The study was conducted between November 2021 and July 2022 in different middle and high schools in Ankara/Turkey. The school administrations were contacted beforehand, and the families were informed about the study by the administration. The adolescents whose families gave permission were included in the study. Before the online survey started, adolescents were given the option of whether they agreed to participate in the study, and if they agreed, the survey section was opened. During the administration of the questionnaire in the schools, a researcher was present in the classroom and answered the adolescents' questions in case of incomprehension.

To determine the sample size power analysis was performed using the G*Power 3.1 program and the number of individuals to be included in the study was calculated as a minimum of 296 based on a Type I error rate (α) of 0.05, a power (1-β) of 0.95, and an effect size f 0.25 [21]. The sample consisted of 614 adolescents between the ages of 11 to 18, including 270 males and 344 females. Adolescents who were receiving medical treatment due to sleep disorders and psychological disorders were excluded.

The data were collected through an online questionnaire prepared by the researchers. The form included questions on the descriptive characteristics of the participants, such as age, sex, weight, and height. Also, to evaluate daily sleep duration, general bedtimes at night and wake-up times in the morning were questioned. The “Power of Food Scale” was used to assess hedonic hunger; the “Yale Food Addiction Scale for Children 2.0″ was used to assess food addiction; and the “Night Eating Questionnaire” was used to assess NES”. BMI was calculated using the weight and height values obtained based on the self-reports of the participants. BMI for age z-scores was evaluated using the references of the World Health Organization (AnthroPlus program).

Power of Food Scale (PFS), which is used to evaluate hedonic hunger status, was developed by Lowe et al. [22] and tested for validity and reliability in Turkish by Ulker et al. [23]. Turkish version of PFS is a 5-point Likert-type scale consisting of 13 items, each item having response options varying from “do not agree at all” to “strongly agree”. PFS consists of 3 subscales regarding food proximity, namely, food available (items 1, 2, 9, and 10), food present (items 3–6), and food tasted (items 7, 8, 11, 12, and 13). PFS total and subscale scores are obtained by summing the item scores and dividing the sum by the number of items. Higher scores indicate a higher tendency toward hedonic hunger [23]. The internal consistency reliability coefficient of the Turkish version of PFS was found 0.92. In this study, Cronbach’s alpha coefficient was found to be 0.91.

The scale assesses eating behaviors in the form of attacks in children observed for the last 12 months, considered to indicate food addiction. The final form of the scale was developed by Schiestl and Gearhardt et al. [24], and Turkish validity and reliability of the scale was conducted by Yılmaz [25]. Yale Food Addiction Scale for Children 2.0 (dYFAS-C 2.0) consists of 16 items, and all items are measured using a 5‐point Likert scale (0 = never, 1 = rarely, 2 = sometimes, 3 = very often, 4 = always). Scores on the dYFAS‐C 2.0 could range from 0 to 64, and the higher scores indicate higher levels of food addiction [24]. The internal consistency reliability coefficient of the Turkish version of YFAS-C 2.0 was found 0.90 in the original study. The Cronbach’s alpha coefficient was determined to be 0.92 in this study.

Night Eating Questionnaire (NEQ) is a scale that was developed by Alison et al. [26], which includes 14 items. The scale includes questions on the frequency of food intake after dinner, the first food of the day and morning appetite, eating in the evening and at night, food cravings, difficulty in falling asleep, the frequency of waking up at night to eat, awareness and mood during night eating, and control overnight eating behaviors. While all participants respond to the first 9 items, there is an instruction for participants who do not wake up at night to eat or have snacks to not continue to answer. Items between 10 and 12 are answered by participants who wake up at night to eat, and items 13 and 14 are answered by participants who have night-snacking habits. Item 7 questions the change in mood during the day, and those who do not experience any change in mood are given 0 points. All items except item 7 are 5-point Likert-type, each with a score range of 0–4. Items 1, 4, and 14 were inversely scored. Item 13 questions the awareness of midnight snacking behaviors, but it is excluded from the scoring. The minimum and maximum scores on the scale are 0 and 52. Two additional questions, which would constitute items 15 and 16, are recommended to be asked, but they are also not included in the scoring. In the original study, it was stated that NEQ scores equal to or greater than 25 indicate NES, and scores below 25 indicate the absence of NES [26]. The Turkish validity and reliability of the scale was conducted by Atasoy et al. [27]. The internal consistency of the Turkish version of the NEQ was found to be satisfactory (0.69) in Atasoy et al. [27] study and also in this study (0.61).

Statistical analyses of the data were conducted using IBM SPSS Statistics 22 (IBM SPSS, Turkey). In the descriptive analyses, categorical data were used as numbers and percentages, mean, and standard deviation values were used according to the normality of the numerical data. Compliance with the normal distribution was examined by Kolmogorov–Smirnov/Shapiro–Wilk tests and histogram plot. One-way ANOVA test was used to compare PFS, dYFAS-C 2.0 and NEQ scores according to the BMI for age z-score classification groups and Tukey’s test were used to compare differences between BMI categories. Pearson correlation coefficient was used to evaluate the relationship between BMI z-score with age, sleep duration, and scale scores. The mediated structural model analysis was performed to evaluate the effect of hedonic hunger on BMI via food addiction and NES. In mediated structural model analysis, data were analyzed with PROCESS v4.0 by Andrew Hayes. Bootstrapping was used to examine the significance of indirect effects. The bootstrap coefficient and confidence intervals were determined by making 5000 bootstraps. In mediation effect analyses conducted with the bootstrap method, the 95% confidence interval (CI) values obtained as a result of the analysis should not contain the value (0) in order for the research hypothesis to be accepted. The significance level was taken as p < .05.