Estimating causal effects on psychological networks using item response theory

Area of Science:

• Psychometric research
• Network analysis
• Causal inference

Background:

• Network models offer an alternative to latent variable models in psychometrics.
• Confirmatory group network comparison faces computational challenges with large Ising models.

Purpose of the Study:

• To estimate causal effects on network state and strength when direct network estimation is infeasible.
• To leverage mathematical equivalencies between Ising and item response theory (IRT) models.

Main Methods:

• Utilized a two-parameter logistic explanatory IRT model.
• Demonstrated method through simulation and application to empirical data from randomized controlled trials.

Main Results:

• The IRT model successfully recovered causal effects on both network state and strength.
• Causal effects on network strength were found to be common and uncorrelated with effects on network state.

Conclusions:

• Causal network models, estimated via IRT, provide valuable insights into intervention impacts.
• The method overcomes computational limitations of traditional Ising models for large networks.