78 1 Density (network level): Represents the proportion of directed relationships to the number of possible directed relationships (Wasserman & Faust, 1994). A density of 1 means that everyone asks advice from everyone else (Borgatti et al., 2013). 2 Centrality (consists of two measures: Network and individual level): a Network indegree centralization represents the proportion of the sum of differences in centrality between the most central member in a network and all other members. This indicates whether there is a center (very central members) and a periphery (members with very low centrality scores) regarding the asking for advice (Borgatti et al., 2013). b Individual indegree centrality is an index that represents the number of ties any specific member has (D’Innocenzo et al., 2016). The members with the largest number within their team perform the most central roles (Sinnema et al., 2020; Smith et al., 2018; Tsai, 2001). 3 Reciprocity (dyadic level): Indicates the proportion of observed directed relationships that are reciprocated in a network relative to the number of possible directed relationships (arc-based; Borgatti et al., 2013). When two members turn to each other for advice, this is a reciprocated relationship. By combining these social network measures, we expect it to represent distributed leadership as follows: Relatively high density, high reciprocity, low indegree centralization, and multiple central members (more than one teammember). We compared the association between these measures and teammembers’ roles (school principal, coach-teacher, teacher) in order to explore which role most commonly takes up the central position. Next, we calculated the correlations between density, reciprocity, and centrality within Ucinet. For each network measure, we attributed the individual data to matrices per school team, for reciprocity and centrality we used ‘difference’ scores between all team members of a team and for density we used the raw scores of ties being absent (score 0) or present (score 1). In this way, we tested the correlation of the three social network measures and especially the added value of the reciprocity measure. Afterward, we compared all school teams’ advice networks with a cross-case analysis, to enhance generalizability and to deepen our understanding of how to describe and measure distributed leadership (Miles & Huberman, 1994). By doing so, we ordered the school teams based on low to high scores of density, reciprocity, and centralization, and examined whether we could exploratively differentiate between school teams, based on face validity and discussions with all authors. Lastly, to visualize and further describe the results, we created sociograms within NetDraw and placed the central members in the center, based on in-degree centrality. We considered the approach used in this
RkJQdWJsaXNoZXIy MTk4NDMw