Figure 2
Beliefs about social relationships and shared beliefs relate to self-reported paranoia. (A) Principal Components Analysis Bi-plot. We performed k-means clustering (\(k=3\) optimal clusters, based on the Elbow test) on 7 different network features (network size, strong ties, centralization, constraint, density, kinship, and assumed shared belief). We visualize the first two principal components as dimensions. The direction of the vector loadings of each feature illustrate which feature drives which cluster. For example, assumed shared belief, strong ties, network size, and kinship distinguishes cluster 3 from the other clusters. (B) High paranoia and Low paranoia networks. For illustrative purposes we depict a typical network described by a high paranoia participant and a low paranoia participant. Paranoia is characterized by more and stronger self-reported ties. (C) Distribution of features across clusters. We note that cluster 3 has the strongest assumed shared belief, kinship, network size, paranoia, and strong ties.
