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

BackgroundThe experience sampling method (ESM), also known as ecological momentary assessment, is gaining popularity in physical activity research. This method involves assessing participants' behaviors and experiences repeatedly over time. One key advantage of ESM is its ability to temporally separate the dependent and independent variable of interest, reducing the risk of reverse causality. However, temporal separation alone is insufficient for establishing causality. This methodology paper illustrates the importance of the identification phase in drawing causal conclusions from ESM data. In the identification phase the causal effect of interest (or estimand) is specified and the assumptions under which a statistical association can be considered as causal are visualized using causal directed acyclic graphs (DAGs).MethodsWe demonstrate how to define a causal estimand and construct a DAG for a specific ESM research question. The example focuses on the causal effect of physical activity performed in real-life on subsequent executive functioning among older adults. The DAG development process combines literature review and expert consultations to identify time-varying and time-invariant confounders.ResultsThe developed DAG shows multiple open backdoor paths causing confounding bias, even with temporal separation of the exposure (physical activity) and outcome (executive functioning). Two approaches to address this confounding bias are illustrated: (1) physical control using the within-person encouragement design, where participants receive randomized prompts to perform physical activity in their natural environment, and (2) analytic control, involving assessing all confounding variables and adjusting for these variables in the analysis phase.ConclusionsImplementing the identification phase enables ESM researchers to make more informed decisions, thereby enhancing the validity of causal inferences in studies aimed at answering causal questions.