Dataset for How AI Advice Helps and Harms Judgment Revision

This dataset accompanies a five-group online experiment involving 540 participants. The study examines how standardized AI-framed feedback influences the revision of existing judgments. The five experimental conditions include independent review, direct recommendation, emotional support, reflective scaffolding, and combined support. The research focuses on judgment improvement, deterioration, advice adoption, subjective confidence, and cognitive offloading.