Reproducibility Package for: Disentangling Service Participation and Hedonic Positioning in Online Reviews — A Structural Confound in Platform Data
收藏资源简介:
This record contains the complete reproducibility package for the manuscript "Disentangling Service Participation and Hedonic Positioning in Online Reviews: A Structural Confound in Platform Data" (under review, Journal of Theoretical and Applied Electronic Commerce Research). Contents: (i) the full independent coding system used to build the study's measurement instruments — written, hypothesis-blind codebooks, coding prompts, every rater output (two large-language-model raters from different model families, a blind human-coded anchor, and blind human adjudication of all contested items), and reliability/validity analysis code; (ii) all 1,302 item weights of the Business Participation Index, tabulated pre- and post-adjudication, with business-level index and positioning scores; (iii) the complete NLP lexicons, regular expressions, and feature-extraction rules; (iv) analysis code for every model, robustness, balancing, and sensitivity analysis reported in the manuscript, with fixed random seeds and package versions; and (v) a deterministic rebuild script that reproduces the analysis panel from the public Yelp Open Dataset. Per the Yelp Dataset Terms of Use, no review-level content is redistributed; DATA_ACCESS.md documents how to obtain the public dataset and rebuild all analysis data deterministically. Start with README.md.



