Geographically weighted regression models for ordinal categorical response variables: An application to geo-referenced life satisfaction data

Abstract

Ordinal categorical responses are commonly seen in geo-referenced survey data while spatial statistics tools for modelling such type of outcome are rather limited. The paper extends the local spatial modelling framework to accommodate ordinal categorical response variables by proposing a Geographically Weighted Ordinal Regression (GWOR) model. The GWOR model offers a suitable statistical tool to analyse spatial data with ordinal categorical responses, allowing for the exploration of spatially varying relationships. Based on a geo-referenced life satisfaction survey data in Beijing, China, the proposed model is employed to explore the socio-spatial variations of life satisfaction and how air pollution is associated with life satisfaction. We find a negative association between air pollution and life satisfaction, which is both statistically significant and spatially varying. The economic valuation of air pollution results show that residents of Beijing are willing to pay about 2.6% of their annual income for per unit air pollution abatement, on average.