Describing and measuring leadership by applying a social network perspective 4 77 4.3.3. Design and procedure 4.3.3.1. Measurement of distributed leadership: Asking for advice In order to measure distributed leadership in school teams, fitting to the definition of leadership as exerting influence, we adopted an advice instrumental network question based on previous social network studies in education (Bryant et al., 2020; Liou et al., 2014; Moolenaar, 2012; Pitts & Spillane, 2009): ‘Who do you turn to for advice on working with the educational program?’ Participants were asked to answer this advice question for each team member from a list of their school team members. This results in a matrix form of data on who turns to whom. Team members were represented with random initials (such as AA, AB, AC etc.) in order to anonymize datasets for analyses (see Appendix 4.2 for the matrix form of the advice question). 4.3.3.2. Procedure We piloted the advice question and the listing procedure within two school teams that were not part of the sample of this study, but work with the same educational program. The participants indicated they experienced no constraints when completing the questionnaire. The research was approved by the ethical review committee for social and behavioral sciences of our university (number 20-056). After the pilot, we started the main phase of our data collection. The participants received a short explanation about the investment required for and the benefits of participating in the study before completing the social network advice question, and all participants agreed. Furthermore, we chose to set the complete network boundary (Knoke & Yang, 2008) to one teacher team per school, as all schools divided their teachers into sub teams to work on this collaborative program. Each school chose one teacher team to participate in this study. 4.3.4. Analysis plan First, we calculated the descriptive networkmeasures (density, centrality, and reciprocity). The advice network question is part of a questionnaire from the larger research project, in which we used a five-point Likert scale. We dichotomized the network measure scores to distinguish between ties being absent (score 0) or present (score 1), by recoding 1 as 0 (absent) and 2 till 5 as 1 (present). Based on matrixes of advice network data, we calculated the social network measures per school team (whole network level) by using Ucinet (Borgatti et al., 2013):
RkJQdWJsaXNoZXIy MTk4NDMw