Gender-Specific Network Structures in Adolescent Eating Behaviour and Body Image: A Network Psychometric Approach

Abstract

Eating disorders represent a significant public health concern among adolescents, yet traditional symptom-based approaches may overlook the complex interrelationships among psychological constructs. This study employed Latent Variable Network Modeling (LVNM) to examine the network structure of eating disorder symptoms, self-related constructs, emotional eating patterns, and social appearance anxiety in 455 Turkish adolescents (61.3% female; Mage = 15.6 years). LVNM extends traditional network analysis by correcting for measurement error, providing more accurate estimates of relationships among latent constructs. Following confirmatory factor analysis to validate measurement models, latent factor scores were extracted and subjected to network estimation using the Extended Bayesian Information Criterion graphical least absolute shrinkage and selection operator (EBICglasso). Centrality analysis identified emotional eating as the most central node (strength z = 1.50), suggesting it plays a pivotal role in activating and maintaining eating disorder symptomatology. Guilt demonstrated highest betweenness centrality (18), positioning it as a key bridge connecting different symptom domains. Bridge centrality analysis revealed that social appearance anxiety (bridge strength = 0.767) served as the primary connector between social-evaluative concerns and eating pathology. Self-compassion occupied a peripheral position with predominantly protective connections (r = -.132 with social anxiety). Gender-specific networks showed similar overall structure (global strength: female = 4.144, male = 4.136), but emotional eating demonstrated higher centrality in males (1.309 vs. 1.031). The network approach reveals that addressing emotional eating may have cascading effects throughout the eating disorder symptom network, offering insights for prevention and intervention strategies among adolescents.