A Componential Model of Human Interaction with Graphs. IV: Holistic and Analytical Perception of Star Graphs

Abstract

Multivariate graphs, such as star graphs, are increasingly popular displays for complex data sets. One advantage claimed for these graphs is that users can engage perceive the data holistically. This experiment used star graphs to examine whether subjects' performance improved when they had an object cue as part of the stars in a task that involved selecting an outlier from a display of four stars. The outlier star differed only on one of five legs with the position of the differing leg varying systematically. Individual subject's response times (RT's) showed two patterns: sensitivity to the position of the different leg (suggesting an analytical strategy) and insensitivity to position (suggesting a holistic strategy). The object cue improved performance for the “holistic perceivers”, but not for the “analytical perceivers”. The research suggests that an object cue can help some, but not all users of a multivariate graph and that assessing users' perceptual processing strategy can be important in evaluating the effectiveness of display design features.