Assessing the Relationship Between Auditor Professional Skepticism and Fraud Detection Efficiency

Posted: Apr 01, 2021

• Micah Allen
• Gemma Cox
• Kai Mitchell

Abstract

This research investigates the complex relationship between auditor professional skepticism and fraud detection efficiency through a novel computational framework that integrates behavioral psychology with machine learning analytics. Traditional auditing research has largely relied on survey-based methodologies and self-reported measures of professional skepticism, creating significant limitations in objectively quantifying this multidimensional construct and its operational impact. Our study introduces an innovative approach by developing a computational skepticism index derived from natural language processing of audit documentation and behavioral analysis of auditor decision-making patterns. We employ a hybrid methodology combining experimental audit simulations with advanced machine learning algorithms to analyze how varying levels of professional skepticism manifest in audit efficiency metrics across different fraud scenarios. The research examines 342 audit engagements from multiple international accounting firms, capturing both quantitative efficiency measures and qualitative skepticism indicators. Our findings reveal a non-linear relationship between professional skepticism and fraud detection efficiency, challenging the conventional assumption that higher skepticism uniformly improves audit outcomes. Specifically, we identify an optimal skepticism threshold beyond which additional skepticism yields diminishing returns and may even impair efficiency through excessive verification activities and analysis paralysis. The study further demonstrates that the effectiveness of professional skepticism is significantly moderated by contextual factors including time pressure, client relationship dynamics, and regulatory environment. These insights contribute to auditing theory by providing a more nuanced understanding of how professional skepticism operates in practice and offer practical implications for audit training programs and firm quality control systems. Our computational approach represents a paradigm shift in auditing research methodology, enabling more precise measurement and analysis of psychological constructs that have previously resisted objective quantification.