Cross-jurisdictional factors linked to gambling frequency in adolescents from 28 European countries: a machine learning approach

Abstract

Adolescents are vulnerable to experiencing problematic gambling, although its prevalence and potential risk factors vary across countries. This study aims to identify cross-jurisdictional factors associated with higher gambling frequency among adolescents in different European countries, using machine learning. Data were obtained from the European school Survey Project on Alcohol and other Drugs (ESPAD). A random forest machine learning approach was used to analyze the data and identify the most significant statistical predictors of gambling frequency across 28 countries. The dataset included 7,765 16-year-old respondents who reported gambling during the past year. Frequency of gambling in the past 12 months served as the objective variable for classification, and 158 potential risk factors for gambling frequency in different domains were analyzed. Gambling was acknowledged by 13.74 % of respondents and was more frequent in boys than girls (11 % vs 3.19 %, respectively). High variability was found across countries. Montenegro reported the highest estimates (26.5 %) and Malta the lowest (5.5 %). Common factors linked to more frequent gambling across countries. Alcohol use measures statistically predicted higher gambling frequency, whereas consumption of tobacco and drugs were significant factors in fewer countries. Other relevant factors for some countries included gaming, social media use, school absenteeism, parental monitoring, and certain leisure activities (e.g. going out at night, computer games, reading less). There is a need for targeted approaches that consider contextual and cultural influences on adolescent gambling. Tailored strategies to effectively prevent gambling and interventions in diverse cultural settings across Europe should be considered for development and implementation.