Node based analysis with network statistics

The function sem.net can be used to fit a SEM model with network data using node statistics as variables. User-specified network statistics will be calculated and used as variables instead of the networks themselves in the SEM.

The following choices of network statistics can be used: 

• degree: Degree is a centrality measure that counts actors/nodes a specific node is connected to.
• betweenness: Betweenness is a centrality measure that counts how many shortest path an actor is crossed by through a random choice. It measures how much an individual control the spread of information.
• closeness: Closeness is a measure of how efficiently a node spreads information and can be calculated by the average inverse distance from a node to all other nodes.
• evcent: The eigenvector centrality is a measure of transitive influence of each node, meaning that a node with high eigenvector centrality tends to connect with other nodes with high eigenvector centrality (Ruhnau, 2000).
• stresscent: Stress centrality is similar to betweenness centrality as it also measures the control of spread. However, while betweenness centrality measures through a random fraction of shortest paths, stress centrality takes into account all shortest paths (Szczepanski et al., 2012).
• infocent: Information centrality is defined as the reduction in network efficiency if a target node is removed. It is a measure of node effectiveness in spreading information (Latora and Marchiori, 2007).
• ivi: Integrated value of influence is a measure that combines different centrality measures (Salavaty et al., 2020a)
• hubeness.score: Hubeness score is a component of IVI and measures a node’s influence in sits urrounding environment.
• spreading.score: Spreading score is another component of IVI and measures a node’s spreading potential.
• clusterRank: Cluster rank is a measure of clustering that takes into account a node, its neighbors, and their clustering coefficients.