Effectiveness Study of the Intervention "Stop Sexual Harassment" in Secondary School

AIMS AND HYPOTHESIS The first aim of this study is to test the effects of "Stop sexual harassment", a sexual- and gendered harassment prevention intervention in secondary school among 8th to 10th grade pupils. The second aim of the study is to test the effects of the intervention on teachers' responses to sexual- and gendered harassment among the pupils. "Stop sexual harassment" is targeted at both pupils and teachers with the intention of reducing sexual harassment and harassment based on sexual orientation and gender expression (gendered harassment) among the pupils. It is hypothesized that the intervention will lead to reduced sexual- and gendered harassment perpetration and victimization and less internalized and externalized problems among the pupils. It is also hypothesized that the intervention will lead to teachers more frequently taking action when sexual- and gendered harassment is encountered among pupils.

DESIGN AND METHOD Recruitment: All secondary schools (8th to 10th grade) in three counties (Vestland, Nordland and former county Buskerud) in Norway will be invited to take part in the intervention.

Design: A cluster randomized controlled trial will be used to test the effects of the intervention. Participating schools will be randomly assigned to intervention schools and control schools.

Assessment: Questionnaires will be administered to pupils and teachers, prior to the intervention to obtain a baseline level (T1), shortly after the intervention (T2), and about six months after the intervention (T3). Previously tested instruments to assess to what extent the intervention leads to reduction in sexual harassment and harassment based on sexual orientation and gender expression among the pupils will be utilized. Teachers' responses to the pupils' sexual and gendered harassment will also be assessed.

Data analytic plan and power analysis: The data will be analyzed with both simple descriptive (such as correlations, frequency tables) and advanced statistical methods. With regard to the latter, 3 level multilevel modelling will be used to assess the impact of the intervention on the outcome variables. Multilevel modelling takes into account the hierarchical design of the data due to cluster randomization by school (level 3) and school class (level 2). Cluster effects were approximated (Intra Class Correlation; ICC) influencing the main outcomes (Sexual harassment, emotional problems, attitudes towards homosexuals) by taking advantage of data from previous studies. These studies indicate that whereas relatively little (often less than 5%) of the total variance of sexual harassment victimization, bullying and emotional problems are at the school and/or school class level, a still limited amount of studies suggests that this might not be the case for instance attitudes towards homosexuals where as much as 12,5 - 30% of the total variance have been explained by school/class levels. Power calculations, by the use the Optimal Design Software, showed that with a .05 level of significance, power = .80, ICC level 2 = .25, ICC level 3=0.05, number of cluster level 2=6 (classes) with a cluster size=20 (pupils), 38 schools are needed to be able to detect effect sizes of .30 (about medium effects according to Cohen (1988). The power of the analyses can be increased by including the pre-intervention measure of the outcome variable. Assuming the average school level of the outcome variable at pre-intervention explains 50 or 30 percent of the variation of the outcome variable at post-test, decreases the number of schools to 30 and 32, respectively, given the same assumptions as above. It was therefore decided to include at least 32 schools (16 intervention schools) to guard against the possibility of type II errors. A minimum of 3840 pupils will thus take part in the study.