An Empirical Evaluation of System Development Methodologies

Many methodologies and techniques may be used in the development of information systems. Two widely used methodologies are: system development life cycle, and proto-typing. In this article, we have extensively evaluated the two methodologies based on field data collected from information systems professionals in business and industry. Specifically, the two methodologies are evaluated for their applicability during various life cycle phases, by types of systems, and by the amount of problem structure. Various attributes associated with the methodologies have also been assessed. Our results support the use of a contingency approach for the selection of a proper methodology for systems development. Article: There are numerous methods available and used for developing information systems (IS). These methods may be categorized into three classes (DB Directions, 1985): methodologies, which are organized, systematic approaches for handling the system life cycle or its major parts; techniques, which are means of accomplishing specific tasks in the system life cycle; and tools, which are software packages to support one or more techniques. Characteristics, usage, and applicability of the techniques and methodologies have been reported in the literature (Colter, 1984; Couger, Colter, and Knapp, 1982; Doke and Myers, 1986; Gingberg and Ariav, 1986; Necco, Gordon, and Tsai, 1987: Palvia and Nosek, 1989; Vessey and Weber, 1986; Wells and Naumann, 1985). In this article, we focus on methodologies for information system development. Several methodologies are reported in the literature (Davis, 1982; Davis and Olson, 1985; Gore and Stubbe, 1983; Gremillion and Pybum, 1983; McDonald, Riddle, and Youngblut, 1986; Naumann, Davis, arid McKeon, 1980; Naumann and Jenkins, 1982; Nosek, 1988; Sumner and Sitek, 1986). The most widely used methodology is the system development life cycle (SDLC), which comprises of a sequence of well-defined linear tasks. The tasks are derived by breaking the system's stages (analysis, design, development, and implementation) into phases, activities, and tasks (Davis and Olson, 1985; Necco, Gordon, and Tsai, 1987). In response to various shortcomings of the SDLC approach as well as an ever-growing backlog of IS applications, alternate methodologies are being utilized. According to Gremillion and Pybum (1983), other approaches for system development are: using commercial application software packages, prototyping, and user-developed systems. Of these, using software packages precludes organizational IS development (although, the package developer must have used some methodology), and user-developed systems are usually ad-hoc, simplistic, narrow in scope, and prone to being of poor quality (Davis, 1982). Therefore, any large-scale comprehensive organizational information systems are developed using one of the two methodologies: system development life cycle and prototyping. The predominant use of SDLC and prototyping is also supported by surveys of information systems professionals and organizations (Palvia and Nosek, 1989). Briefly described, prototyping is a four-stage methodology (Naumann and Jenkins, 1982) consisting of: (a) identifying the user's basic information requirements, (b) developing a working prototype system, (c) implementing and using the prototype system, and (d) revising and enhancing the prototype system. Note that may be a number of iterations of the last two stages. In this article, we focus on the SDLC and the prototyping methodologies. In practice, the system development life cycle may vary (and does vary) from user to user, and organization to organization in terms of the specific details, but the overall character remains essentially the same. In order to ascertain that an organization is using the SDLC approach, we included only those users that used a formal implementation of the SDLC approach, either obtained from a commercial vendor (e.g. SPECTRUM, PRIDE, SDM, CARA) or an in-house implementation of the SDLC. Many of these implementations are generically referred to as application system development methodologies (ASDM); as such we will also refer to it as ASDM. In our view, ASDM and SDLC are synonyms (the only difference is that ASDM is a formal implementation of the SDLC). The previously cited references discuss the characteristics, strengths, and weaknesses of the two methodologies mainly on the basis of views, opinions, and experiences. Our purpose is to objectively evaluate the two methodologies based on the actual experiences of IS professionals in business and industry. Objectives For the two methodologies, ASDM and prototyping, the following objectives are addressed in this article. 1. Assess their utility and applicability during various phases of the system life cycle. Irrespective of the methodology used, there are always activities related to analysis, design, development, and implementation. It is possible that the two methodologies may have different levels of applicability in the various phases. Besides, the literature describes prototyping both as a methodology and as a technique for various activities during the life cycle phases. 2. Assess the utility and applicability of the methodologies for different system types, namely: operational systems, management information systems, decision support systems and strategic information systems. In this work, strategic systems do not specifically refer to systems for competitive advantage, but to a broader range of systems designed to support the strategic/executive level activities. 3. Assess the utility and applicability of the methodologies for structured and unstructured system problems. 4. Determine the perceived values of the attributes associated with the methodologies. The following attributes/dimensions (based on prior literature, experience and pilot tests) are considered for each method:  Project control (i.e. how much does a method facilitate project control)