Interaction network analysis of the six game complexes in high-level volleyball through the use of Eigenvector Centrality

Abstract

Social Network Analysis establishes a network system and provides information about the relationships (edges) between system components (nodes). Although nodes usually correspond to actors within the network (e.g., the players), it is possible to stipulate game actions as nodes, thus creating a network of the flow of game actions. In this study, Eigenvector Centrality (a form of weighted centrality that considers n-order connections) was used to identify differences in the centrality of distinct game actions within each of the six game complexes of volleyball. Thirteen matches (46 sets, 2,049 rallies) of the final round of the 2015 FIVB’s World Grand Prix (Women) were analyzed. Results showed that analyzing actions as actors (i.e., nodes) offers a clear and comprehensive understanding of game flow and poses an interesting alternative to mainstream research where players are considered nodes. Functional differences between the six game complexes were highlighted, denoting the validity of such division. Out-of-system playing (i.e., having to set the attack under non-ideal conditions, e.g., in KI, KII, KIII and KIV), emerged as a regularity of the game and should be translated into the training process.