University of the West of England
Bristol, UK

Within most interactive computer systems, graphical user interfaces (GUIs) and the icons that they embed are important ‘interface agents’ that can be used to facilitate effective and efficient human-computer interaction. However, the design of icons and icon sets for particular applications is often quite problematic because of the many ‘uncontrollable’ variables and design parameters that are involved. Bearing this in mind, in this paper we discuss some of the problems of designing and evaluating icons for use within particular application domains.

Two experiments are described: a design experiment and an evaluation experiment. In the design study, two groups of students each designed an icon set for use within a particular application domain (either electronic mail or information retrieval). In the evaluation study the same students, acting as a single group, evaluated a ‘mixed’ selection of icons. The icons in this ‘evaluation’ set originated either from the design experiment or from a professional source. In order to investigate the influence of domain and prior experience on recall and preference, the data that we have obtained has been subjected to statistical analysis. In this paper we will describe the findings from our research and discuss their implications with respect for developing a basic model for designing icons.