Method

adhere, and of which “The goal is primarily description and explanation not prescription”

(Easton 1992 p. 7).

The default answer to the collection of qualitative data is a case-study design, but case studies do not necessarily yield qualitative data (Damgaard, Freytag & Darmer 2000, Dubois, Araujo 2004). The term case study covers a number of different research designs and is commonly signifying both a process of inquiry and the results of this process (Stake 2000). Different designs yield different results (Damgaard, Freytag & Darmer 2000, Easton 1995, Johnston, Leach & Liu 2000, Stake 2000), but also reflect different purposes and paradigms. The design differs along a number of dimensions of which the discussion of multiple-cases versus case-studies of one or a few cases, seems to be one of the most ardent (Dyer, Wilkins 1991, Easton 2003, Eisenhardt 1991, Eisenhardt 1989).

The number of cases is important for scientists applying a positivist paradigm, because the number is important for generalization. Therefore, proponents of a multi-case approach are to be found among this group of scholars (Damgaard, Freytag & Darmer 2000). The underlying discussion is a matter of the validity of case-studies. For scholars applying a positivist paradigm there are two possible views on this issue (Tsoukas 1989). If focus is on statistic generalization, case studies have low validity, and are only suitable for the investigation of local context, as explorative studies or pilot studies. If focus is on analytical generalization, case studies are regarded to be a valid experimental design on the

condition that multi-case replication logic is applied. Yin (2003) offers two criteria for intentional sampling for multi-case studies; 1) a literal replication; sampling cases for which similar results are expected for predictable reasons 2) theoretical replication; sampling of cases for which contrary results are expected for predictable reasons.

In constrast, Stake (2000), who takes a constructivist approach, favours a single case approach. He argues this on the basis of an underlying difference of purpose between the intrinsic case and the instrumental case. The intrinsic case is one which is interesting in its own right, and the reporting of the case is arranged for the reader to draw his own conclusions. The instrumental case helps the researcher to pursue an external interest and facilitates the understanding of this interest. It is arranged for didactical learning. The generalizations from a number of instrumental cases are arranged as final conclusions by the researcher who presents what has been learned.

A critical realist case study resembles the instrumental case more than the intrinsic case.

The researcher presents data from the empirical and the theorizing of mechanisms as the foundation for a description of the events in the actual. It is claimed that a critical realist

approach is the only one which sustains claims of the existence of general knowledge based on case-studies (Easton 1998, Sobh, Perry 2006). Positivists rely on correlation, and therefore a single case is marginal, while constructivists argue that the case study reflects the researcher as much as the object of interest (Easton 1998). In contrast, critical realists can claim that the knowledge achieved from case studies is epistemologically valid to the extent that it (Tsoukas 1989)

 Establishes similarities between events or differences among events

 Establishes mechanisms and contingencies

 Produces theoretical re-descriptions which increasingly capture new layers of reality This type of knowledge directly informs the issues of a critical realist analysis; structures, powers and contingencies. And for that purpose, one case is enough (Easton 2003).

2.3.2 Interview-data

The interview as a self-reporting method for data collection is much discussed. Behind this discussion there is a concern about

 The data you collect through this method (Gubrium, Holstein 2002)

 The extent to which the techniques and methods applied during the interview influence the data that you collect (Breakwell 1995)

 The analysis of texts (Silverman 2001)

 The reporting of data are further issues of debate(Czarniawska 2002)

These concerns originate in the recognition that it is the researcher’s interest which forms the platform for dialogue. But the effect of this recognition for the data collection

procedures depends on the approach to the respondent. If the respondent is regarded as a vessel of data, which is already there, it is important to avoid contamination of data due to interviewer influence. If the respondent is regarded as a narrator, the interview must be arranged in a way, which enables the respondent to be the active partner in the construction of the narrative (Gubrium, Holstein 2002).

Paradigms influence how you execute, code, and interpret interviews and create a multiplicity of possible methods for the collection, analysis and interpretation of data. But paradigms also help making the choices. Sayer (1992) claims that studies of social structures are theoretically informed and informative narratives rather than formal analysis of timeless mechanisms; i.e. data cannot speak for themselves, theory is needed and applied to interpret data.

Thus, the data sought in a critical realist approach are the qualitative data about meaning (Sobh, Perry 2006). Meaning is the instrument which supports inference between the empirical experience and the real domain of structures and mechanisms, to the conceptualization of the event in the actual. In this study, the inference between the empirical and the event in the actual rests on the respondents’ perception of the value of the triadic constellation of relationships.

