Researchers have identified significantly greater levels of high-risk behaviors among sexual minority youth (SMY). Most studies of SMY's risk behavior's have looked exclusively at individual characteristics to explain high-risk behavior.

This study explored whether school context can explain some portion of the disproportionate heavy alcohol consumption, alcohol related problems, physical fighting and weapon carrying among same-sex attracted (SSA) youth. Bronfenbrenner's Human Ecology Model (1979) was selected as the theoretical framework for its ability to simultaneously consider and measure individual interactions with settings.

This dissertation used data from Wave One of the National Longitudinal Study of Adolescent Health (Add Health) using both the in-home student sample of 18,047 adolescents and 125 school administrators. Same-sex romantic attraction was measured as having both-sex or exclusively same-sex romantic attraction. A two-stage analysis was conducted. The first stage established the person-level factors that differentiated same-sex attracted youth from other youth using STATA 9. The second stage set of analyses was used both student and school level factors in a multi-level analysis (HLM 6) to assess school's contribution to adolescent risk behavior.

The first stage demonstrated significant student differences between same-sex attracted youth and other youth across all four of the dependent variables. In the second stage, the multi-level models identified school characteristics that amplify and moderate risk for all students including: school dropout rates, urban, percentage of students not being treated fairly by teachers, percentage of "Born Again" students, GPA, gender and age, among others. Same-sex attracted youth were no longer at higher risk in any of the four multi-level models.

Inclusion of school factors in a multi-level model eliminates the statistical significance of same-sex romantic attraction, suggesting that school characteristics may offer protection for SSA youth. The findings also suggest that Jessor's Problem Behavior Syndrome may be insufficient to explain adolescent risk behavior among the population.

This study suggests that context measures are necessary to accurately measures SMY's risk behaviors. This study also suggests the need to develop more specific and sensitive measures for school context. This research has implications for studying risk behavior among other vulnerable adolescent populations.