Spatial Patterns as Predictors of Fertility: Change in Rural Egypt

John R. Weeks, San Diego State University
Arthur Getis, San Diego State University
Xiaoling Yang, San Diego State University
Tarek Rashed, San Diego State University
Douglas A. Stow, San Diego State University
Saad Gadalla, San Diego State University
Allan G. Hill, Harvard University

We use data from the 1986 and 1996 censuses of Egypt to show that spatial location is an important predictor of fertility levels in the predominantly rural governorate of Menoufia, Egypt. In an ordinary regression equation, it appears that levels of female education and the percentage of women married in a shiakha (equivalent to a village) are the principal correlates of fertility in each of those two years. However, that model is misspecified because it ignores the substantial spatial autocorrelation that exists in the data. Using both the Getis/Ord spatial filtering approach and the Fotheringham et al. Geographically Weighted Regression approach, we show that a large portion of the variability in fertility in Menoufia is accounted for by where people live, and a smaller fraction is accounted for by sociodemographic characteristics independent of location.