Experience sampling method studies in physical activity research: the relevance of causal reasoning

Area of Science:

• Methodology in health and behavioral sciences.
• Causal inference in observational research.

Background:

• Experience sampling method (ESM), or ecological momentary assessment, is increasingly used in physical activity research.
• While ESM offers temporal separation of variables, it doesn't inherently establish causality.
• The identification phase is critical for defining causal estimands and assumptions.

Purpose of the Study:

• To illustrate the importance of the identification phase for drawing causal conclusions from ESM data.
• To demonstrate the use of causal directed acyclic graphs (DAGs) in specifying causal effects and assumptions.
• To highlight methods for addressing confounding bias in ESM studies.

Main Methods:

• Defining a causal estimand and constructing a DAG for the relationship between physical activity and executive functioning in older adults.
• Utilizing literature review and expert consultation to identify confounders.
• Illustrating physical and analytic control strategies to mitigate confounding bias.

Main Results:

• Directed acyclic graphs (DAGs) revealed open backdoor paths leading to confounding bias, even with temporal separation.
• Physical control via within-person encouragement designs can address bias.
• Analytic control through assessing and adjusting for confounders is also effective.

Conclusions:

• Implementing the identification phase enhances causal inference validity in ESM research.
• Researchers can make more informed decisions by clearly specifying causal effects and assumptions.
• This methodology strengthens the ability to answer causal questions using ESM data.