Investigating the mechanisms of smoking behavior change with statistical mediation analysis

Steven Francis Babbin, University of Rhode Island

Abstract

Investigating and quantifying the mechanisms that underlie behavior change is essential to understanding what drives effective interventions. Cigarette smoking remains a critical concern for public health, and increasing basic knowledge of smoking behavior change can directly lead to improved interventions. This series of six studies represents a comprehensive evaluation of the mechanisms of smoking behavior change with statistical mediation analysis. All studies utilized combined data from five tailored interventions based on the Transtheoretical Model (TTM) for participants in Precontemplation (PC; N = 1145), Contemplation (C; N = 1243), and Preparation (PR; N = 499). Statistical mediation models under investigation were autoregressive, three-wave models (baseline, 12 months, and 24 months) developed within each stage of change. The ten Processes of Change for Smoking were used as independent variables, Pros of Smoking, Cons of Smoking, and Situational Temptations to Smoke were used as mediators, and a behavioral smoking outcome was used as the dependent variable. ^ Studies 1, 2, and 3 investigated single mediator models at PC, C, and PR, respectively. Across the three stages, a total of 25 single mediator models, each with different combinations of variables, demonstrated evidence of statistical mediation. Studies 4 and 5 refined, consolidated, and extended the conclusions from these single mediator models. Study 4 found evidence of statistical mediation in multiple mediator models, and study 5 found evidence of statistical mediation in models with multiple Processes of Change for Smoking, resulting in a total of 20 final models. In study 6, the final models were tested for the presence of statistical moderation. Factorial invariance techniques were utilized to evaluate differences across subgroups for age, education level, gender, race, and original study. The statistical mediation models demonstrated equivalence across subgroups, and this suggests that the models describe mediating mechanisms that are robust across demographic and study-related variables. ^ The 20 final models, as developed in studies 1 through 5 and further validated by study 6, highlight combinations of Processes of Change and mediators that are most related to smoking outcomes. Pros, Cons, and Situational Temptations were all found to mediate smoking behavior, with different combinations of processes, for individuals in both PC and C. The most important Processes of Change for individuals in PC included Consciousness Raising, Dramatic Relief, Environmental Reevaluation, Self-Reevaluation, and Social Liberation. The most important Processes of Change for individuals in C included Counter Conditioning, Consciousness Raising, Dramatic Relief, Environmental Reevaluation, Self-Reevaluation, Social Liberation, and Stimulus Control. Only one combination was found to demonstrate statistical mediation for individuals in PR; Self-Reevaluation was found to mediate smoking behavior through Situational Temptations. ^ Based on the results from the series of statistical mediation analyses, these strategies for smoking behavior change should be emphasized in smoking cessation interventions. Modern interventions can be developed to maximize relevance of intervention contacts and improve effectiveness by tailoring to focus on key behavioral mechanisms. Future interventions can be further refined through new series of statistical mediation analyses.^

Subject Area

Health Sciences, Public Health|Psychology, Psychometrics

Recommended Citation

Steven Francis Babbin, "Investigating the mechanisms of smoking behavior change with statistical mediation analysis" (2014). Dissertations and Master's Theses (Campus Access). Paper AAI3619426.
http://digitalcommons.uri.edu/dissertations/AAI3619426

Share

COinS