The Use of Impact and Self-Impact Correlations In Causal Analysis

Abstract

It has recently been suggested that impact and self-impact correlations be extended to verify analyses drawn on static, crosslag, and dynamic correlations. We argue that this extension to causal analysis contains serious flaws and may be inappropriate in model building. The analysis is based on three aspects of measurement: similarities between the “deficiency score” method and pre/post measurements, shifts in reference points, and patterns by which variables may change.