Chapter 14 Advanced Global Spatial Autocorrelation

In this chapter, I continue with the treatment of global spatial autocorrelation. First, I consider some special applications of the Moran scatter plot principle to time differenced variables and to proportions. These cases provide some short cuts to the computations needed to construct the graph, but otherwise are identical to the standard plot covered in Chapter 13.

Next, I cover the topic of bivariate spatial correlation, a common source of confusion in practice. After presenting the principle behind a bivariate Moran scatter plot, attention focuses on the special bivariate situation where the observation on a variable at one point in time is related to its neighbors in a different time period, i.e., space-time correlation. A critical review is provided of different perspectives and their relative merits and drawbacks.

To illustrate these methods, I go back to the Oaxaca Development sample data set.