An evaluation of three user-system interface specification techniques

TR Number
Date
1988
Journal Title
Journal ISSN
Volume Title
Publisher
Virginia Tech
Abstract

User-System Interface (USI) design is a highly iterative process involving empirical testing and evaluation. The existence of a design specification is implicit in this design process. The quality of the design specification impacts the length and cost of the design cycle. A survey was made of available USI specification tools to determine: which USI components they can specify; their relation to a finite state model of the USI; and if they are usable by the human factors engineer as a basis for an analytic analysis of a USI design to detect violations of excepted USI design guidelines. Four categories of tools were surveyed: semi-formal techniques, Backus-Naur Form grammars, programming languages, and transition networks. An engineering tradeoff analysis was performed based on four pragmatic criteria: understandability, efficiency, expressive power, and fidelity.

The results of the tradeoff analysis suggest that specification tools differ in representational strength and it is best to use a set of tools for a complete specification. A behavioral study using human factors engineers was performed to validate the results of the tradeoff analysis. Human factors graduate students were trained in the use of one of three specification tools and asked to perform an analytic analysis looking for design defects. Fifteen design defects were embedded in each of the specifications. Results suggest that using two of the tools together is the most effective specification technique. In light of these results, current and future software design practices were examined to determine what role the human factors engineer can have on the design team.

Description
Keywords
Citation
Collections