Measurement Case Study 2: Theory-testing research: testing a necessary condition

quickly a customer base. The extent of coupling, the number of inter- faces between an application and the network, was determined based on the technical characteristics of the project. Some applications, such as voice services or person-to-person text messaging, involve applica- tions that are integral parts of the mobile network, i.e. interconnected with many network elements. In contrast, peripheral applications are often connected to the mobile network, or in many cases to an appli- cation platform, through a single interface. For each case i.e. for each successful project we determined the organizational configuration by assessing the four dimensions of the organizational form coordination integration, ownership integra- tion, task integration, and knowledge integration using a qualitative interview with the project manager. Based on the interview data, we characterized each dimension as a low, medium, or high level of inte- gration. To check the measurement validity of our ratings, we com- pared the researcher’s ratings of ownership integration and task integration with the ratings by the project manager for these dimen- sions. The project manager rated these dimensions on a five-point scale using a questionnaire with the statements “the extent that the operator invested in the mobile application development project” and “the extent that the operator performed the project tasks”. No major deviations were found between the assessment of the researcher based on the interview data and the assessment of the pro- ject manager in the questionnaire. We performed the following procedures to collect the data. As indi- cated above, the project managers of the different projects were our key informants for both the dependent and the independent concept and the classification of the project into one of the six types of innova- tion. From each project performed in a single firm, the project manager was interviewed. If multiple firms were involved in the pro- ject, we interviewed only the project manager from the most important firm in some cases we did interview project managers from multiple firms though. At the project manager’s company, each project manager first completed a questionnaire in the presence of the researcher. Our presence allowed us to clarify the questionnaire if nec- essary and also might have acted as a barrier to self-report bias. The questionnaire contained not only questions about the organizational dimensions of the project but also about the respondent’s opinion on the performance of the project. After having completed the ques- tionnaire, respondents were interviewed in a semi-structured way, cov- ering the same topics as in the questionnaire and in the same order. The researchers’ prior experience in the mobile telecommunications industry facilitated the interviews and increased the richness of the data and also enabled us to build the questionnaire using wording familiar to the respondents. The interviews enabled us to validate the answers we obtained with the questionnaire, ensuring measurement validity. We found no serious problems or diverging interpretations of key constructs. The field notes obtained during the interviews were converted into a detailed summary immediately after the interview. In some cases we contacted respondents afterwards to seek clarifica- tions on data that appeared unclear. Letting informants review the case reports was not a standard procedure, however, because the congruence between the questionnaire data and the interview data was considered sufficient to ensure measurement validity. A single researcher mostly conducted the interviews. To improve reliability of the collected data, the first interviews were conducted by the two researchers together to become experienced with the method and to develop an agreed-upon approach to follow. To achieve participation of all firms in our research and to achieve good quality of the data they would provide, we promised all involved firms that we would keep their data confidential. Therefore, we will here present the collected data in such a way that individual projects are “anonymous” and cannot be identified.

5.4.9 Data presentation

The data that we collected are represented in Table 5.5 in the column beneath the case number. It turned out that for each of the six types of innovation we had at least one successful project.

5.4.10 Data analysis

The hypothesis states that each successful product innovation project has an ideal typical organizational configuration. We tested this hypothesis in all 15 cases by comparing the “observed” pattern pre- sented in Table 5.5 in the column beneath the case number with the “predicted” pattern as specified in the column “Ideal type”. We con- sidered the hypothesis confirmed if the observed configuration is a perfect match with the predicted configuration, i.e. if all four values are exactly the same. The hypothesis is rejected if the observation does not perfectly match the predicted configuration. We conducted 15 tests and in each of them the proposition was rejected. Theor y-testing r esearc h Par t II Table 5.5 Data for 15 successful product innovation projects Incremental Incremental change for Radical application change for peripheral application change core application Ideal Case Ideal Case Case Case Case Ideal Case Case Case type 1 type 2 3 4 5 type 6 7 8 Incremental Coordination L L L H H H H L H H L interface Task H L L L H M M L L L L change Ownership H L L L H M L L L L L Knowledge H M L L H M M H L L L Ideal Case Ideal Case Ideal Case Case Case Case Case type 9 type 10 type 11 12 13 14 15 Radical Coordination H M H H H H H H H H interface Task H M L L M M L L L L change Ownership H M L L H M L L L L Knowledge H M H L H M L L L L