Self-discontinuity, smoker and ex-smoker identity, and smoking relapse: A multilevel longitudinal study.

The present study examined longitudinal associations between smoking-related self-discontinuity (i.e. the perception that quitting smoking has disrupted the stability of one's self-concept over time), smoker and ex-smoker identity, and smoking relapse. We tested a mediation model in which increases in self-discontinuity were expected to increase relapse risk through concurrent and prospective increases in smoker identity and decreases in ex-smoker identity. The model was evaluated in a four-wave longitudinal study conducted over 1 year among former smokers who had quit within the previous 5 years. Multilevel analyses isolated within-person associations between self-discontinuity, identity, and relapse over time while controlling for time-invariant between-person differences related to smoking history. Results showed that within-person increases in self-discontinuity were indirectly associated with a higher likelihood of relapse at the subsequent wave through their concurrent association with smoker identity, but not ex-smoker identity. Moreover, no indirect effects emerged when changes in both smoker and ex-smoker identities were modeled prospectively. Sensitivity analyses provided no evidence for reverse indirect pathways from identity to relapse via self-discontinuity. However, within-person increases in both smoker and ex-smoker identities predicted subsequent increases in self-discontinuity, suggesting that identity fluctuations may precede feelings of self-discontinuity rather than the reverse.