Data for "When the Source Is a Machine: Source Type Conditions the Persuasive Weight of Parasocial Interaction in Healthy-Eating Communication"

This record contains the de-identified participant-level data, codebook and analysis code for a between-subjects experiment (N = 412) conducted in Türkiye. Participants viewed an identical social-media-style healthy-eating message attributed either to a human nutrition influencer (n = 206) or, through a disclosure badge, to an AI-generated virtual influencer (n = 206), and then completed Turkish-language measures of perceived expertise, trustworthiness, parasocial interaction, healthy-eating intention and AI literacy (5-point Likert scales). Data were collected online in May–June 2026. Files:- study_data_anonymized.csv: participant-level data (35 variables, no missing values)- codebook.xlsx: item wordings (Turkish and English), coding, reverse-coded items, scale scoring, descriptive statistics and study metadata- analysis_lavaan.R: R (lavaan) script reproducing the analyses reported in the paper- README.txt: study overview, data collection, ethics, file descriptions and citation The study was approved by the Selçuk University Faculty of Tourism Scientific Ethics Evaluation Committee (2026-05-04, decision no. 195/02). The data underlie a manuscript under review at the International Journal of Human-Computer Interaction.