{"id":"W3190138084","doi":"10.1093/jssam/smab022","title":"A Model-Assisted Approach for Finding Coding Errors in Manual Coding of Open-Ended Questions","year":2021,"lang":"en","type":"article","venue":"Journal of Survey Statistics and Methodology","topic":"Reliability and Agreement in Measurement","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"ca_institutions":"Actua; University of Waterloo; Roche (Canada)","funders":"","keywords":"Coding (social sciences); Computer science; Statistics; Natural language processing; Artificial intelligence; Mathematics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.05333626,0.0001081331,0.0007419871,0.0002874966,0.0001096491,0.0001161651,0.0004235581,0.00008699611,0.00002047213],"category_scores_gemma":[0.04390118,0.00008432916,0.00006985809,0.000393866,0.0001094628,0.0001661999,0.0001709918,0.0002289664,1.308047e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005217147,"about_ca_system_score_gemma":0.0002605016,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008526182,"about_ca_topic_score_gemma":0.0004676698,"domain_scores_codex":[0.9933288,0.004125554,0.001484425,0.0002742765,0.0005778907,0.0002090708],"domain_scores_gemma":[0.9816191,0.0158072,0.0009172976,0.0001841605,0.001378087,0.00009420297],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.003986306,0.002708041,0.3685445,0.0007042751,0.00070526,0.0001646069,0.007477638,0.1011885,0.1110697,0.211837,0.02697876,0.1646354],"study_design_scores_gemma":[0.003970564,0.0006481734,0.5165039,0.0002061922,0.000138965,0.0001586504,0.004695265,0.1904042,0.002221254,0.2801836,0.0005090113,0.0003602097],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.07135586,0.0001406674,0.9273774,0.0002445166,0.0002861363,0.0001649582,0.0002880419,0.000001053166,0.0001413317],"genre_scores_gemma":[0.337338,0.0001228901,0.6623521,0.00005782419,0.00001475736,0.00000358836,0.00001034115,0.000004355226,0.00009619544],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2659821,"threshold_uncertainty_score":0.9747896,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.7870611956524755,"score_gpt":0.5441244124235828,"score_spread":0.2429367832288927,"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."}}