Path blocking

As explained in Causal Inference for Data Science

Path blocking is the technique of conditioning on variables so that a graph path cannot transmit statistical association between its endpoints.

Path blocking uses conditioning to stop statistical association from traveling between the endpoints of a graph path. Like cutting a pipe at one point, blocking one variable on a path blocks the entire path. section 7.1.5

Check the path structure before conditioning:

• Conditioning on a chain segment can block the path; conditioning on a collider can instead open an association. section 7.1.5
• A variable can be a fork on one path and a collider on another, so evaluate each path separately. section 7.1.5

Sports and injuries

To test whether Genetics and Coach are independent, inspect every connecting path. Fitness Level is a collider that can create association when conditioned on, whereas conditioning on Previous Injury can leave the endpoints independent because another collider still blocks the path. section 7.1.5

Figure 7.13 A graph with three paths from X to Y: a causal path, a path with a confounder, and another with a collider

Inspect the three-path graph to compare a causal path, a confounded path, and a collider path before choosing conditioning variables. section 7.1.5

How Causal Inference for Data Science covers it

How deep it goes

Explains blocking through chains, colliders, and multiple paths, then applies the reasoning to exercises.

Examples

Uses sports, genetics, coaching, fitness, injuries, fatigue, and proprioception in DAG exercises.

What it assumes

Assumes familiarity with graph paths, conditioning, colliders, and directed acyclic graphs.

What you can do after

Evaluate conditional independence by tracing paths and identifying variables that block or open them.

How it connects

Connects d-separation with chains, forks, colliders, confounding, and conditional independence.

liveBookmanning.com

Read Causal Inference for Data Science on liveBook Get the book at manning.com

Related concepts

• conditioningused by · Conditioning is restricting or guiding a probabilistic or generative process using specified information, variables, values, or events.

This explainer was generated by Manning's knowledge graph from the text of Causal Inference for Data Science. The section links point to the parts of the book it draws on.