intensive longitudinal data
Definition
Intensive longitudinal data refers to repeated measurements collected from individuals at many time points, generating enough within-person observations to estimate separate network models for each participant. In the study by Johal and Rhemtulla, four such datasets were used to fit individual psychometric network models, enabling examination of how network-derived metrics varied across people and predicted outcomes such as life satisfaction. Because each participant contributes a sufficient number of observations, both network-structure variables, including density, global strength, and maximum modularity coefficient, and network-informed variables, such as an individual's average value on the most frequently central node, can be computed at the person level. This design makes it possible to treat network properties as individual-difference variables rather than fixed group-level constants.
Sources: Johal & Rhemtulla (2024)
Related Terms
- psychometric networks (1 shared article)
- prediction (1 shared article)
Applications
Intensive Longitudinal Data and Psychometric Network Models
Psychometric network models estimated from intensive longitudinal data represent a psychological construct as a set of pairwise associations among variables, with each edge reflecting the unique partial correlation between two nodes after controlling for all others. When intensive longitudinal data are available, individual-level networks can be estimated, producing person-specific metrics such as density and centrality that would remain constant across participants if only a single group-level network were fit. In the Johal and Rhemtulla study, this individual-network approach allowed the predictive utility of five distinct network-derived metrics to be compared across four empirical datasets.
Sources: Johal & Rhemtulla (2024)
Intensive Longitudinal Data and Node Centrality
Because intensive longitudinal data permit estimation of an individual network for each participant, the centrality of each node can vary from person to person, making centrality a viable predictor of external psychological outcomes. Johal and Rhemtulla found that an individual's average value on the most frequently central node outperformed network-structure variables as a predictor across most outcomes examined. This result supports using intensive longitudinal data to identify which variable in a system most consistently occupies a central position, and then treating that variable's level as a summary of the broader network-instantiated construct.
Sources: Johal & Rhemtulla (2024)
