D. Unwin (1996) specifically focuses on visualization as a necessary first step in all spatial data analysis, simply because the position of particular attribute values on a map induces associative processes in the analyst, drawing upon analogies, possible prior information, or memory (for instance of possible sources of data error). In geostatistical analysis, Haslett et al. (1991) and Cook et al. (1996, 1997) have introduced linked variogram cloud plots, displaying the values of the squared differences of the pair of head and tail observations 
in one window, and the specific tail to head line on a map in a second window. By moving a pointing device about the variogram cloud plot, the analyst is able to see where on the map display the chosen pairs are located. In general, linked plot technology for dynamic data visualization is becoming an important part of the modern statistical toolbox, perhaps exemplified by XLispStat (Tierney, 1990) and XGobi (Buja, Cook and Swayne, 1996), neither of which is specifically designed for spatial data, but where both have been successfully utilised (Cook et al., 1996, 1997, Brunsdon and Charlton, 1996). Further examples of visualization techniques for socio-economic data are given by A. Unwin (1996).