An investigation of information display variables utilizing computer-generated graphics for decision support systems

The effectiveness of selected computer-generated graphics display variables was examined in a mixed-factors factorial experiment using thirty-two subjects. All subjects performed four different graph reading tasks consisting of point-reading, point-comparison, trendreading, and trend-comparison. In each task, line, point, bar, and three-dimensional bar graphs were investigated under two levels of task complexity, and two levels of coding (color and black-and-white). The effects of these independent variables on measures of task performance errors, time to complete the task, subjective mental workload, and preference ratings were obtained in real-time by a microcomputer control program. Separate MANOVA analyses of these measures for each task indicated significant effects of graph-type for the point—reading task, main effects of complexity and coding for all tasks, and a graph-by—coding interaction for the point-reading, point-comparison, and trend-reading tasks. Subsequent ANOVA analyses showed significance for these effects across several of the dependent measures which are specified in the thesis. Recommendations are made for selecting the most effective graph and coding combinations for the particular types of graph-interpretation tasks and complexity levels encountered.