Spectrally Encoded Two-Variable Maps∗

ABSTRACT The U.S. Census Bureau began publishing spectrally encoded two-variable choropleth maps in the early 1970s. Four experiments with human subjects have been conducted with these controversial maps to examine their efficacy. It has been shown that 1) subjects cannot spontaneously order the colors into the legend arrangement used on these maps, but they can recognize order in that arrangement; 2) the task of judging correlation between the geographical patterns of individual variables is initially more difficult than the task of judging correlation on separate maps, although subjects who do better than guessing on this task perform at least as well as they do with separate maps; 3) subjects do show evidence of having gained information when they are asked to write what they learn from these maps; and 4) subjects find the maps interesting and appealing, particularly if explanatory verbal notes accompany them. Notes ∗The work reported here was supported by the U.S. Bureau of the Census. Acknowledgment is also due to Russell Chisholm and Michael P. Peterson who served as project assistants, the many colleagues who shared their thoughts on both the original proposal and various drafts of this paper, and to the 466 subjects from various colleges and universities in the northeast who participated in the experiments.