Factorial modeling is justified as a method for analyzing correlations in support of causal inference. The method is illustrated and compared to path analysis, LISREL-type analysis, canonical correlation, and commonality analysis. Predictions of impacts of policy manipulations are demonstrated. (Author/CP)

Contrary to Anderson (EJ 187 936), his rule for equation identification is a necessary but not sufficient condition; furthermore, the choice of two-stage or ordinary least squares depends on results and not on methodological properties of estimators. Modification of Anderson's rule and a means for choosing between estimates is offered. (Author/CP)