Table 5.

Augmented Dickey-Fuller test for monthly prevalence of emotions in user comments. A $\mathrm{p}$-value higher than 5% fails to reject the null hypothesis that the time series of the daily amount of user comments with each emotion is non-stationary, thus indicating that the time series probably have time-dependent structures.

Emotion	Test statistics	P-Value	Test result
Anger	−3.4	1.0%	Reject the null hypothesis
Disgust	−3.5	0.9%	Reject the null hypothesis
Fear	−4.7	0.0%	Reject the null hypothesis
Joy	−1.9	35.6%	Fails to reject the null hypothesis
Neutral	−3.3	1.7%	Reject the null hypothesis
Sadness	−3.6	0.7%	Reject the null hypothesis
Surprise	−2.9	4.6%	Reject the null hypothesis