal plots aligned with daily new COVID-19 cases. Measures: Emotional expressions (anxiety, anger, sadness, positive emotion) were quantified as proportions of posts exhibiting each emotion. Social connection expressions were categorized into three types: family, collective, and country. Lexicon-based methods consistent with LIWC and its Chinese adaptation were used to identify emotional and social-connection language. New case counts were transformed as log(Newcases+1) and percentage changes were used for interaction visualizations. Design and models: (1) Descriptive time-series analyses tracked the three-day rolling average of emotion proportions by country across phases. (2) Event-study analysis (following robust designs in the literature) estimated dynamic effects of lockdown policies on the natural log change rate of social-connection expressions over 12 weeks (−5 to +6 weeks relative to lockdown), with 95% CIs. (3) Panel regressions tested how new case counts related to negative and positive emotions, and how social connections moderated these relationships. Separate models included family, collective, and country as moderators. Specifications included city and date fixed effects, control variables, and robust standard errors. (4) Johnson–Neyman plots illustrated regions of significance for interactions between changes in new cases and social connections on emotions. (5) Topic modeling compared thematic emphases within each social-connection category by country (family, collective, country), reporting top topic labels and percentages. Robustness: To enhance comparability, analyses were repeated using the 19 largest Chinese cities (4 municipalities, 15 sub-provincial) matched to 18 U.S. cities and with all U.S. users who mentioned COVID-19; results were consistent with main findings.

- Temporal emotion dynamics: Distinct phase transitions in both countries were observed across baseline, warning, isolation, and normalization, with trajectories aligned to new case counts. Chinese and U.S. users exhibited different patterns in anxiety, anger, sadness, and positive emotion over time. - Topic differences by social connection: Table 3 shows that U.S. “family” discussions centered on Relationships (23.04%), Education (9.31%), Healthcare (8.70%), Finance (7.10%), Daily Activities (5.10%), whereas China’s “family” emphasized Public Health (18.11%), Pandemic (14.65%), Relationships (8.69%), Encouragement (7.88%), Local Affairs (4.41%). For “collective,” U.S. focused on Relationships (11.63%), State Politics (10.55%), Global Affairs (9.59%), Encouragement (6.36%), Entertainment (5.86%); China emphasized Encouragement (13.81%), Pandemic (9.31%), Public Health (7.36%), Entertainment (4.66%), Local Affairs (2.73%). For “country,” U.S. centered on State Politics (23.78%), US Elections (19.2%), Global Affairs (7.29%), Pandemic (5.93%), Finance (4.04%); China emphasized Encouragement (16.98%), Pandemic (13.34%), Global Affairs (10.81%), Public Health (8.43%), Charity (3.95%). - Moderation by social connections (Table 4): • China—Negative emotions: Family b = −0.146 (SE 0.045, p<0.001) and Country b = −0.102 (SE 0.043, p<0.05) were inversely related; Collective b = −0.012 (SE 0.015, ns). Positive emotions: Family b = −0.204 (SE 0.054, p<0.001) inversely related; Collective b = 0.605 (SE 0.018, p<0.001) and Country b = 0.477 (SE 0.013, p<0.001) positively related. • United States—Negative emotions: Family b = 0.312 (SE 0.108, p<0.01), Collective b = 0.183 (SE 0.051, p<0.01), Country b = 1.084 (SE 0.173, p<0.001) positively related. Positive emotions: Family b = −0.246 (SE 0.157, ns), Collective b = −0.679 (SE 0.073, p<0.001), Country b = −0.562 (SE 0.258, p<0.05) negatively related. - Interaction effects (Fig. 8): In China, stronger “collective” ties mitigated negative emotions as new cases changed; stronger “country” ties reduced negative emotions, especially when case growth increased. In the U.S., stronger “family” ties were associated with higher negative emotions when percentage change in new cases was below ~1.5, but lowered negative emotions when growth exceeded ~2. For positive emotions, stronger “collective” ties increased positives in China (especially when case growth was smaller) but decreased positives in the U.S., with the reduction more pronounced at lower case growth. - Lockdown and social ties (Fig. 7): Event-study estimates showed significant, time-varying effects of lockdown on family, collective, and country expression rates around the policy dates in each country (with 95% CIs), indicating different dynamics of social connection salience across weeks before and after lockdown.

Findings indicate marked cross-cultural differences shaped by collectivist (China) versus individualist (U.S.) norms. In China, collectivist values and identification with collective and national entities aligned with lower negative emotions (via family and country mentions) and higher positive emotions (via collective and country), consistent with cooperative, harmony-preserving norms. In the U.S., emphasis on individual rights and present-focused concerns corresponded to higher negative emotions and lower positive emotions associated with mentions of family, collective, and country, and greater attention to state politics and finance in topic analyses. Time orientation also differed: Chinese users showed future-oriented concerns (anticipatory anxiety), whereas U.S. users were more present-focused (sadness/anger tied to current unmet expectations). Interaction patterns suggest that context (e.g., magnitude of case growth) conditions whether social connections buffer or exacerbate emotions. These results underscore the importance of culturally attuned risk communication and policy strategies that leverage culturally salient social connections to support public emotional well-being during prolonged crises.

This study contributes a cross-cultural, longitudinal analysis of public emotions and social connections during COVID-19 across China and the U.S., distinguishing family, collective, and country ties and elucidating their moderating roles. It shows that in China, country and collective connections bolster positive emotions and attenuate negatives, while in the U.S., family connections can buffer negative emotions under severe outbreaks but collective ties tend to dampen positive emotions. Topic patterns clarify differing societal emphases that align with cultural norms. Implications include tailoring crisis communication to leverage culturally resonant social connections and recognizing dynamic effects across pandemic phases. Future research should integrate multi-source data, incorporate fine-grained positive emotion subtypes, and account for confounders through multilevel models to deepen understanding of emotional and social dynamics in crises.

- Reliance on social media data may introduce demographic and behavioral biases, limiting generalizability to the broader population. - Positive emotion was measured at a broad level, potentially overlooking nuanced subcomponents (e.g., joy, gratitude, hope). - Potential confounders (e.g., information transparency, governmental trust) were not explicitly modeled. - Cross-platform and cross-country differences in user bases and platform norms may affect comparability despite fixed effects and controls. - Event timing and policy heterogeneity could influence estimates beyond modeled controls, though robustness checks with matched city samples supported consistency.