Multilevel Modeling as Common Ground for Sociological and Geographical Electoral Studies

Sociological and geographical electoral studies differ, among other things, by the methods they use and the kinds of variables they are interested in, the former focusing on ''sociological'' variables whereas the latter adress ''spatial'' effects. In this paper, we suggest that these oppositions have no deep basis, and we promote multilevel modeling as a possible common ground for elections sociologists and geographers. This allows to make sense of the notion of a ''context'', where the independant variables can act in different manners across space.