separating within- and between-person effects
Definition
Separating within- and between-person effects is the methodological challenge of distinguishing statistical relationships that arise from stable differences across individuals from those generated by fluctuations within individuals over time in intensive longitudinal data. In intensive longitudinal designs, person-wise sample means are commonly used as proxies for true person-wise means to simplify analysis, but this approach can introduce bias wherein observed correlations between person-wise means reflect within-person dynamics rather than true between-person associations. The problem is particularly acute when the number of time points per person is low, between-person variance is small, and within-person effects are strong, demonstrating that adequate separation requires methods that jointly estimate within- and between-person effects.
Sources: Haslbeck & Epskamp (2024)
Related Terms
Applications
Separating Within- and Between-person Effects and Multilevel Vector Autoregressive Models
Two-step approaches that use person-wise sample means to within-person center data and estimate between-person relationships can produce biased estimates of between-person effects when within-person correlations are present. Correlations between person-wise means can be induced by within-person correlations alone, even when no true between-person correlation exists in the data-generating mechanism.
Sources: Haslbeck & Epskamp (2024)
Separating Within- and Between-person Effects and Intensive Longitudinal Data
Intensive longitudinal data collected from multiple subjects measured repeatedly in principle allow for separation of within- and between-person variance, making such designs essential when the research target involves statistical relationships at either level. However, the practical application of person-wise sample means as computational simplifications can compromise this separation through bias in estimated between-person parameters.
Sources: Haslbeck & Epskamp (2024)
