An experimental investigation of the effects of some information system design variables on performance, preference, and learning

摘要

The purpose of this study was to investigate the effects of differences in display, viewer, and task characteristics on interpretation accuracy, viewer preference, and learning in a computerized information system. This research was prompted by the Lockheed-Georgia Company when it considered expanding its executive information system (called MIDS) to middle-level engineering managers. The authors developed an experimental framework which was applied to the design of a carefully controlled field experiment using practicing managers at Lockheed-Georgia as subjects. A mock-up of an engineering subsystem of MIDS was developed and used throughout the experiment. The binary and categorical data were analyzed using logistic regression, enabling the determination of the significant main effects and their interactions. Task-related characteristics had the strongest effect on interpretation accuracy and viewer preference. Display format was never significant as a main effect, but influenced interpretation accuracy only through interaction with task-oriented variables. Personality type, as measured by the GEFT, was never a significant factor. Although improvement in interpretation accuracy occurred over time, no comparative advantage of graphics over tables was found. This study has contributed to the evolution of Lockheed-Georgia's executive information system and its subsystems by improving the level of awareness and understanding of the interactions between task and display characteristics. These findings led to a reduced emphasis and a more critical appraisal of the benefits of computer graphics for future designs.