MEDIATION IN CAUSAL LOG-LINEAR MODELS

Abstract

The analysis of the causality is important in many fields of research. To obtain the causal effects in a causal log-linear model I propose a theory, using the odds ratio and the concepts proposed by Pearl, whenever possible. I start calculating the effects in a simple mediation model where I analyze a new interaction effect, which I call cell. After I calculate the effects in a mediation model with two parallel mediators. If they are correlated, Pearl's theory cannot be used and no alternative theory has been proposed so far. To resolve this issue I present a new concept of causality. The mediation models are essential in the marketing, then to it, for a more complex analysis, I apply my causal theory.

Data:
Received: 15 Jun 2016
Revised: 10 Aug 2016
Accepted: 6 Sep 2016
Published: 20 Sep 2016