Replication Data for 'Convergence without independence: Validating large language model party placements against a local expert panel and the Chapel Hill Expert Survey in Türkiye'

This dataset supports the replication of a validation study examining whether ChatGPT and Grok can accurately estimate the ideological positions of Turkish political parties, benchmarked against expert survey data and the Chapel Hill Expert Survey 2024. A Readme-Codebook sheet provides full variable definitions, party abbreviations, CHES 2024 variable mapping, and methodology notes. All analyses - including MAE, Pearson r, Spearman ρ, and systematic bias - can be replicated from these panels using standard statistical software.