Autocorrelation and Visual Map Complexity

ABSTRACT One important element in the overall look of a quantitative map is the relationship between neighboring values as they appear on the map, i.e., the autocorrelation in the map based on the classes into which the values fall. Methods of calculating this relationship objectively include rank autocorrelation; proportions of identical neighbors; average differences between neighboring values; number of clusters; and a weighted “proportion'’value which includes a contribution from every neighboring pair, the amount of contribution varying inversely with the differences between the paired values. The first lag relationships appear to be most meaningful in terms of the visual appearance of hypothetical blocks of data (“maps”). Some of these measures are related to the way in which subjects order the blocks according to “spatial complexity,'’although checkerboard patterns in particular introduce differences between measured and subjective sequences.