A clustering pattern in the spatial distribution of some variable which seems to be due to the very fact that the occurrences are physically close together, that is, that they are in geographical proximity. Spatial autocorrelation is widespread: rich people move to areas where other rich people live; people only go to parties because other people go, and so on. In any exercise using sampling, the problem with spatial autocorrelation is that if observed values are spatially clustered, samples will not be independent; see T. Bailey and T. Gatrell (1995) and Ehlschlaeger (2000) Computers Env. & Urb. Sys. 24, 5 on measuring spatial autocorrelation. Getis (2008) Geog. Analysis 40, 3, 297 provides a history of spatial autocorrelation.