| Contracting in software business: Analysis of evolving contract processes and relationships | ||
|---|---|---|
| Prev | Chapter 5. Empirical research design | Next |
Klein and Myers (1999, 72) introduce a following set of seven principles for conducting and evaluating interpretive field studies in information systems:
The fundamental principle of the hermeneutic circle
This principle suggests that all human understanding is achieved by iterating between considering the interdependent meaning of parts and the whole that they form. This principle of human understanding is fundamental to all the other principles.
The principle of contextualization
Requires critical reflection of the social and historical background of the research setting, so that the intended audience can see how the current situation under investigation emerged.
The principle of interaction between the researchers and the subjects
Requires critical reflection on how the research materials (or “data”) were socially constructed through the interaction between the researchers and participants.
The principle of abstraction and generalization
Requires relating the idiographic details revealed by the data interpretation through the application of principles one and two to theoretical, general concepts that describe the nature of human understanding and social action.
The principle of dialogical reasoning
Requires sensitivity to possible contradictions between the theoretical preconceptions guiding the research design and actual findings (“the story which the data tell”) with subsequent cycles of revision.
The principle of multiple interpretations
Requires sensitivity to possible differences in interpretations among the participants as are typically expressed in multiple narratives or stories of the same sequence of events under study. Similar to multiple witness accounts even if all tell it as they saw it.
The principle of suspicion
Requires sensitivity to possible “biases” and systematic “distortions” in the narratives collected from the participants.
These steps form a quality standard roadmap for conducting and evaluating a well-defined research setting for case studies. They base their approach on the hermeneutic circle where the central idea is to iterate between the parts and the whole of the studied subject. Other issues, they propose to be considered when evaluating the quality of research, are the principles of contextualization, interaction between the researchers and the subjects, abstraction and generalization, dialogical reasoning, multiple interpretations and the principle of suspicion. However, Klein and Myers warn, not to follow these steps too mechanically as this is only one perspective, which gives more rigour to conduct and report results of the case studies. Further they suggest that none of these principles should be left out arbitrarily and that the researcher should consider how and which of the principles apply in any particular research setting. The steps were capitalized in this study as follows:
The first step of the fundamental principle of the hermeneutic circle was extensively applied. During this iterative process the transcriptions of the interviews were analysed carefully several times. The output of this effort was a good understanding of the data as a whole and also a partition and classification of the data in its smallest units of analysis that were the process activities. After this stage the construction work started from these parts towards the whole; i.e. COTS, tailored and MOTS contract maps. This analysis went further to elaborate new and refined models of contracting in software business.
The principle of contextualization is a more difficult task to describe explicitly as the research subject was understandably a sensitive issue for the companies and the interviewed managers. Therefore in this context it is not possible to discuss the organizations and their operating environments in a too detailed way that could possibly expose their identities. On a general level the research context is depicted in Chapter 1. In Section 5.2.2 the method and reasons of the process for selecting the focal software companies is described.
The question how the researcher creates the data is discussed within the principle of interaction between the researchers and the subjects. This issue is discussed more in Section 5.2.3. The researcher himself participated in seven interview cases (one company was visited three times) and from the remaining seven cases he received only the interview tapes that later were transcribed. Thus in these cases the researcher did not interact directly with the participants as he relied on the transcribed texts. Though in all cases the interviewees were approached by emails in order to get possibility to the member checking procedure. However, the response to the email messages was meagre, only in some cases did the responses give further information.
Abstracting and generalizing process is explained from its essential parts in Section 5.2.5 in order to give the reader the possibility of following or even inferring to the same conclusions as the researcher using the available data. Appendix D shows the main part of the process activity data that has been applied when the researcher has made his conclusions regarding the contracting process. However, it must be emphasized that in reality it may be difficult only by using the available data to infer the same conclusions with the same exactness, as the researcher has been able to do, as the researcher has much more data available and knowledge at hand. However, it should be included so that the readers can follow how the researcher arrived at his theoretical insights.
One of the most difficult tasks is how to separate the researcher’s own intellectual history, i.e. the dialogical reasoning. The researcher has been working in a managerial position in the past during several years in a software company producing COTS, tailored and MOTS software, ref. also the beginning of this chapter. This gives him in a positive way advantage and good pre-understanding when approaching the focal research question. But then the disadvantage is that he also may have strong preconceptions about the subject. This may start already from the basic research setting, including the predefined interview questions as well as the selection process of the companies. The researcher e.g. noticed a personal development and learning process during the interviews towards a better listener from a conversationalist. Further all the time during the analysis process the researcher must be able to make a distinction between his ideas and the opinions expressed by the interviewees.
In this research context it was not possible to find different versions of the story, i.e. multiple interpretations. This resulted from the fact that only a few companies were entered several times and only in some cases the interviews were done with different persons.
