Measuring Age-Related Differences in Cognitive Bias by Developing and Validating a Scenario-Based Questionnaire Using Deep Learning

This study examines the role of cognitive biases in decision-making and addresses the lack of a comprehensive, psychometrically robust assessment tool. Existing instruments often have limited scope and reliability. To fill this gap, a scenario-based questionnaire was developed, based on the dual-system model of cognition, which emphasizes the interaction between intuitive and analytical thinking. Age related differences in decision-making were also explored. Methods: A systematic review identified 230 cognitive biases, refined to 79 items through expert Q-sort content validation and factor analyses (EFA and CFA). A pilot study and formal testing with participants from various age groups assessed the tool’s psychometric properties. The Cognitive Reflection Test (CRT), Decision-Making Competence, and Decision Outcome Inventory (DOI) were used for cross-validation. Deep learning methods classified and validated the scale’s structure. Results: Machine learning and deep learning identified five factors—Belief, Social, Memory, Information, and Calculation—explaining 75% of variance. The scale demonstrated high reliability (Cronbach’s alpha > 0.7) and construct validity. Notably, the “sunk cost” bias was sensitive to age-related differences. Conclusion: This study introduces a novel scenario-based tool for assessing cognitive biases, integrating deep learning to enhance validity. It provides significant insights into cognitive bias structures and decision-making, advancing future research.