Unlike startle responses, participants’ answers to self-report questionnaires (CES-D, PANAS, and WSWS) were nested within a two-level structure of longitudinal data (i.e., sessions nested within participants). Thus, two-level LMMs were conducted to evaluate the effects of session (level 1 predictor) and group (level 2 predictor) on these responses. Again, we used a log likelihood ratio test to assess model fit between one where session was entered as a fixed effect and another where the relationship between session and self-report measure responses was estimated to vary among participants.
