A new study titled “Mental Health, Well‑Being, and Adolescent Extremism: A Machine Learning Study on Risk and Protective Factors” offers insights into the dynamics between mental health, well-being, and extremist attitudes among adolescents.
Utilising an advanced machine learning approach on a nationally representative dataset of 11,397 Norwegian adolescents, the study addresses three questions: the possibility of distinguishing adolescents with extremist attitudes using psycho-socio-environmental variables, identifying the primary predictors of these attitudes, and understanding how these predictors cluster into latent factors.
The findings reveal that 17.6% of the adolescents exhibit elevated levels of extremist attitudes, with a notable prevalence among boys and younger adolescents. The machine learning model achieved a commendable accuracy (AUC of 76.7%), identifying key predictors such as positive parenting, quality of relationships with parents and peers, externalising behaviour, and overall well-being.
This extensive analysis of 550 variables marks a significant advancement in understanding adolescent extremism. By exploring various aspects of adolescents’ lives, including family and school environment, lifestyle choices, and mental health, the study provides a holistic view of the factors contributing to extremism. Notably, the study emphasises the importance of positive parenting and quality interpersonal relationships as protective factors against extremism.
The policy implications of these findings are profound. Interventions aimed at promoting positive parenting, enhancing the quality of peer relationships, and addressing externalising behaviours among adolescents could significantly reduce the susceptibility to extremist attitudes. This research underscores the potential of machine learning in aiding the identification of at-risk individuals and guiding targeted intervention strategies.