Research objective Methodological reflection on Case Study 1

The testing of propositions 1b, and 2b in incremental innovation projects see Table 5.2 can illustrate the advantage of the parallel single case study. The corresponding hypotheses predict not new part- ners, and similar technical capabilities in these projects. If it is assumed that in a serial case study, case 6 would have been selected for the first test, the test would have confirmed the two hypotheses. After this first confirmation, a second case would have been selected for replication. The replication strategy after a confirmation could be to select a case from a very different part of the domain from which the theory is considered applicable. Then the new case in a serial case study would not have been a case from Nokia, but a case from, for example, another economic sector. This would continue until cases were found that were rejected, and then the boundaries of the domain to which the theory applies would be determined. However, by using the parallel case study, rejections of hypothesis 1b were found immediately, indicating that proposition 1b for the small domain of the Nokia cases cannot be supported. The parallel single case study, thus, appears to be an effective and relatively fast way to dis- cover cases in which the proposition is not supported. The replication strategy, after a confirmation, could also be to select a case from the same part of the domain: the new case in a serial case study would be another case from Nokia. Then, after the second test say case 7 or third test say case 8, the conclusion would be justified that proposition 1b could not be supported for Nokia cases, and replications with cases 9 and 10 would not have been needed. This illustrates the dis- advantage of the parallel single case study approach, i.e. the potential to waste time and effort on measurement and hypothesis-testing. The danger of the parallel case study can be illustrated with the results of testing propositions 1a and 3a with respect to radical innov- ation projects see Table 5.1. The test result of case 3 is enough to conclude that proposition 1a which formulates a deterministic rela- tion is not correct, and the test result in case 4 is enough to conclude the same regarding proposition 3a. The danger is that inspection of all five tests together results in conclusions such as “but … the hypothesis is confirmed in the large majority of cases four out of five”. Such a conclusion could only be made after many replications when the hypothesis is rejected in only one case but is confirmed in all other cases, and if one accepts a “pragmatic determinism” view. Normally, a rejection of the hypothesis in a single case from the domain to which the theory is assumed to be applicable is sufficient to reject the hypothesis for that domain although it might be true for a smaller domain.