Materials

All measures and coding are described in Supplementary Material III. A brief explanation is provided below.

Loneliness

In the adolescent samples, the Loneliness and Aloneness Scale for Children and Adolescents (LACA) subscale ‘loneliness’ in relation to peers was used (Goossens & Maes, 2020). A mean score was calculated for the 12 items (range 0–3). Cronbach’s alpha for W1–5 varied between 0.86 and 0.94. In HSL, the Roberts UCLA Loneliness Scale (RULS-8) was used (Goossens et al., 2014). A mean score of the eight items was calculated (range 0–4) for all participants who answered at least 80% of the items. Cronbach’s alpha for W1 was 0.87. In al-RISCO, the Three-Item Loneliness Scale (UCLA-revised: (Hughes et al., 2021; Russell et al., 1980)) was used. A mean score of the three items (range 1–3) was calculated for participants who completed the whole scale. Cronbach’s alpha for W1 was 0.69. Higher scores in all samples indicated more loneliness.

Smoking

In the adolescent samples, various questions on smoking initiation or smoking status were used to categorize participants into smoking initiation no (coded as 0) or yes (coded as 1) (see Supplementary Material III). For each wave after baseline, if participants indicated they never smoked, smoking initiation was coded as 0, while if participants indicated they had ever smoked, smoking initiation was coded as 1 (and the following waves of data were removed from the dataset). We chose to operationalize smoking as smoking initiation in the adolescent samples since adolescence is a period where most adolescents have not yet progressed to regular smoking, and therefore, measuring initiation allows us to capture the onset of smoking behaviour. For the university student samples, a question on smoking frequency was used to define participants’ current smoking status as ‘non-smoker’ (coded as 0; smoking frequency was never, a few puffs, or former), occasional smoker (coded as 1; smoking frequency was monthly or less), and regular smoker (coded as 2; smoking frequency was weekly/daily). We chose to operationalize smoking as smoking status since young adults in university have likely either initiated smoking or had the chance to experiment with it. Smoking status would allow us to understand the amount of smoking beyond a binary value of yes or no.

Other Variables

In the adolescent samples, biological sex (boy or girl) was asked, while in the university student samples, gender (male, female, other) was asked. In all samples, age at W1 was either asked in the questionnaire or calculated based on the date of participation and date of birth. For samples 2 and 3, educational type was categorized into three levels: (1) pre-vocational education, (2) higher general secondary education, and (3) pre-university education. For sample 5 (al-RISCO), the educational type was categorized into two levels: (1) applied and (2) research university.