Intl.'marion and Software Technology 1994 36 (10) 587-596 Modelling the CASE process: empirical issues and future directions Stephen King School of Business and Economic Studies, University of Leeds, Leeds LS2 9JT, UK Robert Galliers Warwick Business School, Universi~ of Warwick. Covent~' CV4 7AL, UK Many current models of information system originated before the widespread use of Computer Aided System Engineering (CASE) tools. The aim of CASE is to automate part of the development process, thus increasing productivity and quality. In this paper, we assess the validity of four significant models of system development against the experiences of three CASE-based development projects. A valid model is a necessary but not sufficient basis for process improvement. The adequacy of the models in addressing the problems encountered in the projects is assessed, and a broader model of information system development is proposed, encompassing strategic planning, organizational learning and the reconciliation of alternative viewpoints. Keywords: CASE software process models, soft systems methodology, strategic IS planning The ever-changing nature of information system development calls into question the relevance of current models of the process. One such change is the increasing use of Computer Aided Systems Engineering (CASE) tools to support or automate some of the stages of system development. Many of the current models were proposed before the widespread use of CASE tools, therefore it is timely to examine the relevance of these models in the context of CASE. In this paper we evaluate four models of system development: the waterfall model, the evolutionary model, the transform model and the spiral model. The models are compared with the experiences of three small-scale business system development projects undertaken recently for a major UK manufacturing group by researchers from the Advanced Technology Centre (ATC), University of Warwick. Each project made extensive use of CASE tools. Research questions We take the view here that a model can be either descriptive The research reported here was undertaken whilst the first author was with the Advanced Technology Centre, University of Warwick. An earlier version of this paper appeared as Warwick Business School Research Paper No. 78. February 1993. or prescriptive. A descriptive model must be a faithful representation of what is, i.e. reality. A prescriptive model describes what should be. In other words, if the current situation could be transformed into the situation described by the model, some benefit will result. Descriptive models can provide insight into complex systems and serve as the foundation for prediction, comparison and even optimi- zation. The first research question is therefore: (1) Do current models of information system development describe the CASE process as experienced in the three projects? The descriptive worth of each model will be assessed by mapping the stages of each project against the stages defined in the model. A highly descriptive model will map closely onto the project experiences. Prescriptive models should provide a framework by which to improve current practice. The second research question is therefore: (2) Do current models of information system development prescribe solutions to the problems encountered in the CASE projects'? A useful prescriptive model should identify the problem areas encountered in the projects and suggest solutions. For 0950-5849/94/010587-10~ 1994 Butterworth-Heinemann Ltd 587