This chapter gives important methodological, theoretical, and practical guidelines to the information system (IS) researchers to carry out a historical study. This study shows how a new theory can be discovered inductively from historical studies using a methodological guideline from Mason, McKenney, and Copeland (1997b), using multiple data collection methods, such as semistructured interviews, archival files, and published news, and using novel data analysis methods from learning and intelligent systems, such as the Self-Organizing Maps (SOMs), SOMs combined with U-matrices, and the Bayesian network modeling. It also outlines the benefits, the main problems, the characteristics, and the implications of historical research in the information system field. Finally this chapter gives future some research directions of historical research.
Introduction And Background
The goal of this chapter is to outline how a new theory can be discovered inductively using historical research approach. Historical research is difficult to carry out because its nature is interpretive, the research of which has been largely ignored so far (Bannister, 2002). There are studies outlining the importance of historical research in the area of information systems (IS) (Ein-Dor & Segev, 1993; Locker, Miller, Richardson, Tebeaux, & Yates, 1996; Mustonen-Ollila, 2005; Sandberg, 2005), studies involving single or multiple time periods and being longitudinal in its nature (Barley, 1990; Heiskanen, 1994; Pettigrew, 1989, 1990), and a study of the historical evolution of information system and time generation categories’ development (Friedman & Cornford, 1989). Historical research can be characterised as longitudinal research from the past, because both longitudinal research and historical research are characterised by the measurement of differences or change in a variable from one time period to another, and longitudinal data analysis permits insights into the processes of change (Davies, 1994; Menard, 2002). In the historical studies, the research begins far away in the past when the distant causes for the events did not mean anything to each other, but later on effected to the current effects.
The fundamental assumptions of the historical research are as follows. First, in historical research, several simultaneously used data collection methods are needed. If the bulk of the gathered data is qualitative, consisting of interviews and archival material, historical research methods should be adopted (Copeland & McKenney, 1988; Mason, McKenney, & Copeland, 1997a, 1997b; McKenney, Mason, & Copeland, 1997) such that the suggestions of Pettigrew (1985) should be followed mostly when gathering and organising the data.
Pettigrew (1985) studies the management of organisational change, which is the phenomena in its real context. Pettigrew (1985) points out the aspects of continuity and change, and his study illustrates how change as a continuous incremental process can be interspersed with radical periods of change. These major decisions are associated with major changes in business market conditions (Dawson, 2003).
Second, the terminology and concepts between the historical research and the other forms of over time dependent research, such as longitudinal research and time series research are interchangeable, because they all study the phenomena of change and patterns over time but from different perspectives.