Test graduate-level ability to analyze and apply methodological approaches for addressing measurement error, confounding (including time-varying confounding), and causal mediation in longitudinal studies linking accelerometer-derived physical activity patterns to cardiometabolic outcomes in older adults. Assess knowledge of accelerometer data processing (non-wear, epoch aggregation, intensity metrics, compositional constraints), validation/calibration and error-in-variables corrections, use of DAGs to identify confounding/mediating pathways, g-methods (marginal structural models, g-computation), longitudinal mediation estimands (natural and interventional direct/indirect effects), model selection, diagnostics, and sensitivity analyses. Expect formulation of analytic plans for given scenarios, identification of bias sources, choice of estimators, and rigorous interpretation of causal claims.