# Identifying causal pathways with and without diagrams

- James Corter,
*Teachers College, Columbia University*
- David Mason,
*Teachers College, Columbia University*
- Barbara Tversky,
*Teachers College, Columbia University*
- Jeffrey Nickerson,
*Stevens Institute of Technology*

## Abstract

Causal modeling generally involves the construction and use of
diagrammatic representations of the causal assumptions, expressed as directed
acyclic graphs (DAGs). Do such graphs have cognitive benefits, for example by
facilitating user inferences involving the underlying causal models? In two
empirical studies, participants were given a set of causal assumptions, then
attempted to identify all the causal pathways linking two variables in the model
implied by these causal assumptions. Participants who were provided with a path
diagram expressing the assumptions were more successful at identifying indirect
pathways than those given the assumptions in the form of lists. Furthermore, the
spatial orientation of the causal flow in the graphical model (left to right or
right to left) had effects on the speed and accuracy with which participants made
these inferences.

Back to Table of Contents