Application of Causal Machine Learning to Compute the effects of video game intensity on students' academic performance and psychosocial well-being

Abstract

This study applies Double/Debiased Machine Learning (DML) to estimate the quasi-causal effects of video game intensity on students’ academic performance and psychosocial well-being using survey data from university students in Kazakhstan, Kyrgyzstan, and Uzbekistan. While prior research often reports negative correlations between intensive gaming and academic outcomes, such associations may reflect confounding and self-selection rather than causal effects. Using a causal machine learning framework that combines flexible nuisance estimation with orthogonalized treatment effect estimation, the analysis provides more credible causal inference under conditional unconfoundedness. The results indicate that the negative association between gaming intensity and academic performance is largely explained by confounding rather than a direct causal effect. However, gaming shows a consistent negative effect on sleep quality, suggesting sleep disruption as a key mechanism influencing student well-being. These findings demonstrate the importance of causal inference methods in digital games research.