HUMAN PROCESSING OF 2-DIMENSIONAL GRAPHICS - INFORMATION-VOLUME CONCEPTS AND EFFECTS IN GRAPH-TASK FIT ANCHORING FRAMEWORKS

This report discusses the findings from three related experiments on the effects of information volume in graph-task fit anchoring frameworks reported in the literature. Information volume is operationally defined as the size of a data matrix (SDM), that is, the total number of points in a graphical display. The anchoring frameworks specify that an extraction task has high or low x-value anchoring depending on whether or not the x-component is represented in the question (as a given or unknown value). A total within-subject repeated measure experimental design was used to test the effects of SDM on speed and accuracy of data extraction. These experiments also integrated different frameworks to relate the information-volume effects. Results indicated that increased SDM adversely affected only data extraction time, not accuracy. A significant graph format by information volume interaction was observed; and training did reduce perceived information complexity, especially for high data volume displays. Also, effects of information volume on graph types differed: For vertical bars, a steeply rising monotonic performance-information volume trend was observed for all tasks. Symbols produced flat nonmonotonic trends for high x-value anchoring tasks and a gently rising monotonic trend for low x-value anchoring tasks. In contrast, line graphs produced a gently rising monotonic trend for high x-value anchoring tasks and a nonmonotonic trend for low x-value anchoring tasks. Such evidence suggests that information volume effects on human processing of two-dimensional graphical displays are influenced considerably by the character of the graphic format used for representing quantitative data and by the ''fit'' in anchoring characteristics between tasks and graphical formats.