System Analysis as Scientific Inquiry

Information systems are understood as models or representations of an application domain. System analysis is a mode of inquiry for purposes of understanding the domain and effecting change in it. Scientific inquiry also aims for understanding and description of domain. In contrast to system analysis, scientific inquiry is based on a 4000-year history. Its processes and methods are well accepted and arguably successful. This paper explores the parallels between the two processes and shows implications of viewing system development as a kind of scientific inquiry. A descriptive survey of system development in organizations presents empirical indications as to whether these parallels are in fact observed in system development practice. INTRODUCTION Information systems are models of an application domain (Iscoe et al., 1991; Jackson, 1995). For example, a production planning and control system is intended to represent, with its software structures and its data, the production facilities of the business. System development begins with system analysis, an inquiry into and documentation of the application domain, and continues to the implementation and use of the final software system. System development is a relatively young field, and there still exists much debate about its methods and processes. In contrast, there is a rigorous process of inquiry into a domain that is much older, is well accepted in the community practicing it, arguably successful as measured by its results, and well regarded by its wider audience. This is the process of scientific inquiry. This paper explores the parallels between the processes and artifacts of science and system analysis. As both are concerned with the investigation and description of a domain, we suggest that much can be learned by applying well-accepted and successful scientific methods to the process of system analysis. At the same time, we are interested to what extent scientific practices are already present in system development. We have therefore operationalized the recommendations in the form of a survey instrument. A questionnaire was initially sent out to the CIOs of the 100 largest users of personal computer systems in New Zealand. From these, 15 responses were obtained. The survey and the descriptive results are presented in the following sections of the paper. The paper begins by examining the development process, and then continues with the role of conceptual models as theories of a domain, modeling languages as paradigms, and social aspects. We conclude the paper with a brief discussion and outlook towards further work.