*2.2. Instruments*

Apart from sociodemographic information (e.g., sex, age, migration background, living situation, diagnosed somatic and psychiatric disorders in the family) that was used to describe the sample, we obtained data from three validated and often used instruments to assess psychopathological symptoms and quality of life. The strength of a network analysis lies in exploring complex associations among a large number of diverse psychological features. Thus, we selected instruments that assess many different aspects of psychopathology and quality of life in a dimensional/continuous way. While network estimation approaches to handle categorical and ordinal data exists, such data types are still regarded as suboptimal [36]. Moreover, the selection of instruments was driven by the limited time provided by the schools to complete the entire questionnaire battery (max. one school hour of 50 min).

Specifically, data obtained through the following instruments were included in the network analysis:

The Youth Self-Report (YSR, [37,38]) is a widely used self-report instrument to measure a wide range of behavioral and emotional problems (112 items rated on a three-point scale). Item ratings are summed up in eight syndrome scales ('socially withdrawn', 'somatic complaints', 'anxious/depressed', 'social problems', 'thought problems', 'attention problems', 'dissocial behavior' and 'aggressive behavior'). Additionally, the items can be aggregated to three broadband scales, a total problem score, an internalizing problem score and an externalizing problem score. The YSR raw scores were used in this study. However, we also calculated the percentage of clinically relevant problem scores by using the available norms (cut-off: *T*-score > 63 for broadband scales). Internal consistencies were excellent for the broadband scales (Cronbach's alpha > 0.86) and acceptable for the syndrome scales (Cronbach's alpha between 0.56–0.86).

We used the SCOFF questionnaire [8,39] to screen for eating disorders, an aspect that is not covered in the YSR questionnaire but which is particularly relevant for overweight and underweight populations). It assesses five core features of eating disorders, including significant weight loss, intentional vomiting, body dissatisfaction, loss of control over food and food intrusive thought. The presence/absence of each symptom is rated on a dichotomous scale ('yes' vs. 'no'). The total score representing the number of present eating disorder symptoms (possible score range 0–5) was used in the present study. According to the authors of the original SCOFF version [39], a score ≥ 2 represents an increased risk for eating disorders. A recently published meta-analysis reported a pooled sensitivity of 86% and specificity of 83% using full-syndrome eating disorders or other established eating disorder questionnaires as reference [40].

Moreover, we used the KIDSCREEN scales [41] to obtain well-being and quality of life in different domains including 'self-perception' (satisfaction with own body and appearance), 'parent relation and home life' (assessing the quality of relationship with parents, feeling understood by them, being able to talk with them), 'social support and peers' (assessing the quality of peer relationship, spending joyful time with friends, helping each other, being able to rely on friends), 'school environment' (satisfaction with the school environment, getting along well with teachers, being able to concentrate well) and 'social acceptance' (assessing the absence of bullying). Items are rated on a five-point scale; higher subscale scores indicate higher levels of well-being/quality of life. In addition to the subscales, a general measure of well-being ('KIDSCREEN-10 ) was calculated. In the present study, the raw scores of the general and subscale measures were used. Internal consistencies of the scales ranged from 0.77 to 0.89.
