Data for "Paying for the human touch: How human versus AI digital actors influence purchase intention in short dramas"

This dataset accompanies the manuscript "Paying for the human touch: How human versus AI digital actors influence purchase intention in short dramas" by Xinzhi Chen. It contains three separate English-language Excel workbooks, each with a Data sheet and a Codebook describing variables, response coding and data preparation. Study 1 (Study_1_Data_English.xlsx): 712 platform poll votes, comprising 319 votes in the human-actor condition and 393 in the AI digital-actor condition. The workbook includes aggregate frequencies and 712 analysis rows expanded from those frequencies. These rows are not original respondent-level platform records. The outcome is stated willingness to pay, not an observed purchase. Use either the unweighted Data sheet or the Frequencies sheet weighted by frequency, without combining the two representations. Study 2 (Study_2_Data_English.xlsx): 250 final analysis cases. Variables include actor type, purchase intention (PI1-PI5), perceived verisimilitude (PV1-PV6), narrative transportation (NT1-NT4), baseline emotional state (BES1-BES4), short-drama viewing frequency in the past three months, sex and age. The Study 2 covariates are the mean baseline emotional-state score and viewing frequency. Study 3 (Study_3_Data_English.xlsx): 420 final analysis cases. Variables include actor type, story genre, PI1-PI5, PV1-PV6, NT1-NT4, sex and age. Sex and age are the Study 3 covariates. Two source frequency variables not used in the manuscript were omitted from this shared workbook. Coding: actor_type is 0 for AI digital actors and 1 for human actors; story_genre is 0 for science fiction and 1 for contemporary stories; sex is 0 for female and 1 for male; age is reported in completed years. Study 1 pay is 0 for unwilling and 1 for willing to pay. Scale items in Studies 2 and 3 use seven-point responses, with no reverse-scored items. Construct scores are arithmetic means of their respective items; composite scores are not included as additional data columns. Retained numerical values and row order are unchanged from the supplied analysis workbooks. Headers, labels and worksheet names were translated or standardized in English. Full item wording and scale endpoints are provided in each Codebook.