Quantifying the effect of Behaviour Self-Regulation on well-being through causal analysis: A methodological framework...

Area of Science:

• Biomedical Informatics
• Computational Health
• Behavioral Science

Background:

• Longitudinal behavioral data analysis is crucial for understanding well-being.
• Traditional methods often confuse correlation with causation, limiting intervention insights.

Purpose of the Study:

• To apply causal inference to disentangle well-being drivers from longitudinal self-report data.
• To introduce a novel metric for quantifying behavioral consistency.
• To identify causally-informed targets for digital health interventions.

Main Methods:

• Developed the Behaviour Self-Regulation Score (BSRS) for trait-like and state-like consistency.
• Utilized causal graphical models and propensity score methods for causal effect estimation.
• Controlled for motivational and perceptual confounders in the analysis.

Main Results:

• Short-term behavioral consistency (BSRS-S) showed a stronger causal impact on well-being than long-term consistency (BSRS-L), despite low correlation.
• Causal feature selection improved well-being prediction accuracy in machine learning models.
• Identified distinct causal pathways influencing daily well-being.

Conclusions:

• Causal inference provides a robust framework for analyzing longitudinal self-report data.
• Short-term behavioral consistency is a potent, causally-driven factor in daily well-being.
• Causally-informed modeling enhances the identification of effective digital health intervention targets.