Do cognitive attributions for smoking predict subsequent smoking development?

To develop more effective anti-smoking programs, it is important to understand the factors that influence people to smoke. Guided by attribution theory, a longitudinal study was conducted to investigate how individuals' cognitive attributions for smoking were associated with subsequent smoking development and through which pathways.Middle and high school students in seven large cities in China (N = 12,382; 48.5% boys and 51.5% girls) completed two annual surveys. Associations between cognitive attributions for smoking and subsequent smoking initiation and progression were tested with multilevel analysis, taking into account plausible moderation effects of gender and baseline smoking status. Mediation effects of susceptibility to smoking were investigated using statistical mediation analysis (MacKinnon, 2008).Six out of eight tested themes of cognitive attributions were associated with subsequent smoking development. Curiosity (β = 0.11, p < 0.001) and autonomy (β = 0.08, p = 0.019) were associated with smoking initiation among baseline non-smokers. Coping (β = 0.07, p < 0.001) and social image (β = 0.10, p = <.0001) were associated with smoking progression among baseline lifetime smokers. Social image (β = 0.05, p = 0.043), engagement (β = 0.07, p = 0.003), and mental enhancement (β = 0.15, p < 0.001) were associated with smoking progression among baseline past 30-day smokers. More attributions were associated with smoking development among males than among females. Susceptibility to smoking partially mediated most of the associations, with the proportion of mediated effects ranging from 4.3% to 30.8%.This study identifies the roles that cognitive attributions for smoking play in subsequent smoking development. These attributions could be addressed in smoking prevention programs.

► We test associations between smoking attributions and subsequent smoking behaviors.
► 6 out of 8 smoking attributions were associated with subsequent smoking development.
► The associations varied between genders and at various stages of smoking.
► Most associations were partially mediated by susceptibility to smoking.
► These findings should be considered in smoking prevention programs.