Modelling change in adolescent smoking behaviour: stability of predictors across analytic models

Area of Science:

• Adolescent health behaviors
• Behavioral science
• Psychological modeling

Background:

• Adolescent smoking is a significant public health concern.
• Understanding the dynamics of smoking initiation and cessation is crucial.
• Existing models may not fully capture the complexity of behavioral change.

Purpose of the Study:

• To examine the variability of predictors for adolescent smoking changes.
• To compare linear and nonlinear analytical models in predicting smoking behavior.
• To assess the explanatory power of different modeling approaches.

Main Methods:

• Three models were tested: standard linear, pre-post linear, and nonlinear (cusp catastrophe).
• Longitudinal data from 779 Australian adolescents over 12 months were analyzed.
• Variables included demographics, smoking status, social influences, self-esteem, neuroticism, coping, stress, and risk-taking.

Main Results:

• Predictors of smoking change were consistent across linear and nonlinear models.
• Model explanatory power varied significantly.
• Models incorporating past behavior (initial conditions) explained over four times more variance than those without.

Conclusions:

• Confidence in predictors of adolescent smoking change is justified across models.
• Past behavior is a critical factor in dynamical models of adolescent smoking.
• Further research on the role of past behavior in health behavior dynamics is recommended.