A Method for Estimating Indirect Effects in Path Analysis

Abstract

The decomposition of causal components into direct and indirect effects maybe substantively important, because the decomposition allows the consideration of how causal effects occur. Causal models are useful analytic tools because they allow both the author and reader to understand explicitly the assumed order of effects. The interpretations of decompositions calculated as a part of the analysis depend on the assumed causal order of variables. Which associations are to be decomposed depends on the purpose of the analysis and the presentation of results. It would serve little purpose to use the methods explicated in this paper to calculate a wholesale collection of indirect effects; unless of course, these were required by the research questions which motivated the analysis. The methods explicated herein should ease the burden of analyzing causal models, but they are not substitutes for reflective analyses of social data.