Hidden Variable Discovery

Context

• Hidden variables, especially unobserved confounders, can be a challenge for estimating effects.
• The Hidden Variable Discovery helps you search for such variables.
• It is based on the following principle:
  • Graph theory stipulates that in a serial connection of variables, such as A-B-C, the variables A and C are marginally dependent.

This report computes the G-test and the independence probability between two variables of the network that are connected by a path of length one or two and are not part of a V-structure.

All existing paths of length one or two (without a V-structure) will be tested.

The independence probability is computed, and only paths that are not independent are kept in the report.

Computing independence between variables:

• G-test: The value of the independence test G is computed from the data associated with the network between each pair of variables that are the ends of the paths.

• Degrees of freedom: Indicates the degree of freedom between the ends of each path.

• p-value: Represents the independence probability of the G-test between the ends of each path.