Physical activity as a system: A network perspective on behavioral attributes and self-efficacy

Abstract

There is no gold-standard method for measuring physical activity (PA), partly because it is a dynamic system shaped by interactions among behavioral and contextual attributes. Network analysis offers a systems-based framework to model these relationships. To illustrate its use, we modeled PA attributes, assessed gender invariance, and examined links with PA-related self-efficacy in 40 healthy adults (18 men, 22 women; mean age = 27.9 years) monitored for 1 week with self-report and smartphone accelerometry. Three network models were estimated: (1) relationships among PA attributes; (2) gender comparison; and (3) integration with self-efficacy variables. The latent structure revealed both positive and negative connections, with 53.42% being positive; the strongest positive edge linked weekday and weekend sitting time. Moderate PA, sitting time, and self-efficacy when access to facilities was limited emerged as central nodes, with some gender differences. Overall, network analysis provides a promising approach for modeling PA as an interacting system.