Viewpoint: Action research

In an earlier issue of the Journal oflnfofmation Systems, Gilbert Mansell articulates the use of action research in information systems (Mansell, 1991). This short note is intended to complement that paper and reflect on the differences between action reseach and the positivistic approach. The author has been involved in action research for many years and has developed the Multiview methodology based on this tradition (Avison 8, Wood-Harper, 1990). This methodology has been formalized and modified through action research and that text illustrates this through six action research studies. The methodology has been further modified for use in developing countries and for nonspecialist analysts (Bell & Wood-Harper, 1992). Action research is client-centred and contextual. The researcher who wishes to investigate an organization consults with members of the organization and the goals for the research are negotiated between the client group and the researcher. The research involves a cycle in which the framework that informs the exercise and the action research process are employed. The cycle is a learning process that ideally ends with outcomes acceptable to both client group and researcher. The aim is to produce practical outcomes for immediate use. It also aims to inform and develop the conceptual framework within which the action research is operating. Its assumption is that each social context is unique rather than an instance of a general case and it does not produce law-like generalizations about organizations. It is useful to contrast action research with the positivistic methods of the natural sciences as an illustration of action research concepts and because positivism is used in much of the research in information systems and in the practice of developing information systems. Table 1 compares positivism and action research on the criteria of scope, methods, role of the researcher, goals and outcomes.