In the final principle of suspicion the researcher questions critically possible false preconceptions and aspects of reality presented to him. This is a problematic question and in this study the researcher has taken the standpoint that all the interviewed managers have given their opinions as truthfully as possible, i.e. they have not deliberately tried to change the truth of the story. Of course, this is quite difficult to prove, but here again the pre-understanding of the researcher may be an advantage. However, the interviewees may leave out some interesting and perhaps difficult issues without the researcher noticing.
Ferlie and McNulty (1997, 368) characterize high quality research as built up of two aspects, namely of internal validity and external validity. The internal validity deals with the rigour of the theoretical, methodological and empirical bases underlying the research project, whereas the external validity reflects the relevance and accessibility to user groups outside the research community. The evaluation of concepts involve completeness, simplicity, understandability and ease of use and the assessment of the models are done again in terms of their fidelity with the real world phenomena, robustness and internal consistency among other things (ibid). According to March and Smith (1995, 260) if the building activity is first in its domain then the artefact and the adjoining claims itself provide a contribution to the research community. In other words "Actual performance evaluation is not required at this stage".
Dawson (1997) emphasizes the importance of the researcher getting involved in the data collecting and observation of the studied phenomena. He calls this "getting his hands dirty". He further prefers processual research the longitudinal element included in the research design, as this is important in collecting the data and becoming an accepted part of the workplace under study. Further he stressed the importance of different data collecting methods to contextualise, validate and crosscheck the interview data. In the processual analysis the data is broken down, categorized and sub-categorized (labelled) and then reconstructed, i.e. built up again to a form quite different from the original and analyse possible emerging observable patterns. This helps the researcher to explain and understand the object of study better (ibid). Bryman (1988) states that research mainly utilizing semi-structured interviews for data collection is generally characterized by not having the processual sense compared to action and participative research methods.
One form of validation is the method of triangulation where comparison of data from different sources and by different methods are done (Silverman 1993). The second method is the so-called member checking where the transcribed interviews and interpretations based on them are presented to the interviewees in order to correct errors or to add issues not found in the texts and their interpretations (Hirsjärvi & Hurme 2000). The investigation of several events within one or several cases provides good opportunities for comparison and insight for theory development (Halinen & Törnroos 1995, 513). For the empirical material of this study an effort was made for member checking but the results were, however, quite meagre as there were only three cases out of twelve that responded to the enquiries with more information.
Yin (1994) emphasizes the following three principles in the data collection:
Use as many data sources as possible. This is especially important in case research, using several sources gives different angles to the object studied and this may even reveal conflicting arguments. This includes the triangulation of data.
Prepare a case study database where the raw data is entered as well as all the data and documents that the researcher produces.
Maintain a logical and clear chain of reasoning so that every finding can be traced back to the original source where the fact was first encountered.
In this study the above principles were implemented by collecting data from interviews and other documents were gathered. Two interviewers were used, in two cases the companies were interviewed two to three times. The database was formed to include all digital material, among these were the transcribed tapes, field notes, letters, emails, telephone conversation memos, analysis memos and contracts. All the individual process activities were indexed and marked in the documents where they were found. See e.g. the next chapter and the quotations.
Qualitative study conducted properly gives possibilities for a local explanation of the observed phenomena (Alasuutari 1995). To further generalize the findings and the elaborated model to reflect the contracting process of the software producing companies universally more research on the subject is needed.
Next, the possible sources of error in this study and how they may affect the results are discussed:
Selection of the analysed software companies and chosen analysis perspective
The selected twelve software companies are presumed to well represent their line of business. The selection process tried to include a sample of companies using different business modes. Thus, the companies have not been selected using any statistical random sampling technique. In order to get a wider perspective for the study the research body was selected to consist of local Finnish firms with international business and Finnish software companies established in the USA. In both environments the companies have to take into account demanding contractual processes.
The analysis perspective was defined to be the perspective of the software company. The customer procurement practices concerning software acquisition and relationships with software companies is available in abundant literature. Thus the customer side of the relationships was left out from the interviews in order to get a more precise viewpoint. This was a conscious choice and the justification for this was that this study analyses the interplay of the contracting and software development processes and their evolvement together with their relationship development as seen from the software developing company point of view.
The coding of the process activities
The method for coding was developed for this specific data and it was based on the used models and analysis tools. This perspective had been developed during several trials with the empirical data as described in Section 5.2.5. From literature no other suitable coding method was found or proven for this purpose.
Opinions presented by the interviewees
All the interviewees held in their respective companies a managerial position, i.e. they were key-persons who had a clear and wide understanding of contracts, contract negotiations and exploitation of contracts in software production. With the anonymity of the interviewees as well as of the companies were pursued to enable to establish a confidential atmosphere during the discussion sessions. Thus the interviewees were assumed to hold better possibilities to discuss the sensitive contractual issues.
Based on the limited nature of the used models
Applied models naturally represent only one kind of possible model, but they have been previously used and tested in other cases. However, there is no evidence, of it used in the same way and in the same line of business, closest similar use to e.g. the Möller and Wilson model is done by Halinen (1994). Some of the problems encountered applying these models are discussed more in Section 8.1.