JMIR Publications

ABSTRACT

Background:

Wuhan city in China was sealed off on 23 January 2020 in response to the COVID-19 pandemic. Extreme measures such as lockdowns were found to be associated with intense negative emotional responses that can be detrimental to psychological well-being. However, past studies about the emotional responses to the Wuhan lockdown have analyzed social media data using lexicons based on limited emotion types.

Objective:

To map Chinese people’s emotional responses to the Wuhan lockdown and compare Wuhan residents’ emotions with those of people elsewhere in China by analyzing social media data from Weibo using a lexicon modeled on the circumplex model of affect.

Methods:

Social media posts on Weibo from two weeks before to two weeks after the Wuhan lockdown (9 January to 6 February 2020) were collected. Each post was coded with a valence score and an arousal score. In Study 1, 359,190 Weibo posts were used to map the overall emotion trajectories. To control for between-subject differences, Study 2 (N = 11,802) analyzed posts from the same individuals with repeated ANOVAs. Study 3 compared emotions between Wuhan residents (N = 1,241) and non-Hubei residents (N = 12,851) using differences-in-differences regressions.

Results:

Majority of the posts made during the lockdown period were pleasant and of low arousal. Across the three studies, a gradual increase in both valence and arousal before the lockdown was observed. Study 2 found a temporarily boosted valence (F2.000, 23573.290=36.600, P<.001) and arousal (F1.990, 23443.010=359.292, P<.001) on the lockdown day. In Study 3, the Wuhan group showed similar valence (δ=-0.010, SE 0.030, P=.759) and arousal (δ=-0.036, SE 0.026, P=.210) changes after the lockdown as the non-Hubei group. Both valence (γ=0.062, SE 0.011, P<.001) and arousal (γ=0.071, SE 0.009, P<.001) were higher in the two weeks post-lockdown than the two weeks pre-lockdown, whether being in Wuhan or not. Wuhan residents reported lower valence (β=-0.101, SE 0.028, P<.001) and lower arousal (β=-0.072, SE 0.023, P=.004) than non-Hubei residents over the same period.

Conclusions:

Our results show that even during the pandemic, most Weibo posts were pleasant and of low arousal. Among the Chinese, there was both a temporary boost and a longer-term increase in valence and arousal from the lockdown. However, Wuhan residents’ emotions were not immediately affected by the lockdown: Their valence and arousal were systematically lower, but showed a similar escalation after the lockdown as did the non-Hubei residents. The overall increase in valence and arousal after the lockdown might suggest collective cohesion and mutual support in online communities during a public health crisis. Policymakers could consider how online social connections can provide mutual encouragement during a lockdown, thereby alleviating adverse impact of lockdowns on psychological well-being.