EMA, or Experience Sampling Method (ESM), addresses the ecological validity limits of laboratory studies by surveying participants in their natural environments multiple times a day via smartphones. This yields intensive longitudinal data that captures within-person fluctuations.

EMA data is inherently nested (Level 1: measurement occasions; Level 2: persons). Ignoring this violates the assumption of independence. A basic random intercept model partitions variance into within-person and between-person components: $$Y_{ij} = \gamma_{00} + u_{0j} + e_{ij}$$ where $\gamma_{00}$ is the grand mean, $u_{0j}$ is the person-specific deviation (between-person variance), and $e_{ij}$ is the occasion-specific deviation (within-person variance).