no code implementations • EMNLP (CINLP) 2021 • Daniel Low, Kelly Zuromski, Daniel Kessler, Satrajit S. Ghosh, Matthew K. Nock, Walter Dempsey
We use propensity score stratification, a causal inference method for observational data, and estimate whether the amount of comments —as a measure of social support— increases or decreases the likelihood of posting again on SW. One hypothesis is that receiving more comments may decrease the likelihood of the user posting in SW in the future, either by reducing symptoms or because comments from untrained peers may be harmful.