Feasibility of dynamic structural equation modeling for capturing micro-level temporal dynamics in adolescent physical activity.

backgroundWorldwide, physical activity (PA) levels among adolescents remain insufficient, highlighting the need for effective interventions. To improve these efforts, a nuanced understanding of habitual PA patterns and their fluctuations is essential. This feasibility study demonstrates the application of Dynamic Structural Equation Modeling (DSEM) to analyze micro-level temporal dynamics in adolescent PA behavior. Building on data from Beck et al. (2024) (BMC Public Health), one-week diary-based activity records from 15 adolescents (8 boys, 7 girls; mean age = 13.04 ± 1.28 years) were re-analyzed to examine: (1) the degree to which actual PA deviates from habitual PA, (2) the temporal dynamics of PA, and (3) the situational locus of control regulating these deviations.

methodsParticipants first outlined their usual weekly schedules and then documented their actual behaviors for one week. This intensive longitudinal design yielded 1,785 hourly observations and descriptive data of habitual and actual PA was analyzed. A two-stage analytic approach was employed. First, given the small sample (N = 15), we used an idiographic approach with DSEM to analyze each adolescent’s longitudinal data, employing simple models to capture within-person dynamics and summarizing results descriptively by counting credible intervals excluding zero, stratified by gender. Second, situational locus of control underlying deviations from habitual PA was assessed using Pearson residuals.

resultsN-of-1 DSEM analyses showed that both habitual plans and actual activity were relatively stable, with strong autoregressive patterns, evening reductions, and consistent positive alignment between habitual PA and action; cross-lagged and weather effects were minimal, highlighting stable idiographic dynamics.Self-controlled increases in PA occurred mainly toward the end of school days and early weekend mornings, whereas self-controlled decreases were most prominent in late afternoons.

conclusionThis feasibility study highlights the utility of DSEM for modeling fine-grained behavioral dynamics in small adolescent samples. Findings underscore stable yet context-sensitive PA patterns and demonstrate DSEM’s value as a methodological framework for future research on adolescent PA.