I acknowledge upfront the difficulty in explaining simply and correctly a method that ‘happens sequentially, subsequently, simultaneously, serendipitously and scheduled’ (Glaser, 1998, p. 1). The spiral, and at times simultaneous, nature of grounded theory is a powerful and satisfying feature of the research method; it allows flexibility and continuous sharpening of emerging constructs via deep familiarisation with data, validation, and progressive expansion of knowledge and skills. This nature is represented in Lehmann’s (2001) research model.

Lehmann (2001a) describes the grounded theory process as a spiral that starts by collecting ‘slices of data’ in a substantive area of enquiry, which are then codified and categorised in a continuous process that moves toward saturation and results in the theoretical densification of concepts represented by a substantive theory. Figure 5.2, “Grounded theory’s building process (Lehmann 2001a).” represents this iterative process.

Although this model provides a good overview of the process of grounded theory, it fails to include the significant role of extant literature external to the substantive area in the formulation of the substantive theory, and the role of memos.

To help explain the activities that developed the substantive theory in my study, I expanded Lehmann’s (2001a) model, adding components from Eisenhardt’s (1989) and from the Glaserian literature. By doing so, it is possible to present a picture (Figure 5.3, “Expanded Lehmann’s research model.”) that includes the important role the literature played in my research and to acknowledge the key role of theoretical memos.

Entering the field is the first research action to be conducted in the context where the phenomenon is found. To enter the field I considered three important aspects:

After entering the field, access was unrestricted and I became involved in theoretical sampling. Theoretical sampling was a data collection process that continued until the very end of the research (including the write-up stage). This allowed me to take advantage of emergent themes, to acquire data continuously and to maximise observation opportunities.

All interviews were recorded in both digital and analogue forms. The tape recording was then transcribed and ATLAS.ti, a software application for qualitative data analysis, facilitated open coding and other coding activities.^[6] Open coding involves ‘running the data open’; that is, analysing the data to extract a set of categories and their properties. This is done by coding for as many categories as possible without a preconceived set of codes (Glaser, 1978). During open coding, I labelled the text of each interview, detecting new lines of enquiry, which guided subsequent data acquisition activity. Open coding generated 337 codes.

The writing of theoretical memos starts almost in parallel with open coding. Because memos are ‘the theorising write up of ideas about codes and their relationships as they strike the analyst while coding’ (Glaser, 1978, p. 83), memos are produced constantly in grounded theory, from the beginning of the analysis process until reaching closure, capturing the thoughts of analysts while they progress through the work. Memos raise the theoretical level via a continuous process of comparison and conceptualisation. They also provide freedom, flexibility, and enhance creativity (Glaser, 1978; Urquhart, 2001).

As codes and memos accumulated, I started to perceive relationships between them. This process, called theoretical coding, conceptualised the interrelation of substantive codes by generating hypotheses for integration into a theory. Therefore, theoretical codes emerged from open coding and theoretical memos, weaving a new story from the fragmentation of open coding (as suggested by Lehmann, 2001b). The grounded integration of concepts is a flexible activity that provides broad pictures and new perspectives. However flexible, theoretical codes must remain grounded on data, they cannot be empty abstractions. The concept of flexibility implies theoretical sensitivity to a number of possible coding paradigms, or coding families, consciously avoiding over-focusing on one possible explanation. Glaser (1978; 1998) provides a comprehensive (but not definitive) list of code families allowing for this flexibility.

The emergence of a pattern, in my study’s case ‘resolving conflicts’, marks the beginning of selective coding. This process refers to delimiting the theory to one or two core variable(s) which act as a guide for further data collection and analysis (Glaser, 1978 p. 61-72). By doing so, the research focused on one of the several basic social processes or conditions that are present in the data. The delimitation of the analysis to those significant variables affecting the core variable contributes to parsimonious theory (Glaser and Strauss, 1967).

At this stage in the process, the role of the extant literature becomes very important because researchers need to acquire sensitivity and knowledge on grounded concepts. The literature is therefore read as a source of more data to be compared with existing grounded data. For example, in my study, readings about trust, shared mental models, conflict, psychological contracts, transaction cost economics, and organisational psychology raised the theoretical level and improved construct definitions (as suggested by Eisenhardt, [1989]). Most of these readings were outside the substantive area of research, yet they were made relevant by the actors’ main concerns and the emerging theory.

The researcher achieves theoretical saturation when the main concern of the research can be accounted for, and further sampling fails to add significant value to the study through adding new categories or properties.

At this stage, when the theory becomes dense with concepts and enriched by relevant extant literature, the researcher has ‘discovered’ a substantive theory. Substantive theories are applicable to the particular area of empirical enquiry from which they emerged (Glaser and Strauss, 1967). They can be classified as ‘middle-range’ theories; that is, between ‘minor working hypotheses’ and ‘grand theories’ and they are relevant to the people concerned as well as being readily modifiable (Glaser and Strauss, 1967).

The objective of this section was to present an overview of the activities involved in this study. However, some concepts require further explanation, as discussed in the next section.