In this paper, we introduce spatially weighted context data as a new approach for studying the contextual dimension of factors that shape social behaviour and collective worldviews. First, we briefly discuss the current contribution of multilevel regression to the study of contextual effects. We subsequently provide a formal definition of spatially weighted context data, as a complement to and extension of the existing multilevel analyses, which allows studying contextual influences that decrease with increasing distance, rather than contextual influences that are bound within discrete contexts. To show how spatially weighted context data can be generated and used in practice, we present a research application about the impact of the collective experiences of war across the former Yugoslavia. Using geographically stratified survey data from the TRACES project, we illustrate how empirical conclusions about the collective impact of war events vary as a function of the scale at which context effects are being modelled. Furthermore, we show how observed geographic patterns can be explained by underlying patterns of social proximity between the concerned populations, and propose a procedure to estimate the part of spatial dependency explained by models applying specific definitions of social proximity. In the final section, we discuss the boundary conditions for the use of spatially weighted context data and summarise the contribution of the proposed approach to existing methods for the study of context effects in the social sciences.