This study explores a refined method for performing spatial autocorrelation analysis with spatial datasets processed for dasymetric mapping, or what is pertinently termed here as “dasymetrically disaggregated spatial data”. Conceptually, the refined method applies a spatial weight assignment method where neighborhood weights are assigned based on the area units’ spatial configuration, as well as the parent choroplethic zones from where the area units originate. This method is mathematically carried out by algebraically combining two spatial weight matrices: one coming from the conventional, spatial-neighborhood-based spatial weight matrix; and the other from a custom weight matrix based on similarity of choroplethic origins. Using Siquijor Island, an island province in the Philippines, as test area for conducting a case study, it was affirmatively shown that the revised spatial weight assignment method can yield substantially distinct spatial autocorrelation analysis results. In particular, two different effects were identified, depending on the type of input spatial variable used from the dasymetrically disaggregated spatial data. When the spatial variable used is in the form of raw counts, the refined method produces a filtering effect, resulting in lowered levels of spatial autocorrelation detected in the analysis output. On the other hand, when the spatial variable is in the form of densities, the method oppositely gives an amplifying effect, where intensified spatial autocorrelation levels are instead observed. These two opposite effects are consistently observed whether spatial autocorrelation analysis is measured at the global level, i.e. in terms of the global Moran’s I, or the Geary’s C index; or at the local level, i.e. in terms of the local Moran’s I, or the Getis-Ord Gi* index. Keywords: spatial analysis | spatial autocorrelation | dasymetric mapping | spatial weight matrix | cluster and outlier analysis | hotspot analysis | Moran’s | Geary’s | Getis-Ord | spatial statistics     «

This study explores a refined method for performing spatial autocorrelation analysis with spatial datasets processed for dasymetric mapping, or what is pertinently termed here as “dasymetrically disaggregated spatial data”. Conceptually, the refined method applies a spatial weight assignment method where neighborhood weights are assigned based on the area units’ spatial configuration, as well as the parent choroplethic zones from where the area units originate. This method is mathematically carr...     »