The Display of Multivariate Information: An Experimental Study of an Information Integration Task

Abstract

Process control systems typically involve many variables that (a) can be intercorrelated with each other (cross-correlated), (b) can be correlated with themselves over time (auto-correlated), and (c) that are represented by displays possessing varying degrees of reliability (i.e., displays that must be weighted by their importance or trustworthiness). We examined these variables in an information integration task, which also compared the relative advantages of integral and separable displays (pentagons and staggered bargraphs, respectively). We intended to investigate how the predicted advantage of the integral display would be modulated by the information's correlational structure and reliability. The results are interpreted within the framework proposed by Wickens and his colleagues of the display proximity advantage.