Time-dependent association between the discussions of mass protests and mental health on social media during the 2019 anti-government social unrest in Hong Kong: A text mining study

PurposeSocial media are increasingly pivotal as the platform where activists and observers plan, promote, and respond to collective actions. To examine how mental health discourse might be impacted by mass protests, this study analyzed their time-dependent association during the 2019 anti-government social unrest in Hong Kong.MethodsConsecutive day-by-day user-generated content on online forums and social network sites (SNS) from June to November 2019 was obtained. A Cantonese term-list was created to identify terms related to mass protests and mental health discourse. The frequency of comments containing such terms was analyzed using time series models.ResultsThere were 3,572,665 social media comments in the investigation period. As hypothesized, the frequency of comments with mass protest terms was higher on days with mass protests than on days without. Frequency of comments with both mass protest- and mental health-terms was also higher on days with protests than days without. A time-lagged effect of protest-terms was found on online forums but not on SNS. Our results suggest a positive association between offline protest activities and online psychological reactions.ConclusionsSocial media content reveals discussion of mental health concerns stemming from, or exacerbated by, social unrest. The potential mutual influences between mass protests and online reactions, as well as the functional differences between online forums and SNS in this regard, are discussed. Street protests and their associated mental health discourse can be readily detected on popular online forums. Mental health services should consider such dynamic relationships between on- and offline activities.