A comparison of random coefficient modelling and geographically weighted regression for spatially non-stationary regression problems

The problem of locally varying coefficients in geographical applications is considered. Two approaches to this are then discussed-geographically weighted regression and random coefficient models. The latter is considered in two forms: firstly, the case that only the intercept coefficient is random; and then the case in which all coefficients are random. All these techniques are applied to a data set derived from the 1991 UK Census of Population relating to limiting long-term illness.