The Within-Between Myth: Rethinking the Conceptual Overreach of Multilevel Terminology

Intensive longitudinal data have renewed enthusiasm for studying psychological processes “within persons”.This trend is accompanied by a subtle conceptual slippage: the statistical labels “within-person” and “between-person” are increasingly treated as if identifying the level at which psychological phenomena occur, and as if multilevel models were uniquely suited –or even required– to study those processes. While variance-based reasoning and the transfer of multilevel terminology to repeated-measures designs make these interpretive leaps appear natural, we argue that this is unwarranted. Variance decomposition is an analytic device with no claim to where processes originate nor privileged access to them. Psychological phenomena unfold within individuals regardless of data structure or statistical analyses. Clarifying the distinction between statistical labels, models, and conceptual claims helps prevent mistaking analytic conventions for theoretical insights. By maintaining conceptual precision, researchers can capitalize from ILD without granting multilevel terminology and models an explanatory status they do not possess.