Identified or conflicted: A latent class and regression tree analysis explaining how identity constructs cluster within smokers

Identity, or ‘who I am’, is important for smoking behaviour. Identity constructs (parts of a person’s identity) are typically examined as separate entities, but emerging evidence suggests that the multifaceted nature of identity is relevant in the context of smoking. This cross-sectional study examined how smoking-related self- and group-identity constructs cluster within adult daily smokers (N = 231), whether classes of smokers can be distinguished based on clusters of identity constructs, and which factors explain class membership. Data were collected online in The Netherlands and Belgium, 2017–2018. Latent class and regression tree analyses showed that participants in Class 1 of ‘Identified smokers’ (estimated population share 54%) reported stronger smoker self- and group-identities, stronger expected identity loss when quitting smoking, and weaker quitter self-identities and non-smoker self- and group-identities (vs. Class 2 of ‘Conflicted smokers’). Class membership was explained by the interaction between mental smoking dependence (dominant explanatory variable), consideration of future consequences, age of smoking onset, self-efficacy, and future self thought clarity. Models had good fit. The identity of more dependent smokers is more strongly oriented toward smoking. Smoking is also more strongly embedded in the identity of smokers who started smoking young, are less inclined to think about the future, and have lower self-efficacy.