Spatial Distribution and Modelling of the Total Fertility Rate in Turkey Using Spatial Data Analysis Techniques

Abstract

The fertility rate has been declining for over four decades in Turkey. However, the fertility rate has shown regional variability due to ethno-cultural differences. While the fertility rate is low in the Western part of Turkey, the Eastern and Southeastern parts have still shown moderate to high rates. This study focuses on the spatial patterns of the total fertility rate. Using variables that may affect the fertility rate, such as economic and socio-cultural parameters, we performed spatial data analysis techniques to represent, analyze, and model the spatial data. The results show that according to Moran's scatter plot, Turkey's total fertility rate falls into two groups: high-high and low-low. On the other hand, local Moran's I results have shown that while the East and Southeastern regions have positive auto-correlations, Marmara, the Aegean, the West Black Sea, and the Middle Anatolia regions have negative auto-correlations. In this study, we applied both the ordinary least square (OLS) and geographically weighted regression (GWR) models and compared the results. In GWR analysis, variance of the dependent variable was shown to be 93%, and we achieved a high success rate in modeling Turkey's total fertility rate. In the limitation of this study, using an illiterate female population rate and Kurdish female population rate variables, one can obtain more accurate models that show the total fertility rate and show where the fertility rate is high. As a conclusion, spatial data analysis methods bring a new perspective to socio-demographic studies.