{"id":"W4415779721","doi":"10.33423/jabe.v27i5.7857","title":"Approaching Human-Level Data Coding? A Systematic Comparison of OpenAI, DeepSeek, and Human Coders in Qualitative Analysis of Customer Reviews","year":2025,"lang":"","type":"article","venue":"Journal of Applied Business and Economics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Coding (social sciences); Qualitative analysis; Qualitative research; Scalability; Qualitative property","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01085571,0.0002415792,0.003282888,0.001333552,0.0003088216,0.0001519892,0.0006167154,0.0001258862,0.00002100874],"category_scores_gemma":[0.0002667757,0.0002135655,0.000207768,0.001982766,0.0004455871,0.0004069586,0.0003232837,0.0002542567,2.93561e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001090789,"about_ca_system_score_gemma":0.0001939262,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009727816,"about_ca_topic_score_gemma":0.002787838,"domain_scores_codex":[0.995269,0.0006082142,0.003362584,0.0003614716,0.0002001256,0.0001986198],"domain_scores_gemma":[0.993863,0.00107324,0.004344387,0.0002997253,0.000333616,0.00008602076],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.0005866345,0.002200076,0.03871265,0.04879981,0.01908732,0.000003523088,0.2673581,0.08949818,0.001006664,0.511172,0.0002699679,0.02130511],"study_design_scores_gemma":[0.006161178,0.000185398,0.09897821,0.01631631,0.03891791,0.000004835905,0.4645339,0.3311657,0.0001100783,0.04097092,0.0009918523,0.001663727],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9689824,0.004491,0.02327553,0.0002058014,0.0001198695,0.0005993194,0.00005010107,0.000001758546,0.002274211],"genre_scores_gemma":[0.9931206,0.001916745,0.004822548,0.0000355028,0.00003716975,0.000004992053,0.00002330241,0.000007506198,0.00003160186],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4702011,"threshold_uncertainty_score":0.8708948,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2436110177352338,"score_gpt":0.4710197405948601,"score_spread":0.2274087228596263,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}