3.3.1. General Network Structure

Figure 1 shows the network plots based on the EBIC gLASSO estimation for the (**a**) overweight and (**b**) underweight group. In the overweight group, anxious/depressed mood is placed very central in the network with strong associations to socially withdrawn behavior and moderate associations to other psychopathological symptoms (e.g., thought problems) and well-being variables (particularly self-perception). Attention problems was another node with several associations to other nodes in the network, especially aggressive behavior and social problems. Aggressive and dissocial behavior were clustered at the periphery of the network. Interestingly, eating disorder risk obtained with the SCOFF was one of the most peripheral nodes in the network. However, it was placed next to self-perception (assessing satisfaction with body and appearance) and somatic complaints. Moreover, nodes related to the contact, problems and satisfaction with peers were plotted next to each other.

**Figure 1.** Network plots of the estimated EBIC gLASSO networks of (**a**) overweight adolescents; (**b**) underweight adolescents. Each node represents a variable of the YSR, SCOFF or KIDSCREEN scales. Each link ('edge') represents the partial correlation (blue = positive correlation, red = negative correlation). Thicker edges represent stronger associations. Variable abbreviations: ACCEPT Social Acceptance, AGG Aggressive Behavior, ANX\_DEP Anxious Depressed, ATT Attention Problems, DISS Dissocial Behavior, PARENT Parent Relation and Home Life, PEERS Social Support and Peers, SCHOOL School Environment, SCOFF SCOFF Score, SELF Self Perception, SOM Somatic Complaints, SP Social problems, SW Socially Withdrawn, THOUGHT Thought Problems.

In the underweight group, again, anxious/depressed mood was strongly associated to several other nodes in the network including socially withdrawn behavior, self-perception, somatic complaints and thought problems. Attention problems were strongly positively associated with aggressive behavior and social problems and negatively linked to satisfaction with the school environment. With the exception of the social acceptance domain of the KIDSCREEN questionnaire, well-being variables seem to form a cluster within the network, with well-being regarding school being strongly negatively associated with attention problems and self-perception being negatively associated with anxious/depressed mood and eating disorder risk. As in the overweight group, the SCOFF was one of the most peripheral nodes in the network.
