This paper examines the impact of reproductive externalities on fertility behaviour in Kenya by quantifying the effects of group membership on the number of children born. We focus on the identification of structural forms of social interaction operating across individuals in the context of fertility behaviour. While structural forms of dependence may be separated from residual dependence, we also highlight the importance of difference expressions of structural dependence, including multiple expressions of social interaction. Using this idea of multiple social interactions, we use the 1998 Demographic and Health Survey on 5994 women from Kenya to examine whether the ‘local’ effect of household-level influences and cluster-level residential settlement is important relative to the more ‘global’ effect of ethnicity on fertility behaviour. In so doing, we conclude that the importance of multiple social interactions is that the assumption of a single model of interaction can lead to incorrect inferences.