psychometric networks

Psychometric networks refers to graphical models that represent psychological constructs as webs of pairwise associations between variables (termed nodes) and their mutual connections (termed edges), typically estimated as partial correlations. These network models visualize how items assessing a construct—such as negative affect, personality traits, or psychopathology symptoms—are interconnected, allowing researchers to examine the structure and properties of the system rather than treating the construct as a single latent or observed variable. To relate a network-instantiated construct to external psychological variables, researchers must summarize the network using either network-structure variables (such as density, global strength, or modularity) derived from the network's topological properties, or network-informed variables (such as the most central node) identified from the network's architecture. Recent empirical work has demonstrated that network-derived metrics vary substantially in their predictive utility, with an individual's average value of the most frequently central node emerging as the most consistent and effective predictor of external psychological outcomes across intensive longitudinal datasets.

Sources: Johal & Rhemtulla (2024)