A special case was simulation A1, in which we simulated a 1-factor model for a mix of dichotomous, ordinal (3 categories), and continuous phenotypes with factor loadings ranging from .60 to .90, to show that TATES also works well for phenotypes of different measurement levels. In this specific simulation, the correlation matrix between the phenotypes, used to approximate the correlations between the p-values, was mixed with the type of correlation (Pearson, polyserial, polychoric) depending on the measurement levels of the phenotypes involved.
