Two types of networks were simulated. First (F1,F2), all regression weights in matrix B were set to .04202, or .08187, resulting in phenotypic intercorrelations of .56 or .12, respectively, i.e., the phenotypic variance-covariance matrix of the network simulations mimics the phenotypic variance-covariance matrix of two Rasch models discussed above(A2/E1,A3/E3). Second (F3), a network was simulated with 4 clusters of strongly associated phenotypes (correlation = .55), and weaker associations between clusters (correlation = .13).
