<i>Human Factors and Behavioral Science</i>: Experiments on Quantitative Judgments of Graphs and Maps

Recommendations are made to use pastel rather than highly saturated colors on statistical maps; standardize the point cloud size relative to the frame on a scatterplot; scale circles by making the circle area proportional to the variable represented, but expect widely varying judgments of the areas.

Abstract

Behavioral studies are essential for devising guidelines for effective communication of quantitative information from graphs. Three experiments in which subjects made quantitative judgments from three different kinds of graphs lead to several recommendations: use pastel rather than highly saturated colors on statistical maps; standardize the point cloud size relative to the frame on a scatterplot; scale circles by making the circle area proportional to the variable represented, but expect widely varying judgments of the areas.