Mixed-effects model results.

In (a), the models study the effect of session length on the quality of the first comment C1 in a session; i.e., data only contains the first session comments. In (b), the models investigate the effect of the session index i on the quality of respective comment Ci; data includes all comments in sessions with more than a single comment. Each table highlights the most appropriate models for each quality features based on extensive model analytics—lmer refers to linear mixed-effects models while glmer refers to generalized linear mixed-effects models. All coefficients are strongly significant as derived from model comparisons based on BIC statistics.