Conclusion, Contributions, and Suggestions 6 135 were not included in the Foundation leerKRACHT datasets. Furthermore, not every concept that would have been relevant to study leadership in collaborative innovation, such as asking teachers more specifically about their leadership practices, was included in the questionnaires. To compensate for possible biases, we used additional data gathering instruments such as semi-structured interviews with school principals and teachers and a social network questionnaire, which provided the opportunity for new topics to emerge. A strength of using data from a larger research project is that schools were not additionally burdened with data gathering. Another strength is that all schools worked with the same program, making it easier to compare schools. On the other hand, it might be that particular kinds of school decided to work with the program and that we have a slight sample bias. However, we found that the schools were representative of Dutch schools in terms of urbanity, denomination, and school size (De Jong et al., 2021). A question that arises for follow-up research is whether our findings apply in schools using collaborative innovation-oriented programs other than the specific program studied here. 6.2.4.2. Self-reports In most of the studies, only self-reported data were used, which may be sensitive to response tendencies and social desirability: these included a questionnaire on horizontal and vertical working relations (Chapters 2 and 5), a questionnaire on collaborative innovation (Chapter 2), interviews with teachers (Chapter 2), and interviews with school principals (Chapter 3). However, we used a mixed-methods design in Chapters 2 and 5 in which we combined questionnaire and interview data, enhancing the reliability of the results. In Chapter 4, we used social network data, being data about other teammembers, in Chapter 5, we triangulated school principals’ self-reports with data from teachers: this confirmed the self-reports. 6.2.4.3. Sample sizes of individual studies The sample sizes used in our studies differ. Chapter 2 has a large sample size with regard to the questionnaires. However, we also selected a smaller number of extreme case studies to study further the links between working relations and collaborative innovation. The strength of selecting case studies is that it increases the reliability of the analyses (Seawright, 2016). Chapter 4 has a sufficient sample size, since the unit of analysis in social network research is relationships rather than the number of participants. Chapters 3 and 5 are based on smaller samples sizes of 22 schools (Chapter 3) and 14 school teams (Chapter 5). However, our aim was to understand leadership processes within schools and to compare schools and relate them to several contextual characteristics. With a smaller sample size but rich datasets, we were able to study these relationships thoroughly.
RkJQdWJsaXNoZXIy MTk4NDMw