Application of spatial autocorrelation in the analysis of fertility distribution and fertility patterns in Croatia

The paper addresses the measurement of variability of the total fertility rate (TFR) at the NUTS3 and LAU levels for the year 2021. By applying spatial autocorrelation methods, the Global and Local Moran’s Index, the hypothesis of TFR value clustering at the regional and local level was tested. In other words, the study answers the question of whether TFR values are spatially grouped or heterogeneously distributed at the selected levels. The research procedure determined the degree of spatial autocorrelation at both observed levels and defined statistically significant clusters. Based on the Getis-Ord Gi* hotspot analysis, spatial patterns of TFR distribution at the LAU level were identified, and in addition to the analysis of z-values, the level of statistical significance of TFR hotspot and coldspot areas was also analyzed. The results of the research provide a deeper insight into the understanding of differences in TFR values between regions and within regions, which are often not detectable using classical statistical-analytical approaches.