Bridging Disciplinary Boundaries (January 11 - 14, 2007)

This study examines whether and how the ethnic composition within schools influences the efficacy of a youth substance use prevention program. Data come from a randomized trial of a proven efficacious prevention program, the keepin' it REAL Model Program, which was administered to a predominantly Mexican American sample of 4,622 middle school students in Phoenix, Arizona, beginning in 1998. Students were randomly placed into one of four treatment categories – a Latino-based treatment version, a non-Latino version emphasizing mainstream cultural values, a multicultural version that combined elements from the first two, or a control group that received no treatment program. Given that substance use varies by ethnicity, we hypothesized that school ethnic composition could operate through the school-level normative environment to influence students' responsiveness to prevention programming and thus their substance use attitudes and behaviors. As in other parts of the nation, Latino students represented a large and growing presence in the study schools. Some research suggests that predominately Latino schools may create uniquely protective environments for Latino students and perhaps even for individuals from other groups. Yet, Latinos are diverse, particularly in terms of acculturation. Thus, we explore whether and how students' receptivity to the prevention messages of keepin' it REAL was affected by Latino predominance in the school, and the degree to which it reflected the presence of more recent immigrants and less acculturated students. The use of hierarchical linear modeling, paired with multiple imputation techniques, manages issues of clustered data and attrition. Two measures of school composition were tested in interaction with dummy variables for treatment versus control groups: the percent Latino among all students, and the percent that were non-English speaking at home. Pre-test to post-test self reports of last 30-day use of alcohol, cigarettes and marijuana were used to assess program impacts. An analysis of the entire sample revealed significant interactions between percent of students who do not speak English at home and treatment version, but no effects of the overall percentage of Latino students. These findings led us to conduct subgroup analyses by individual level of acculturation, which revealed even larger effects of school ethnic composition on program efficacy for selected groups. Among less linguistically acculturated Latino individuals, program effectiveness of the Latino version in reducing alcohol, cigarette, and marijuana use was enhanced in schools with higher percentages of students from non-English speaking homes. These effects were not found for more acculturated Latinos, nor for non-Hispanic Whites, nor did the effects occur with the other versions of the prevention program. For all the subgroups, the percentage of Latino students at school had no moderating influence on the effectiveness of any treatment version. Results are discussed in light of theories of immigrant adaptation, cultural matching of program content, and social processes in schools.