Target Mean Analysis — Direct Effect

Context

This analysis plots the Direct Effect Curves of the selected Driver Nodes. It works like Target Mean Analysis for Total Effects, with one difference: while the x-values of a Driver Node are simulated across its range, the distributions of the other nodes in the network are held fixed. For instance, when computing the curve of the Target Node $\mathit{Pleasure\_(4)}$ as a function of $\mathit{Intensity}$, the marginal distributions of $\mathit{Pleasure}$, $\mathit{Corresponds}$, and $\mathit{Easy\ to\ wear}$ are maintained.

The Direct Effect concept page explains what holding the other nodes fixed means, which nodes are exempt (the Non_Confounder class and Not-Observable Factor nodes), and how the single-number Direct Effect is derived from these curves.

In the following plot, the Direct Effect Curves of all Driver Nodes are displayed at the same time. However, please note that all of these curves are calculated individually.