While macro-level and repeated cross-sectional studies of tobacco control policies and smoking rates, as well as evaluative studies of new policies in specific geographic areas, provide some evidence of policy effectiveness, little research has examined how policies at various levels implementation (e.g. local, county, and state) affect within-individual trajectories (initiation, prevalence, duration, and cessation) of smoking during the critical teenage and young adult years. Without an approach that examines the full range of policies simultaneously nationwide on the same individuals over time, policymakers cannot adequately assess the efficacy of particular tobacco control policies and, thus, risk investment in non-effective regulations. Using nationally representative longitudinal data, a complete record of tobacco control ordinances, and innovative statistical methods, the goal of this project is to provide policymakers and health researchers with the tools to make the most informed tobacco control policy decisions regarding youth smoking prevention and intervention, including which individuals or locales are least susceptible to changes in behavior resulting from policy implementation. To fully understand how tobacco control policy affects youth smoking patterns over time, we require a geocoded, repeated observations dataset of youth cigarette use and predictive risk factors, a comprehensive database of tobacco control regulations, and a statistical method capable of handling a complex hierarchical data structure. We address those needs by merging data from the annually collected National Longitudinal Survey of Youth 1997 (NLSY97) with the Americans for Nonsmokers'Rights (ANR) national repository of tobacco ordinances and regulations, and using recent developments in hierarchical cross-classification models. To identify the main effect of tobacco policies on trajectories of youth cigarette use, we will combine the two geocoded datasets and analyze them with recently developed hierarchical cross- classification models. After controlling for well-known risk factors and local structural conditions, the efficacy of particular tobacco policies will demonstrate which policies should be adopted on a wider basis, ultimately impacting health and mortality and reducing costs to health care and society. To assess whether the effect of policy varies by characteristics of the individual or context, we will analyze interaction effects between individua-level characteristics and policy measures and local-level characteristics and policy measures, which will reveal if a given policy is more effective for certain types of individuals or those residing in similar areas. The lack of effectiveness of policies on certain individuals will expose the need for those policies to incorporate a component that addresses particular groups or areas. This proposal's innovation lies in both its examination of within-youth changes in smoking behavior over time as it relates to both previously examined and unexamined tobacco restrictions at various levels of implementation and in the development and furthering of innovative statistical approaches necessary for datasets that contain repeated observations on multiple levels.
This study will examine the effect of various forms of tobacco control policy on youth trajectories of smoking over time and variations in the effect of policy by characteristics of the individual and their local context. Specifically, through innovative hierarchical cross-classified models using a repeated observations dataset and a complete collection of local tobacco control policies, we will examine the ways in which changes in policies influence youth initiation, duration, prevalence, and cessation of cigarette smoking. By influencing the proliferation of the most successful tobacco control policies, the findings from this study will influence mortality rates and costs to public health as fewer youth become adult smokers.