{"id":"W1729977867","doi":"","title":"Beyond the Scores: Using Candidate Responses on High Stakes Performance Assessment to Inform Teacher Preparation for English Learners.","year":2009,"lang":"en","type":"article","venue":"University of Washington Tacoma Digital Commons (University of Washington Tacoma)","topic":"Multilingual Education and Policy","field":"Social Sciences","cited_by":43,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Pedagogy; Psychology; Mathematics education; Variety (cybernetics); Teacher education; Sociology; Teacher preparation; Quarter (Canadian coin); History; Archaeology; Mathematics","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.000817681,0.0002486606,0.0003916434,0.0004689829,0.002131372,0.0002082509,0.001206056,0.0001846212,0.00009627357],"category_scores_gemma":[0.0003791644,0.0002835307,0.0002272423,0.000689025,0.0005096477,0.001753679,0.0001598561,0.0003321219,0.000008760648],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005044178,"about_ca_system_score_gemma":0.0006713157,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00494195,"about_ca_topic_score_gemma":0.002936019,"domain_scores_codex":[0.9980354,0.0002180175,0.0002535509,0.0003675573,0.0006567132,0.0004688137],"domain_scores_gemma":[0.9977716,0.0004844886,0.0004919716,0.0005024807,0.0004730979,0.0002763421],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.007577365,0.002283737,0.03939461,0.0002285915,0.0004574694,0.0000194707,0.7078262,0.006574169,0.00173079,0.1148847,0.02961004,0.08941287],"study_design_scores_gemma":[0.004573255,0.003370725,0.1139339,0.0005101638,0.0003433788,0.000003213422,0.4634492,0.006442487,0.001156544,0.001312241,0.4033474,0.001557461],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9501374,0.000009331572,0.0003788378,0.004185435,0.0002093051,0.0007198137,0.0002132683,0.0001111753,0.04403548],"genre_scores_gemma":[0.9912489,0.00002382427,0.001677559,0.0002128129,0.00009371778,1.699636e-7,0.00008286785,0.00001407642,0.006646063],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3737373,"threshold_uncertainty_score":0.9999617,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03961253567595996,"score_gpt":0.3355352411582134,"score_spread":0.2959227054822534,"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."}}