Research Approach

---

This research is a real-time, broadly focused, non-quantitative longitudinal study of a single organisation. Data was collected via participant and non-participant observation, interviews and the examination of associated documentation. As this project utilises structuration theory as an ontological sensitising device, it employs longitudinal, historical and contextual research methods (Orlikowski, 1988). Nundakumar and Jones (1997) highly recommended this participant observation research approach but noted it was rarely used.

The case study involved observations covering three-and-a-half years, with concentrated periods of involvement in the first six months and periodically throughout the project. This involved attending meetings, shadowing participants, conducting interviews, observing activities, examining documentation, and sometimes active involvement. Such real-time longitudinal observation is the best way to look at change in its organisational context, but is quite rare (Pettigrew, 1995). Most longitudinal studies tend to examine change retrospectively, so that the reconstructed events tend to ignore the ambiguity and untidiness of real world situations.

A Case Study Using Ethnographic Methods

The use of a single primary study relates to an interpretivist stance. The LSP case study was investigated via ethnographic methods to create a "thick" description of events and issues (Atkinson & Hammersley, 1997). There is no attempt to completely empathise with the subjects in the case study, as the many differing perspectives and world views would have made this impossible and limiting. The subjects of the research primarily included legislative drafters and others in the drafting office, and the separate systems development unit. Separate groups, and individuals within the groups, often had different perspectives on events and issues. The aim was to understand what Geertz (1988, in Harvey and Myers, 1995) termed the "webs of significance". In other words the thick description of the LSP aims to reflect the opinions and world views of those directly involved with the LSP. Participants' responses to early drafts of sections of this chapter suggest that this has been done.

As Applegate and King (1997) point out, the intellectual foundations of case study research are robust. Demonstrable rigour is achieved by making data gathering and analysis techniques explicit and transparent, as is done below. In addition to this, the validity of the descriptions and discussions of this research project are enhanced by:

• aiming to find multiple sources of evidence for each observation (Yetton, Johnston et al., 1994). Each piece of information was checked against other sources to ensure its validity. For example, an observation might be checked against minutes of a meeting (another person's observation) and a question at a later point in time;

• actively looking for negative evidence (Dey, 1993);

• feeding back observations and interpretations to those in the field (Pettigrew, 1995);

• maintaining a log of events and interpretations to aid reflection; and

• ongoing analysis through discussions with research colleagues.

The Role of the Researcher

It is important to reflect on the role of the researcher and how this may have affected the phenomena under investigation (Prasad, 1993). As in any research project, it was impossible not to interact and be part of the observed phenomena even when trying to be a non-participant observer. Giddens (1989) termed the way in which the actions and perceptions of the researcher become part of the social interaction being observed the "double hermeneutic cycle".

On one level, the double hermeneutic cycle describes how "lay members of society routinely reincorporate social science concepts into the world" (Held & Thompson 1989: p 251). When people in the situation under investigation, or similar situations, start using the concepts developed by researchers, it illustrates the utility and plausibility of the concepts developed by social researchers. Hence, towards the end of this longitudinal project, the concepts being developed here were presented to those involved in the study for comment. As well as being appreciated by those in the field, this was a good validity test. The double hermeneutic cycle can also occur within social research projects, so that the actions and perceptions of the research directly influence their observations and results (Dey, 1993). Interpretivists recognise this is unavoidable but that it must be at least partly addressed by actively analysing the role of the researcher as part of the research context.

Initially the research focused on the LSP project as the unit of change. However, later multiple levels of analysis were used, and the relationship between the different levels was an integral component of the research. As Pettigrew (1995) noted, focusing on a project as the unit of change can be problematic, and it can prove insightful to examine change at multiple levels.

Data Analysis

Data analysis involved two primary steps. The first was the distillation of the wealth of information derived from the case studies into a rich description of incidents and issues of the project as it progressed via a process of summarisation. The second step involved the consideration of these incidents and issues against existing and emerging themes. These two steps have been iterative. An iterative approach to theory development is highly recommended (Miles & Huberman, 1994; Walsham, 1995).

Field notes provided an abundance of detail and in order to gain some initial insight into it, it was necessary to try and summarise them. Thick descriptions are an important part of qualitative observations, forming the foundation for further theory building (Dey, 1993; Walsham, 1995). These thick descriptions display the subtleties and some of the multiple perspectives which make up "reality". Many of the recent developments in the social study of technology have produced thick descriptions which have provided a wealth of detailed information, but this needs to be structured if it is to be of use (Bijker, Hughes and Pinch, 1989).

The structure of the thick description was not closely aligned to existing theoretical frameworks and emerged over time. Initially a grounded approach was employed, with observations from the field categorised initially according to incidents and issues which the systems developers and/or users had to contend with as the project progressed.

Over time, some of these incidents and issues seemed to naturally group together around certain themes. Miles and Huberman (1994) refer to this activity as noting patterns and themes, and clustering them.

In the latter stages of the research project, ongoing data from the field and reviews of my notes and other documentation were used to strengthen or discard emerging concepts. During this stage, concepts become better articulated and were sometimes adapted to ongoing observations or further information from other sources. Miles and Huberman (1994) referred to this as building a logical chain of evidence and making conceptual/theoretical coherence.

---