ABSTRACT
The chapter on advanced global spatial autocorrelation considers some special cases of the Moran scatterplot and broadens the concept of spatial correlation to a bivariate setting. A differential Moran scatterplot is a shortcut to compute Moran's I for the change in a variable observed at two points in time. In GeoDa, instead of having to compute the difference first and then apply the Moran scatterplot, this can be done simultaneously. In all other respects, this is a standard Moran scatterplot.
The Moran scatterplot for EB rates extends the approach of Empirical Bayes rate smoothing to a special standardization of the rates that corrects for the inherent variance instability. Again, the resulting Moran scatterplot is standard.
Finally, the concept of bivariate spatial autocorrelation and its special case of space-time correlation is broached. The difficulty of separating in-situ correlation with the spatial form is highlighted, and different perspectives on relating observations at locations over time are illustrated.
