What are other names for causal diagrams?

• A. Linear regression graphs
• B. Directed acyclic graphs or modified path models
• C. Scatter plots
• D. Bayesian networks only

Answer: (B)

Causal diagrams are representations of how we think variables causally relate to one another. They are most commonly called directed acyclic graphs (DAGs) or modified path models.

A DAG uses nodes for variables and arrows to show direct causal influence, with no cycles allowed. This acyclic, directional structure helps in reasoning about which paths convey causal effects and which paths create confounding, making it possible to apply ideas like the backdoor criterion to identify estimable causal effects.

Modified path models refer to path diagrams used in path analysis or structural equation modeling to illustrate direct and indirect effects along specific pathways. They convey the same kind of causal structure but may incorporate more complex relationships or latent variables.

Why the other options don’t fit: linear regression graphs typically display estimated relationships from a regression model rather than a full causal structure; scatter plots show associations without implying direction or causal pathways; Bayesian networks are a broader framework of probabilistic graphical models, and while they can instantiate DAGs, they’re not the sole or exclusive term for causal diagrams.