Structural Equation Modeling in Aviation Is It Too Good to Be True?

Abstract

It is difficult for aviation researchers to perform complex experiments to demonstrate various types of causation such as mediation, moderation, and so on. Therefore, structural equation modelling (SEM) applied to correlational studies is increasingly becoming de rigueur in the aviation area. To engage in SEM, it is merely necessary to measure all the constructs of interest, and the SEM program provides the path coefficients, significance tests, goodness of fit indices, and all the rest. It is extremely convenient to be able to replace difficult-to-perform experiments with relatively easy-to-perform SEM studies. Therefore, it is not surprising that SEM is becoming increasingly favored. However, the easiness is suspicious. The present article examines carefully whether SEM soundly fulfills its promise to provide strong evidence of causation or of the extent of the causation, as indexed by path coefficients.