Implications of graphics enhancements for the visualization of scientific data: dimensional integrality, stereopsis,...

This study examines the effectiveness of four techniques to assist scientists in evaluating multidimensional data. Subjects viewed a series of complex 3D data sets, each representing an underlying complex surface, from which a set of discrete points or observations were sampled. From each sample they answered questions that required focus of attention on certain data points or integration across varying numbers of data points and dimensions. After a number of samples were viewed from each surface, subjects were tested for their retention of the surface characteristics. In Experiment 1, 3D (perspective) representations were found to support superior performance to 2D (planar) representations, but only for more integrative questions. Animated motion provided no benefits. In Experiment 2, stereoptic views of a 3D display were also found to support performance, particularly for integrative questions, but the ability to rotate the data space (motion parallax) and the presence of a mesh surface connecting the points did not. The posttests revealed some evidence that 3D representations improved the ability to visualize the surface, but neither 3D renderings nor stereopsis led to a better abstract representation of the data.