Participants

Sample selection was carried out using the convenience sampling method. The sample consisted of 7723 in-school adolescents between the ages of 13 and 18 years at the time of recruitment (Median = 15, SD = 1.48), of which 55% were girls, from four countries (Chile = 15%, Spain = 24%, Mexico = 29%, Peru = 32%). Participants were from mixed schools (50%) and single-sex schools (50%). Regarding their academic performance, 42% of participants had obtained good grades, 42% had passed all subjects, and 16% had failed one or more subjects. Over 98% lived in an urban area while 2% lived in a rural area. Socio-economic levels may be quite high because the percentage of parents (at least one of the two parents) with university studies was high (Chile 39%, Spain 81%, Mexico 74%, and Peru 69%).

Instrument and variables

The evaluation instrument used was the YOURLIFE project self-report questionnaire [34], with three different versions depending on age (13, 15 and 17 years). This instrument has been used in several national and international surveys carried out among adolescents and based on other questionnaires [35] (see Appendix).

Socio-demographic variables

Information concerning the socio-demographic data of the participants was collected, among them sex, age, country of residence, or parents’ educational level.

Problematic technology use

This was assessed on the basis of four questions that included time spent using mobile devices, interacting with peers through social networks, writing emails, chats or tweets, and eating while looking at one’s smartphone (e.g., I spend my time distractedly, looking at my smartphone, tablet or computer, even when I could be doing more productive things). The response format was a Likert-type scale from 0 (Strongly disagree) to 4 (Completely agree). Principal component analysis was performed on four items, one factor based on the standard eigenvalues > 1 criterion was found, explaining 51% of the total variance. The internal consistency of this scale (α = 0.68) was acceptable.

Structured leisure activities

The frequency of structured leisure activities (volunteer, artistic, sports and family activities) during the last year was measured. The response format had five response options (Never; Less than 1 day a month; 1–3 days a month; 1–2 days a week; and 3 or more days a week).

Unstructured leisure activities

The frequency of unstructured leisure activities (in public spaces, leisure centers, and youth venues or nightclubs) during the last year was measured based on four questions, with the same response format as for structured leisure activities.

Inhibitory control

This executive function was measured on the basis of responses to one statement (I do things without thinking about them) with five options (Never; Almost never; Sometimes; Almost always; Always). In order to measure inhibitory control, an inverse item was used and scores were inversed.

Goal setting

Two executive functions (planning, and achievement of goals) were measured on the basis of responses to two statements (I plan the things I do; I usually finish what I start) with five options (Never; Almost never; Sometimes; Almost always; Always). Principal component analysis was performed on two items; one factor based had eigenvalues > 1, which explained 69% of the total variance. The internal consistency of this instrument (α = 0.55) was weak.

Procedure

Educational centers from four countries were invited to take part by email which provided the link to the website designed to offer detailed information to the participants (http://www.proyectoyourlife.com/). Schools agreeing to participate in the project received a protocol with instructions on the survey process, and on the specified date, each school administered the questionnaire in person during school time. The general design of the study was approved by the Ethics Committee of the University of Navarra, and each new participant school was asked to follow the project’s specific ethical guidelines. The respective ethics committee of each participating country had access to the questionnaire prior to application.

The rationale of the study was explained verbally to the students in all schools involved in the study. Moreover, participants received written information detailing the objectives of the project as well as their rights. The framework questionnaire was administered after parental permission for this research was received. A self-administered anonymous questionnaire was administered. No incentive for participation was offered, but each school was sent a report with the overall results of their center, and the implementation of specific educational programs was encouraged to prevent the problems detected in the study. The data of the present study were collected before the COVID-19 pandemic, and the detailed data collection procedure can be found in a previous publication [34].

Data analysis

The prevalence rate of PTU was calculated on the basis of participants scoring 4 (Totally agree) on any of the four questionnaire items. Statistical control of potential confounding variables was applied across the whole set of ANCOVA analyses, including as covariates adolescent age and adolescent gender. First, one-way ANCOVA with Country of origin (Spain, Chile, Mexico, Peru) was conducted for PTU. Then, a mixed model ANCOVA 4 (Country of origin) × 4 (Structured leisure: volunteer, artistic, sports, family) was conducted, with Structured leisure as a repeated measures factor. A similar mixed model was executed, with Unstructured leisure (public spaces, nightclubs, leisure centers, youth venues) as a repeated measures factor. In these analyses the Greenhouse–Geisser degrees of freedom correction was applied when necessary. As an index of effect size, the partial eta-squared statistic was used (small effect η2 = 0.01; medium effect η2 = 0.06; large effect η2 = 0.14) [36]. For the posterior contrast analysis, the Bonferroni correction method was used to control the type I error rate, with α < 0.05. The correlation matrix between PTU and the study variables (inhibitory control, planning goals and leisure activities) was conducted. These data analyses were performed using IBM SPSS Statistics 27.

Robust statistics are more appropriate when the data are not multivariate normal (Mardia’s normalized coefficient exceeded 23.76). Goodness-of-fit of the model was assessed with the normal theory maximum-likelihood (ML). A number of fit indices were calculated, including: (a) the overall χ2, (b) Satorra and Bentler [37] (1994), robust maximum-likelihood (S–B), (c) the comparative fit index (CFI), (d) the Satorra–Bentler robust comparative fit index (RCFI), (e) the root mean square error of approximation (RMSEA). The most commonly used criterion for an acceptable fit is CFI ≥ 0.90, and RMSEA ≤ 0.06 [38]. A confirmatory factor analysis (CFA) assessed the adequacy of the hypothesized measurement model and the associations among the latent variables: Problematic technology use (indicators: four items), Goal setting (indicators: planning and achievement of goals), Structured leisure activities (indicators: volunteer, sports, family), and Unstructured leisure activities (indicators: public spaces, shopping centers, youth venues, and nightclubs). The artistic activities indicator was deleted for Structured leisure activities because the factor loading was smaller than 0.32 in an initial CFA. Inhibitory control was included as an observed variable.

In the structural model, Structured leisure activities, Unstructured leisure activities were predictors of PTU. Moreover, Structured and Unstructured leisure activities predicted Goal setting and Inhibitory control, which served as the intervening variables in the relationship between leisure activities and PTU. Significant correlations were allowed among the Unstructured and Structured leisure activities. The estimation of indirect effects was accomplished using a SEM model. These analyses were performed using the EQS 6.2 Structural Equation Program.