ABSTRACT
Regression analysis can provide information regarding the extent the predictor variables in the model either directly or indirectly influence the response variable, the target. The key idea is that for any predictor variable or feature in the model, its slope coefficient is estimated, given that the values of all other model features are held constant. If all variables for which the given predictor variable correlates were included in the model, then the slope coefficient estimates the direct causal impact. This ideal state is likely never reached, but the better, more inclusive models can begin to approximate the causal impact. Moderation models analyze the impact of the interaction of two (or more) predictor variables in the model on the target. Mediation models formally analyze the direct and indirect effect of a predictor variable on the target. The indirect effect is the impact of the predictor on the target via the mediator variable, a basic path analysis model. The presented analysis includes a path analysis of the corresponding path model.
