Predictors of fraud victimization among the older adults in China: A machine learning analysis

Abstract Purpose Fraud victimization among older adults is a growing concern, yet the factors influencing susceptibility remain underexplored. This study examines the predictive role of individual factors in fraud victimization among 5499 older adults in China using machine learning (ML) techniques. Methods Data were drawn from a 2015 dataset, including 22 demographic, financial, and psychological variables. Random forest models were applied to identify key risk factors contributing to fraud susceptibility. Results Random forest models showed that demographic characteristics, particularly income, consumption, education, and age, are the most significant predictors of fraud victimization. Additionally, factors, such as chronic illness, lack of pension benefits, financial knowledge, and psychological traits like risk appetite and generalized trust also influence victimization risk. Moreover, psychological traits, although predictive, exhibit weaker explanatory power than demographic variables in this study. Additional ML analyses showed that fraud type influences victimization, with telephone and SMS fraud associated with lower risk, while fraud by acquaintances in person increases susceptibility. Conclusions These results suggest the necessity of multifaceted prevention strategies, including financial education programmes, targeted policy interventions, and social support mechanisms to mitigate fraud risks among older adults.