Predicting graph reading performance

Performance and preference measures are commonly used in the assessment of visualization techniques. This is important and useful in understanding differences in effectiveness between different treatments. However, these measures do not answer how and why the differences are caused. And sometimes, performance measures alone may not be sensitive enough to detect differences. In this paper, we introduce a cognitive approach for visualization effectiveness and efficiency assessment. A model of user performance, mental effort and cognitive load (memory demand) is proposed and further mental effort and visualization efficiency measures are incorporated into our analysis. It is argued that 1) combining cognitive measures with traditional methods provides us new insights and practical guidance in visualization assessment. 2) analyzing human cognitive process not only helps to understand how viewers interact with visualizations, but also helps to predict user performance in initial stage. 3) keeping cognitive load induced by a visualization low allows more memory resources to be available for high level complex cognitive activities. A case study conducted supports our arguments.