Adding Structural Variables to Space-Time Interaction Tests: An Application to Fertility Transition in Brazil

Carl P. Schmertmann, Florida State University
Renato M. Assuncao, Universidade Federal de Minas Gerais
Joseph E. Potter, University of Texas at Austin

A long debate in the fertility transition literature concerns the relative strengths of ideational diffusion and economic development. Spatial statistics have always been implicitly important in this debate, but they have recently moved to the foreground. Bocquet-Appel and Jakobi (1998) suggested using Knox’s space-time interaction test for a study of demographic transition in Great Britain. The Knox test is widely used in epidemiology, but it has a serious drawback for demographic applications. Specifically, it does not account for structural variables that also change in space and time. Consequently, the Knox test is too likely to find evidence for ‘epidemic’ patterns of fertility diffusion when local economic development levels are changing. We propose an improved space-time interaction test and illustrate with data from four Brazilian censuses. We also show how this method can be adapted to produce maps showing where and when diffusion effects seem strongest, net of covariate effects.