Extending Dow and Eff?s (2007, 2009) innovative DEf01-R for modeling evolutionary network effects, a Complex Social Science http://intersci.ss.uci.edu (CoSSci) Gateway co-developed with Argonne provides complex analyses of ethnographic, archaeological, historical, ecological, and biological datasets with easy open access. Analyses of dependent variable y with n observations, X independent and other variables; imputes missing data for all variates. Several (n x n) W* matrices measure evolutionary network effects such as diffusion or phylogenetic ancestries. W* is row-normalized to sum to 1 to obtain a W, multiplied by X as WX, and allowing X and y multiplication by W: Wy measures the evolutionary autocorrelation portion of y discounting evolutionary effects of close or related neighbors. Tested for exogeneity, error terms uncorrelated with the independent variables, the 2-stage OLS results include measures of propinquity and phylogenetics and independent variable predictors that are relativized for autocorrelation.

CoSSci uses DEf01 imputed variables to let users explore complex networks of variables including terms for Wy evolutionary autocorrelation results. The user keeps the dependent and independent variables in the model, adding Unrestricted independent variables relevant to the dependent variable. Group significance tests select among these Unrestricted variables those that pass the Holm-Bonferroni (H-B) test with respect to the dependent and independent variables. All variables are then analyzed for potentially Bayesian causal networks using AIC and R library(bnlearn). Supercomputer HPC is needed because random 50% or more subsets of observations or variables using library(bootstrap) may identify alternative causal networks of variables.