Star plots: How shape characteristics influence classification tasks

Our research addresses the question of how to design interfaces for spatial analysis such that they support cognitive processes. In this paper we specifically target the question of map symbol design for the analysis of multivariate data, which is a common problem in cartography and related fields. We focus on star plots and the largely unaddressed question of how to assign variables to rays in a star plot and which consequences specific shapes have-as the result of data characteristics and the assignment of variables to rays-on interpretation and classification. We conducted an experiment with two conditions that were designed to shed light on the question: Does the shape of a star plot influence the interpretation (meaning) of the data it represents in a classification task? While previous research on multivariate point symbols has addressed this question for Chernoff faces, for example, few connections have been made to the shape of a star plot and its potential influence on meaning. We found that certain salient shape characteristics induced by variations along the horizontal and vertical axis increase the classification speed. However, we also found that salient shapes, such as has one spike, introduce a perceptual similarity that overrides the assumed similarities in the meaning of the represented data.