Skip to main content
. 2019 May 29;10:1186. doi: 10.3389/fpsyg.2019.01186

Figure 1.

Figure 1

R code of the five multi-level models (M0, ME1, ME2, MCP1, and MCP2) using package lme4. In all models, we estimated the post-therapeutic measurement of CES-D (cesd) based on a random intercept for each client (id). In ME1 we estimated the post-therapeutic effect of the number of positive emotion words as the interaction effect between the number of positive emotion words (posemo) and an indicator variable (post). The other models are similar, in ME2 we estimated the effect of the number of negative emotion words (negemo), in MCP1 we estimated the effect of the number of insight words, and in the MCP2 we estimated the effect of the cause words. M0 is nested under each of these models.