Perception, which is a mental process, is not directly accessible. Instead I have asked questions which yield data on behaviour in the triadic relationships (the activities, the division of labour). And I have asked question which encourage respondent to express their attitude to the other actors, and to evaluate the other actors’ contribution to the value creation process. By combining these three types of data, I reach an understanding of the meaning; how the actors perceive the value of the triadic constellation of

relationships.

The ability to extract meaning can be supported by analysis of semiosis, defined as the intersubjective production of meaning (Fairclough, Jessop & Sayer 2004), but it is not the ability to interpret the text as text, which produces the result. The result rests on the ability to re-contextualize the narratives collected through the interviews and other sources of information into a story of the field (Czarniawska 2002). The construction of the narrative of the situation or event conceptualizes the actual (Mutch, Delbridge &

Ventresca 2006).

2.3.3 Data collection and analysis

Four triadic channel relationships have been selected as cases of value creation in three party constellations. They are not sampled for literal or theoretical replication with the purpose of confirming the working of this process for predictable reasons. This is not possible due to the irregularity of mechanisms in a critical realist approach. Instead, the cases are sampled on the basis of the common empirical features. They all involve a supplier, a merchant and a customer, who are all communicating with each other. The respondents are representatives of the supplier, the merchants, and the customer. They are the individuals who are involved in the actual interaction related to value creation in the triad. The cases are sampled in order to learn about value creation in triadic business relationships and to explore this phenomenon. One case may have been enough, but the choice of four cases enables the observation of similarities and differences among four instances of triadic value creation, which facilitates the establishment of mechanisms and contingencies.

The data was collected through semi-structured interviews conducted on the basis of an interview guide (appendix A). The interview guide, which directs the interviews, is constructed with the purpose of extracting data from the respondents. However, the dialogue is conducted in a way which leaves as much room for the respondents as possible in order to encourage them to take the lead in the exploration of the questions raised during the interview. As described above I have collected data on the behaviour in the triadic relationships (the activities, the division of labour). And I have collected data on the respondents’ attitude to the other actors, and their evaluation of the other actors’

contribution to the value creation process. These data have been combined to reach an understanding of the actors’ perception of the value potential of the triadic relationship in which they participate.

Data on the behaviour is collected through the discussion of activities and division of labour. Activities have been organized in a number of value dimensions. Background interviews point to the product as a distinct and separate element of value. Therefore, he interview guide is organized so that the product is introduced as the first value dimension.

In this way the dialogue is initiated on the basis of a concept which is familiar to the respondents. Afterwards the other value dimensions are introduced and discussed. In order to find out how the value of each dimension is evaluated relatively to other value dimensions, respondent are asked to prioritize the value dimensions. For the prioritization each value dimension is printed on a separate piece of paper and handed to the

respondents in a randomized order. This procedure is applied in order to avoid the indication of any preferred ranking.

Data on the respondent’s attitude towards and evaluation of the other actors’ contribution are collected through a dialogue on

 the respondents’ assessment of the relationships

 the role of the merchant in triadic relationship and

 the respondents’ interpretation of the ways in which the relationships influence each other

The discussion of the role of the intermediary is supported by illustrations of three possible roles; a provider for the customer, a distributor for the supplier or a trader (Gadde, Snehota 2001). Likewise, the conceptualization of channel relationships as triadic micro-networks is introduced to the respondents. The dialogue on the structure of channels is discussed with the respondents on the basis of illustrations of channels as chains of dyadic relations, and as triangles. This issue is included in order to shed light on the respondents’

visualization of the triadic relationship. Do they see it as a chain or is their interpretation corresponding with the theoretical conceptualization of the relationship as triadic?

Background interviews have also pointed towards a number of additional issues of importance for the understanding of value creation in the triads:

 Pricing issues related to discounts and profits

 The concept of the proficient timber merchant

 Competition in the industry

 Decision making in the building material industry

 The importance of / influence from merchant chains

These issues are included in the interview guide in order to get data on the atmosphere of the relationships, and on the environment.

At the end of each interview the respondents are asked if further value dimensions influence the value creation process. No respondents indicated so, but an un-anticipated administrative activity occurred during the first customer interview: In some situations the merchants make thorough basic calculation of tenders and quotations for their customers, based on the material that the customers receive from the ordering parties. Consequently, this element of administrative services was included in all further interviews.