Repository for "Grounded Conversational AI in Mixed Reality: Calibrated Trust in a Cultural Heritage Deployment"

Repository accompanying the manuscript "Grounded Conversational AI in Mixed Reality: Calibrated Trust in a Cultural Heritage Deployment" (submitted to the International Journal of Human-Computer Studies), a voice-driven Meta Quest 3 museum guide deployed at the Zoological Museum "Cambria", University of Messina. This deposit contains the study protocol and statistical analysis plan (written before data collection; analyses added afterwards are marked as post hoc), participant-facing materials (informed consent, demographic questionnaire, researcher script, post-session questionnaire, session sheet), R analysis scripts (calibration model, design-sensitivity simulation, inter-rater reliability, with a synthetic-data smoke test), a table of the aggregate results reported in the paper, the pre-generated area-visit-order schedule, an implementation excerpt for the specimen-assignment balancing procedure, and the specimen knowledge records used by the LLM. The collected response-level dataset is deliberately not included, given the re-identification risk of a small (P = 20), university-affiliated convenience sample. The reported aggregate results are tabulated for checking, but the deposited code requires the response-level data, which are available from the corresponding author to qualified researchers upon signature of a data-use agreement. The system prompt is available from the corresponding author on request.