Lex localis - Journal of Local Self-Government_11(1)_January

348 L EX LOCALIS - J OURNAL OF L OCAL S ELF -G OVERNMENT H. Wanivenhaus, J. Kovač, A. Žnidaršič & I. Vrečko: Vienna Construction Projects: Redirection of Project Management Critical Success Factors—More Focus on Stakeholders and Soft Skills Development 1994). A two-mode network consists of two sets of units in which ties exist only between units of different sets (and units inside a particular set share no connection). In our initial data, one set of units indicated respondents (191 of them) and the other set consisted of 22 project management methods and measures (presented in Figure 4). We represented the two-mode network with a rectangular matrix, presenting respondents in rows and project management methods and measures in columns. A tie existed (denoted with 1) if a respondent nominated the particular project management method; otherwise it was 0. Therefore, the dimensions of our rectangular matrix were 191 x 22, and the matrix had 680 ones representing undirected ties (or edges) in the two-mode network. From the two-mode network, we constructed the valued one-mode network (for detailed instructions, see de Nooy et al., 2011): If two respondents in the two- mode network nominated the same project management method, there is a line in the one-mode network. Therefore, the new one-mode network has 22 vertices representing project management methods and measures (Figure 5). The values (drawn with different shades of gray) represent the number of common nominations of each pair of methods by the respondents (e.g., coordination of the project with other projects and motivation of the project members together nominated 19 respondents). Obviously, the matrix is symmetric as the relationship showed no direction. The values on the diagonal of the matrix represent the number of nominations of each project management method and measure (e.g., 52 respondents nominated analysis of stakeholders). In the next step we performed blockmodeling analysis, where we focused mainly on the partition of 22 methods and measures. Based on the dendrogram (not presented here), we selected four clusters. Units— in our case, methods and measures—from one cluster have similar connection patterns within a cluster and to other clusters. We presented the obtained clusters in a matrix representation in Figure 5 and in the graph representation of the obtained valued network in Figure 6. We depict project management methods and measures with vertices, where four shapes (circle, triangular, diamond, and box) represent four clusters obtained with indirect blockmodeling. The size of a vertex is proportional to the number of nominations of each project management method and measure, while the width and shade of gray correspond to the number of common nominations by pairs of respondents. The smallest cluster (denoted with white triangles in Figure 6) consists of two project management methods and measures: analysis of stakeholders and coordination of the project with other projects. A large number of respondents