Anonymized Dataset for "Beyond Trust Increase: Explainability and Accuracy Feedback for Calibrated Human–AI Reliance"

This dataset contains anonymized survey responses from a between-subjects online experiment examining the effects of feature-based explanations and explicit accuracy feedback on trust change, trust calibration, and intended reliance in AI-assisted decision-making. The final sample includes 310 adult participants distributed across three conditions: Control (n = 100), Explanation (n = 105), and Explanation + Feedback (n = 105). The dataset includes pre-task trust, post-task trust, perceived AI accuracy, normalized post-task trust, calibration gap, trust change, willingness to use AI, and selected demographic and AI familiarity variables. All data are anonymized and contain no personally identifiable information.