Direction of dependence analysis for pre-post assessments using non-Gaussian methods: a tutorial.

Area of Science:

• Psychology
• Causal Inference
• Statistical Modeling

Background:

• Understanding the causal dynamics between depression and social/occupational functioning is crucial for effective psychotherapy.
• Existing methods may not adequately capture the complex interplay of variables during therapeutic processes.

Purpose of the Study:

• To introduce and validate novel methods for inferring causal directionality from observational psychotherapy data.
• To investigate the causal relationship between changes in depressive symptoms and social/occupational functioning.

Main Methods:

• Utilized a linear non-Gaussian structural vector autoregression model, suitable for non-normally distributed data.
• Employed simulations to test estimator performance across various scenarios.
• Analyzed data from 1428 adult patients using the Finnish Psychotherapy Quality Registry.

Main Results:

• The proposed methodology successfully identified causal directions in simulated data.
• Real-world psychotherapy data showed no evidence of a unidirectional causal link.
• Findings suggest shared or reciprocal causation between depression and functioning.

Conclusions:

• Presented a novel analytical tool for examining psychotherapy processes using observational data.
• Patient data indicates a reciprocal relationship or influence of third variables on depression and functioning during therapy.
• Challenges the notion of a single dominant causal pathway in psychotherapy outcomes.