Investigating User-Generated Content in an Online Drug Recovery Forum: Lessons for Successful Computer-Mediated Communication of Social Support.

Benefiting from social support in online health communities requires maintaining textual communication. Investigating the process and identifying successful patterns can guide devising interventions to help online support seekers. We propose new methods to investigate the relationship between support-seeking requests and response messages in an online drug recovery forum. We use LIWC2015 text analysis software to quantify the support-seeking messages and apply machine learning algorithms to code the amount of informational and emotional support in the responses. Our work has several findings regarding the language in request messages that would increase or decrease the chances of receiving more informational or emotional support in response. For example, expressions of positive emotions and self-reference in request messages were associated with receiving more emotional support, and messages that used words indicating close relationships received more informational support. These findings contribute to the current understanding of computer-mediated communication of social support in online health communities, identifying strategies to mobilize maximal social resources. Moreover, our proposed methods can be used in other studies to investigate the exchange of social support or similar topics on online platforms.