We constructed LMMs to examine the effects of Group and Session on CES-D, PANAS, and WSWS scores. We found no significant results when we compared postcessation depressive symptoms, as measured by CES-D, in controls, abstainers, never-quitters, and relapsers. However, we did find a significant Group × Session interaction effect on negative affect, as measured by the PANAS negative subscale, for controls, abstainers, never-quitters, and relapsers, F(3, 323) = 2.74, p < .05 (see Figure 3). The results indicated that adding a random slope effect to session did significantly improve the goodness-of-fit for a model that used group and session to predict self-report negative affect. Our results showed that abstainers, simple slope = 0.07; t(323) = 2.04, p < .05, and never-quitters, simple slope = 0.69; t(323) = 2.35, p < .05, reported a significant increase in negative affect across the precessation and postcessation periods compared with control participants who continued to smoke (simple slope = —1.04). No significant difference in negative affect changes was found among abstainers, never-quitters, and relapsers.
