A Note On Standardized Coefficients as Structural Parameters

Abstract

Claims that standardized coefficients, such as path coefficients, are inappropriate for assessing the invariance of a causal process across populations are critically examined. It is argued that such claims are usually based on specious reasoning and that there are instances when theoretical considerations dictate the comparison of standardized coefficients. It is more fruitful to view this issue in terms of the scales used to represent variables in structural-equation models than to view it in terms of choosing between standardized and unstandardized coefficients.