The Use of Structural Equation Models in Interpreting Regression Equations Including Suppressor and Enhancer Variables

摘要

It is shown that the usual interpretation of "suppressor"effects in a multiple regression equationassumes that the correlations among variables havebeen generated by a particular structural (causal)model, namely, Conger’s (1974) two-factor model. Adistinction is drawn between the technical definitionof "suppression," which is more fittingly labelledenhancement, and suppression as the appropriateinterpretation of a regression equation exhibitingenhancement when that equation has been generatedby the two-factor model. It is demonstratedthat a number of models can generate enhancementbut cannot sensibly be interpreted in terms of themeasuring, removing, or suppressing of irrelevantor invalid variance. How a regression equation isinterpreted thus depends critically on the structuralmodel deemed appropriate.