Modelling Time-Varying Contextual Effects in Family and Fertility Research

Hill Kulu, University of Liverpool

There is increased interest in examining the impact of contextual characteristics on partnership and fertility behaviour of individuals. The conventional multilevel approach provides guidance on how to measure time-constant contextual effects; the proper way of modelling time-varying contextual effects, however, remains unclear. This paper proposes a framework, that allows us to distinguish between time-constant and time-varying contextual effects in multilevel event history analysis and to properly control for clustering in multilevel models with time-varying contextual factors. We use retrospective survey data from Austria to show how regional time-constant and time-varying characteristics influence fertility behaviour of individuals who live in these regions. The framework can be applied to also examine how country-level factors influence demographic behaviour of individuals.