An analysis of the predictive factors of digital gaming addiction on psychological well-being in young people using machine learning
Related Experiment Video
Updated: Aug 24, 2026
Transcranial Direct Current Stimulation for Online Gamers
Published on: November 9, 2019
Ender Özbek1, Bekir Çar2, Ahmet Kurtoğlu3
1Department of Travel, Tourism and Entertainment Services, Diyarbakır School of Social Sciences, Dicle University, Diyarbakır, Türkiye.
Aim:
The main objective of this study is to examine students' attitudes toward physical education (PEA) and sports classes within the framework of digital game addiction (DGA) and values education (VE).
Methods:
The research was conducted based on a quantitative and relational survey model. The study group consisted of 1,012 students (507 male, 505 female) aged 14-18 who were enrolled in private high schools during the 2023-2024 academic year. Participants were selected on a voluntary basis, and data were collected through an online survey. The data collection tools used were the DGA Scale, the PEA Scale, and the VE Scale. The data were evaluated using both classical correlation analyses and machine learning algorithms.
Results:
Correlation analyses revealed no significant linear relationship between DGA, VE, and PEA (p > 0.05). In contrast, analyses conducted using machine learning models yielded high levels of accuracy. The model that best predicted DGA scores was the Hybrid GBM + RF algorithm (R2 = 0.90128, RMSE = 1.58326). Similarly, the Hybrid GBM + RF model also showed the highest performance in predicting PEA scores (R2 = 0.90215, RMSE = 1.61893). According to interaction models, the relationship between VE and DGA varies depending on the student's PEA; the protective effect of VE becomes more pronounced in individuals with high attitudes. Similarly, in individuals with high levels of DGA, the effect of VE on PEA weakens or reverses.
Conclusion:
It demonstrates that traditional linear analytical approaches may be limited in explaining the complex and multidimensional relationships between variables; conversely, machine learning models are able to reveal patterns between DGA, VE, and PEA in a more comprehensive manner. This result suggests that values education may serve as a significant explanatory variable in understanding negative trends associated with DGA, and that students' PEA should be assessed within this multi-variable framework.