The participant adolescents were recruited through schools in five different communities in the middle of Sweden located, due to convenience, within commuting distance of the university. Fifty-four schools were approached for participation and eight agreed to participate constituting 34 different classes covering different socioeconomic groups and study programs (theoretical and vocational).
In this study, 650 adolescents aged 15–17 years were asked to participate. Out of this sample of 650 adolescents, 22 were excluded because they had passed 18 years of age or had given nonsense answers, and 12 did not complete the questionnaire, rendering a study group of 616 adolescents, a participation rate of 94.8%. There were 295 girls and 319 boys, while two answered ‘other’ on the question of gender. The mean age was 16.4 years (SD = 0.50), 70 adolescents were born in another country (11.4%), of whom 47 were born in a country outside Europe (7.6%) and 23 (3.7%) in another European country.
Data was collected in Spring 2020, during February and March. The study was designed as a community/school based, cross sectional study to investigate the psychometric qualities of ARQ.
A letter was sent out by email to the headmaster/mistress of each of the schools, followed by telephone contact one or two weeks later. When a school decided to participate, a letter with information about the research was sent out. The headmaster/mistress who had said yes gave contact possibilities to the teacher of the class and a date for the research was decided.
Before all survey occasions, teachers and mentors in all classes were given oral information about the study, and information sheets were issued about where students can apply for support if needed. At all data collection sessions, at least one of the authors was present throughout the session (F.H. or E.K.) and was available to answer questions and respond to any reactions. Before the students' participation in the study, an oral review was held about the purpose of the study and information about the study with emphasis on voluntariness, anonymity, and instructions for implementation. After the oral review, information letters and consent forms were distributed to all students. After signing consent forms, the students were referred to the questionnaire survey via the website www.iterapi.se., which is connected to the platform at Linköping University and is considered completely safe and secure. Since students in Sweden use laptops, both during lessons and for homework, the online solution was decided on together with the headmasters of the schools. It took less than a lesson (60 min) to complete the questionnaire.
The questionnaire package consisted firstly of a page covering demographic data and then the measures.
Adolescent Resilience Questionnaire (ARQ)
The ARQ is a self-report instrument intended to identify the adolescent’s ability to be resilient even when he/she is experiencing difficult circumstances . It was developed by Gartland and colleagues  to investigate internal and social resilience factors in adolescents aged 11–19 years during the previous six months. The instrument has 88 items, with 12 inherent scales divided into five domains. The domains are: a) Internal: with subscales self-esteem, emotional insight, negative cognition, social skills, and empathy/tolerance, b) Family: with subscales connectedness and availability, c) Peers: with subscales connectedness and availability, d) School: with subscales supportive environment and connectedness, and e) Community.
The items are formulated into claims such as: “My family listens to me”, or “My life has a sense of meaning to me”. It covers the previous six months and the answer options lie on a five-point Likert scale, ranging from 1 (all the time) to 5 (never). In the Australian study , 451 students from 11 schools answered the formula and the authors showed it to have good internal consistency, with a Cronbach’s alpha ranging between 0.70 and 0.90 for all scales except the subscale Friends – availability.
The ARQ was translated into Swedish by three researchers with great knowledge of the subject and with permission from D. Gartland . When consensus was reached by the researchers, the ARQ was back-translated by a native English professor. The necessary corrections were then made until a final version was developed.
Sense of coherence scale-13 (Soc-13)
A Swedish-translated version of the Sense of Coherence Scale (SOC), created by Antonovsky , was used in the present study. This is a well-known scale and has been used in many studies. The questionnaire is based on Antonovsky’s early theoretical model, which aimed to increase understanding of the relationship between coping strategies, stress, and health. SOC is understood in terms of three components: comprehensibility, manageability, and meaningfulness . In the present study, the SOC-13 was used. SOC-13 is a shorter version of SOC-29 . The questions relate to different areas of life and are answered on a seven-point Likert scale, ranging from 1 (very seldom or never) to 7 (very often). A high overall score suggests a strong SOC. It has been shown that SOC-13 has good validity and reliability and is valid as a cross-cultural instrument , and validated in Sweden . In this study, Cronbach’s alpha was found to be 0.86.
The Rosenberg Self-Esteem Scale  is a scale designed to measure the concept of global self-esteem. This concept is understood in its common definition as a person’s overall sense of worth . Individuals report how true the ten statements are for them. Four answering options are used, ranging from 0 (strongly disagree) to 3 (strongly agree). The total score ranges from 0 to 30, with high values indicating high self-esteem. In a study where 16,988 participants in 53 countries answered RSES, the total average alpha level for all countries, was 0.81, which indicates good internal consistency . The scale has been validated in Sweden  In this study, Cronbach’s alpha was found to be 0.91.
Relationship questionnaire (RQ)
RQ is a self-report instrument which has four items, intended to indicate participants’ attachment style . The four attachment styles identified in the RQ are thought to measure how an individual looks upon and behaves in his/her relationships. They are believed to be an effect of the person’s relational history. The four attachment styles are: secure, dismissing, preoccupied, and fearful, where the last three indicate an insecure attachment style . On the questionnaire, the participant marks on a seven-point scale ranging from 1 (disagree strongly) to 7 (agree strongly) how much they recognise themselves in the description of each attachment style. Question number 2 is designed to identify a secure attachment style and questions 1, 3, and 4 indicate insecure styles. The scale has been translated into Swedish  and was validated in 2001  on a sample of adults.
This study was approved by the This study was approved by the regional ethical review board at Linköping University (Ref. no. 220–08) and the authors have followed the ethical codex concerning information, consent, and usefulness (Swedish Research Council, 2002). There were no agreements, rewards, or payments to participate.
Internal consistency for the ARQ domains were examined using Cronbach’s alpha. To test for construct validity, the Gartland  model was fitted with the study sample using confirmatory factor analysis (CFA) with the mean- and variance-adjusted weighted least squares (WLSMV) estimator . Adjustment of the model to Swedish adolescents was made by excluding items not loading on their respective factor. Model fit was examined by Overall model fit and was tested by: Root Mean Square Error of Approximation (RMSEA), Comparative Fit Index (CFI), Tucker-Lewis Index (TLI) and Standardised Root Mean Square Residual (SRMR). Using guiding principles set out by Brown  and Schreiber et al. , several different fit indices were used. For convenience of reporting, a model was judged as having good fit when the overall picture of fit indices indicated good fit, and excellent if all of them indicated good fit: (χ2/df < 3, RMSEA ≤ 0.05, CFI and TLI ≥ 0.95, and SRMR < 0.08 [43,44,45].
Concurrent validity was examined by means of correlations between factor scores generated from CFA and a) SOC-13, and b) RSES, using Kendall’s tau. Notably, Kendall’s tau values are generally 66–75% of the size of Pearson correlations , and for comparative purposes the more conservative 75% was used, i.e. Pearson correlations of 0.10, 0.30, and 0.50 (often considered as small, moderate, and large) are comparable with Kendall’s τ values of 0.075, 0.225 and 0.375.
Factor analyses were performed using Mplus version 8.4 , while other analyses were performed using RStudio  with R version 4.0.3 .