. 2022 Nov 14;6(11):e37698. doi: 10.2196/37698

Table 2.

Immediate emotional responses based on a multilevel model predicting valence or arousal with slope 2 and slope 3 at level 1 by location at level 2 (data set 2).

Variable		Fixed effect				Random effect
		Coefficient	SE	P value	SD		Chi-square (df)^a	P value
Valence
	Intercept (γ₀₀)	0.337	0.022	<.001	0.452		656.8 (3)	<.001
	Slope 1 (γ₁₀)^b	0.123	0.013	<.001	—^c		—	—
	Slope 2 (γ₂₀)^d	0.118	0.021	<.001	0.259		3.8 (3)	.28
	Slope 3 (γ₃₀)^e	−0.146	0.020	<.001	0.260		24.7 (3)	<.001
	Slope 4 (γ₃₀)^f	0.000	0.001	.99	—		—	—
	Location (γ₀₁)^g	−0.037	0.032	.24	—		—	—
	Slope 2×location (γ₂₁)^d,g	−0.172	0.052	.001	—		—	—
	Slope 3×location (γ₃₁)^e,g	0.109	0.047	.02	—		—	—
Arousal
	Intercept (γ₀₀)	2.365	0.019	<.001	0.265		183.3 (3)	<.001
	Slope 1 (γ₁₀)^b	0.089	0.011	<.001	—		—	—
	Slope 2 (γ₂₀)^d	0.293	0.022	<.001	0.569		333.2 (3)	<.001
	Slope 3 (γ₃₀)^e	−0.315	0.021	<.001	0.607		403.4 (3)	<.001
	Slope 4 (γ₄₀)^f	0.000	0.001	.30	—		—	—
	Location (γ₀₁)^g	−0.042	0.026	.11	—		—	—
	Slope 2×location (γ₂₁)^d,g	−0.262	0.053	<.001	—		—	—
	Slope 3×location (γ₃₁)^e,g	0.218	0.050	<.001	—		—	—

^aThe chi-square value is calculated as the difference in –2loglikelihood between a model with and a model without the random effect.

^bSlope 1: emotional trend before the lockdown (between January 20, 2020, and January 22, 2020).

^cThe cells were kept empty when the random effect of a variable was not included in the model.

^dSlope 2: difference in emotion between January 23, 2020 (the day of lockdown) and January 22, 2020.

^eSlope 3: difference in emotion between January 23, 2020 (the day of lockdown) and January 24, 2020.

^fSlope 4: longer-term emotional trend in the 2 weeks after the lockdown was imposed.

^gLocation: dummy coded as 1 denoting the Wuhan group and 0 denoting the non-Hubei group.