Performance with tables and graphs: effects of training and a Visual Search Model

摘要

After more than 70 years of research it is still not clear under what conditions graphic presentations of information have an advantage over tables. A minimum assumption Visual Search Model (VSM) was designed to predict the performance of various tasks with tables and graphs that show data with different levels of complexity. An experiment tested the performance of five tasks with tables, bar- graphs and line-graphs, showing data with various levels of complexity, over the course of nine experimental sessions in order to assess possible changes in the relative efficiency of the displays after practice. Tables had an initial advantage over graphs for all tasks, and there were complex interactions between the variables. The initial differences between the displays disappeared for some tasks after users gained experience with the displays, while for other tasks the differences continued to exist even after extended practice. The VSM predicted the results for tables well. For graphs the model was adequate for tasks that involve single data points, such as reading values or comparing pairs of values. The performance of tasks that require the analysis of data configurations, such as reading a trend, could not be predicted with the VSM. Hence the VSM can predict task performance with tables and graphs for low-integration tasks.