Candidate cases Methodological reflection on Case Study 5

9.3.3 Case selection

The criterion of maximum variation of the value of the known concept required that at least two instances of a purchase per type of service should be selected. It would have been acceptable for this theory-building study to select cases i.e. instances of a service purchase from the small population of all such purchases in one company. This would enhance the chance of finding specific relations between type of service and kind of representative, but the resulting propositions would next need to be tested in other groups or populations i.e. other companies and types of companies. It could be possible that the propositions built in the study would only hold in the one company in which they were found. In order to avoid this scenario, cases were selected from two different companies. By doing this, the chance of finding an effect of the type of purchased service on the kind of representative involved in the communication was reduced which could be overshadowed by the effects of differences in policy between the companies in the way they select representatives but at the same time it increased the likelihood that found propositions would be robust in replication tests.

9.3.4 Extracting relevant evidence

Data were collected by asking informants which departmentsfunctions were involved in managing the ongoing supply after the purchase of a service. The answers that were obtained are listed in Table 9.7. In order to translate these various answers into values of one concept, the seven value-creating functions described by Porter 1985: 45–48 were applied. An eighth value of this concept was added based on the data that had been collected, namely the value “internal customer” i.e. a representa- tive of the users of the service as distinct from persons with a functional expertise. The result of the study thus far is depicted in Figure 9.4. Type of service Representatives of different value- creating functions according to Porter ??? Independent Dependent Figure 9.4 Proposition with discovered dependent concept

9.3.5 Coding

Although Porter’s functions are not defined in this text, there is no reason to doubt the valid and reliable coding of the collected data represented in Table 9.7A in terms of these functions as repre- sented in Table 9.8. No information is given about how the level of perceived risk was coded. Probably the company informants were asked to rate the level of risk for each purchased service as either high, or medium, or low. The reclassification of some services that originally had been considered semi-manufactured services as component services suggests that the independent concept had been measured validly and reliably.

9.3.6 Data presentation

The obtained scores are presented in the format of a data matrix Table 9.8.

9.3.7 Data analysis

In 9.1 “How to design and conduct a theory-building case study”, we advise to assess first whether there is evidence for a sufficient condi- tion. A sufficient condition exists if a specific value of the independent concept always results in a specific value of the dependent concept. This means that it must be assessed for every value of the independent concept whether a specific dependent concept is always present. In this study, inspection was needed of the three different types of service that were left after the semi-manufactured services were reclassified as component services and at the level of buyer-perceived risk of each project. It appears that a representative of procurement is always present. Because this appears to be a constant, not a concept, it can be left out of the analysis. Table 9.9, which is identical to Table 9.8, clearly shows in the first two columns case 1 and case 2 that a representative of the internal customer was always present in consumption services CNS. Based on these findings it can be argued that purchasing a consumption service is a sufficient condition for having a representative of the