Individual explanatory variables and adjusted variables

Data of each participant were collected through a questionnaire that was answered by the participants in electronic device (PDA, Personal Digital Assistants) – there was no intervention by researchers.

The quantitative variable ‘number of operational drinking fountains” was found based on the following questions: “Are there drinking fountains in the school?” and “How many of them are working?”, which were answered by school principals.

The proxy school-environment variable for exercising comprised variables “outdoor courtyard”, “indoor courtyard”, “pool” and “sports availability after school shift”, which were built through Principal Component Analysis (PCA). Variables used in this composition were provided by school principals, who had provided information on the number of sports spaces within the questionnaire (outdoor courtyard, indoor courtyard and pool), as well as on sports activities provided by the school to students after the school shift (wrestling, soccer, volleyball, swimming and athletics). Two categories were taken into consideration, namely: “less favorable conditions” and “more favorable conditions” for exercising. The “more favorable conditions” category was featured by the largest number of indoor courtyards, larger number of sports modalities offered by the school (wrestling, soccer, volleyball, swimming and athletics, among others) and having a pool in the school.

Variable “how often legal guardians have supper with the adolescent” was measured through the question: “Does your father (or step-father) or mother (or step-mother), or legal guardian, have supper with you?”—the following answers were available: “almost never”’, “sometimes”, “almost every day” and “every day”. Variable “living with parents” was calculated by combining variables “living with the mother (yes/no)” and “living with the father (yes/no)”. Variable “having breakfast” was calculated based on the question: “Do you have breakfast?”. Answer options were “I do not have breakfast”; “I have breakfast sometimes”; “I have breakfast almost every day” and “I have breakfast every day”. Exercising was classified as inactive (0 mn/week), insufficiently active (1 – 299 min/week) and active (≥ 300 min/week), according to WHO’s recommendations [16].

The socioeconomic score was assessed in ERICA based on questions about owning assets and comfort items according to the social stratification of the Brazilian Economic Classification Criteria (CCEB), which was elaborated by the Brazilian Association of Research Companies [17]. However, no information about mother’s schooling was provided in 20.41% of questionnaires. Thus, by taking into account that the exclusion of these individuals would represent significant loss for the sample, the option was made to only sum the score recorded for “owning assets” to “having a housekeeper” (which was renamed as socioeconomic score in the current study). The socioeconomic score was featured based on three equal intervals, as defined by Moura et al., [18], namely: low socioeconomic level, 0 -12; medium socioeconomic level, 13–25; and high socioeconomic level, 26–38.