Regression tree analysis of ecological momentary assessment data

Area of Science:

• Health Psychology
• Behavioral Science
• Data Science

Background:

• Ecological Momentary Assessment (EMA) is a popular method for collecting intensive longitudinal data in health psychology.
• EMA is increasingly used to study the links between individuals' momentary functioning and health outcomes like binge eating and alcohol use.
• Analyzing complex EMA data presents challenges for traditional statistical methods.

Purpose of the Study:

• To introduce and evaluate Regression Tree Modelling (RTM) as a viable alternative to multilevel modelling for EMA data analysis.
• To demonstrate the utility of RTM in identifying complex interactions between variables influencing health outcomes.
• To explore the application of RTM-generated decision trees for clinical assessment and intervention tailoring.

Main Methods:

• The study applies Regression Tree Modelling (RTM) to analyze real-world EMA data.
• RTM is compared with traditional multilevel modelling for its ability to handle complex interactions.
• Methods for extending RTM to analyze nested data structures common in EMA are presented.

Main Results:

• RTM effectively models associations between predictor variables and continuous outcomes, accommodating higher-order interactions without explicit term creation.
• The generated decision trees offer interpretable insights for guiding health assessments and interventions.
• The paper illustrates both the advantages and limitations of RTM in practice, including extensions for nested data.

Conclusions:

• Regression Tree Modelling (RTM) is a valuable tool for analyzing intensive longitudinal data from Ecological Momentary Assessment (EMA).
• RTM's ability to capture complex interactions and generate decision trees enhances its utility for health psychology research and practice.
• The demonstrated extensions allow RTM to effectively handle nested data structures inherent in EMA studies.