8. Long-Term Effects of Gay Straight Alliances On Perceived School Safety Among LGB Youth In Western Canada, Using A Novel Analytical Method To Enhance Causal Inference For Population Health Interventions

Abstract

Gay Straight Alliances (also known as Gender and Sexuality Alliances, or GSAs) are school-based social clubs designed to support LGBTQ youth inclusion and wellbeing. While intervention studies of participants show promising short term results, studies of their wider effects on school climate, including for youth who do not participate, are limited to cross-sectional correlation studies; randomized trials to infer causality are generally infeasible, because it is difficult to randomize interventions across large populations to start at the same time, and to track changes over equal years. To address these limitations, we aimed to identify potential long-term effects of GSAs on LGB students’ perceived safety in Western Canadian schools, using a new method to evaluate site-level longitudinal effects of non-randomizable population health interventions. Data were from 1,625 lesbian, gay, and bisexual (LGB) students in 135 public schools in the province-wide cross-sectional BC Adolescent Health Surveys of 2003, 2008, and 2013 (grades 7-12; 67.4% girls; 77% bisexual; mean age 15.7 years). Individual level variables included a perceived safety scale (6 items, about washrooms, classrooms, hallways, etc., Cronbach as>.90) and age as a covariate. School level information about GSA presence and length in existence was gathered in 2008 and 2014. We used the Site-Level Longitudinal Effects of Population Health Interventions (SLEPHI) novel design and analytical approach, involving nested multi-group, multi-level modeling of repeated cross-sectional student data (Level 1) within schools (Level 2), across survey years. Length of time since GSA started was calculated for each survey; schools with no GSA at any time point were coded 0. Repeated MG-ML models constrained varying parameters to account for secular trend, cohort effects, measurement error, and measurement equivalence, to address requirements to support causal inference. In 2003, only 9 schools (7%) had a GSA; by 2013, 64 (52%) did so. Length of GSA ranged from 0-14 years. Controlling for age, the MG-ML models found GSA length was linearly related to increased school-level perceived safety among LGB students (b=0.17, SE=0.07, p=.021), i.e., the longer a school had a GSA, the greater the perceived safety among its LGB students. This relationship remained consistent across all 3 survey years, and for the longest running GSA (14 years). Even when GSAs started in different years, with potentially different societal influences affecting the survey cohorts, we found LGB students in schools with longer established GSAs reported higher levels of perceived safety throughout various parts of the school, even after 14 years, suggesting a robust protective effect of GSAs on overall school climate, not just for individual participants. Results affirm both cross-sectional school-based survey findings and shorter-term longitudinal studies of GSA participants; health professionals should encourage schools to support GSAs and work to maintain them. As well, this new SLEPHI method evaluates population health intervention outcomes over longer time, with larger samples and lower costs, than typical non-randomized longitudinal studies of individuals over time, a distinct benefit for school-based interventions, where students attend only 4-5 years.