Structural Equation Modeling (SEM) denotes a group of statistical methods which are predominantly applied to metric scaled and normal distributed data. Recently SEM has often been applied to categorical data, yet this demands specific parameter estimates and particular data matrices. Models in SEM with exclusively dichotomous variables have practically not been reported in the literature until now. It was tried to confirm an empirical model generated on the basis of ordinal data, after dichotomization of all variables by means of SEM. The dichotomous SEM approach was based on tetrachoric correlations and asymptotic covariances. A data set from anaesthesia on 820 patients served as an example. The path model consisting of three exogenous and four endogenous variables was tested thereafter by means of Prediction Configural Frequency Analysis (P-CFA). The comparison of SEM and P-CFA showed that a combination of both methods increases the scope for interpretations, i.e. despite congruent findings, each method yielded non-redundant information. The advantages and disadvantages of both methods are discussed.