Contextual research on adolescent offending starts primarily from the idea that residential areas provide the key ecological setting that (indirectly) shapes observed differences in adolescent offending. The known observed geographical concentrations of adolescent offenders in economically disadvantaged areas and the influence of the social disorganization/collective efficacy perspective are
responsible for scholars’ focus on the structural context of areas of residence. On the other hand contextual research has also been conducted from the perspective of the school context. The school with its organization, structure and social climate is also a meaningful setting that shapes attitudes and behaviours of adolescents. This separate contextual approach is discussed in this contribution because
it does not sufficiently take into account the contextual diversity that characterizes contemporary reality. Adolescents do not spend their leisure time exclusively at school that often is situated in another area than the residential area in which they live. They often spend leisure time on many other places then their residential area. This study contributes to the empirical research literature for three reasons. First, simultaneously and not analytically separate the impact of neighbourhoodand school context on adolescent offending is assessed. Second, this study starts from a theoretical framework, allowing us to control for the effects of confounding variables. Third, this contribution makes use of non-hierarchical multilevel modeling, that is a statistically correct method for answering research questions involving multiple contexts. From the framework of the situational action theory, with its strong emphasis on the interaction between social mechanisms and the micro-situational context, block wise regression analyses give support for the existence of small but substantial contextual effects of school level disadvantage, whereas the effect of neighbourhood level disadvantage is completely due to neighbourhood composition. Key words: contextual effects, non-hierarchical multilevel modeling, adolescent offending, cross-classified design