Outcome evaluation of Arizona’s school-based smoking prevention program – a multilevel study

Objective. The purpose of this study was to conduct a quantitative outcome evaluation of Arizona’s school-based tobacco use prevention efforts and identify school-level factors indicative of a high smoking prevalence. School-based prevention services reach a sizeable share of Arizona’s in-school youth every year and cost about $4,000,000 annually. Method. Data were obtained from two sources, the Arizona Youth Survey for outcome data and process implementation records for intervention data. Because intervention data were only available on the school level, a series of hierarchical linear models was constructed to link outcome data to intervention data, based on Wittmann’s general evaluation framework of the five-data-box conceptualization. This study used a quasi-experimental design to evaluate implementation efficacy. Missing data were addressed with a number of methods to minimize bias. Results. Parental smoking, parental approval of smoking, age and American Indian ethnicity were individual-level risk factors for self-reported smoking and favorable attitudes towards smoking. A school-level intervention index showed a small relationship to school-level means in ecological and HLM models on reported number of friends who smoked and perceived normativity of smoking, but nil effects on self-reported behavior. School-level academic achievement was a strong predictor for self-reported smoking, attitudes and number of friends who smoked. School-level poverty was a redundant predictor once academic achievement was taken into account. Conclusions. Smoking prevention programming showed very weak effects on outcomes of eighth grade students. If programming is supposed to reach schools with the highest smoking issue, program administrators should target schools with low ratings on academic achievement. More rigorous study designs could shed a light on exact magnitude of effects