“There Is No Wrong Time to Do the Right Thing”: Examining Factors Influencing the Offender’s Decision to Confess to Fraud and the Victim’s Decision to Refer to Authorities

ABSTRACT This study employs data mining techniques to investigate two important aspects of fraud cases: the likelihood of perpetrators confessing before discovery and the propensity for victim organizations to refer cases to law enforcement. Utilizing data from the ACFE, we apply Random Tree and Naive Bayes algorithms to develop predictive models. Although our findings reveal challenges in predicting pre-discovery confessions, where the best model achieved only 16.33 percent accuracy in identifying true confessions, we present a more promising model for predicting law enforcement referrals, with 66.34 percent accuracy and correct classification of 74.84 percent of actual referrals. The study identifies key demographic, situational, and behavioral factors that influence referrals and confessions. By expanding theoretical frameworks beyond the fraud triangle and addressing gaps in managerial and nonfinancial variables in fraud detection models, this research contributes important insights for fraud examiners and organizations and improves our understanding of post-fraud decision-making processes. Data Availability: Data are available from the ACFE Research Institute. Restrictions apply to the availability of these data, which were used under license for this study. JEL Classifications: C55; G41; K42; M42.