Multi-group segregation indices for measuring ordinal classes

Most current methods used in geographic segregation analysis can be regarded as measures developed to examine the spatial segregation of nominal groups that are typically defined by ethnicity or race. However, the increasing polarization and spatial sorting of urban communities, especially in developing countries, along lines defined by the accumulation of material wealth necessitate the development of new approaches to socio-spatial segregation analysis. In particular an approach is required that explicitly considers the assessment of groups based upon ordinal level classes such as, in a three group case, high, medium and low. This paper introduces a social difference coefficient that modifies three measures used for multi-group spatial segregation indices in previous research, namely the Local Getis index, the Proximity index and the Exposure index. The social difference coefficient accommodates different interactions between ordinal-level categories. An empirical study of residential segregation based on housing quality classes in the northern cone of metropolitan Lima, Peru is used to demonstrate the utility of the new measures.