{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":31,"total_is_capped":false,"direct_labels_cover":1,"predictions_cover":31,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"9fdb53ddc261","filters":{"venue":"The Stata Journal Promoting communications on statistics and Stata"}},"results":[{"id":"W3013330736","doi":"10.1177/1536867x20909688","title":"The random forest algorithm for statistical learning","year":2020,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Financial Distress and Bankruptcy Prediction","field":"Business, Management and Accounting","cited_by":1264,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Random forest; Computer science; Key (lock); Artificial intelligence; Credit card; Machine learning; Algorithm; Statistical learning; World Wide Web","authors":[{"name":"Matthias Schonlau","is_ca":true},{"name":"Rosie Yuyan Zou","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03533706308133225,"gpt":0.2758883826175839,"spread":0.2405513195362517,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001296731,0.0001675908,0.0001910193,0.00003981505,0.004009705,0.001255985,0.0008348807,0.00003133359,0.00001772114],"category_scores_gemma":[0.002028254,0.0001018539,0.00004445439,0.0001673219,0.0003056511,0.0002764711,0.0003803409,0.0006444007,0.00002066402],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001834812,"about_ca_system_score_gemma":0.00003925151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006417658,"about_ca_topic_score_gemma":0.00007038097,"domain_scores_codex":[0.9986544,0.0001054005,0.0004729895,0.0001789982,0.0002762469,0.0003119517],"domain_scores_gemma":[0.9969028,0.001893851,0.0004181362,0.0004306064,0.0003120439,0.00004255335],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001971397,0.00006918341,0.0003876244,0.00006741741,0.00006969002,0.000004938136,0.0004423705,0.0001375085,0.000009247097,0.3407848,0.0261665,0.6316636],"study_design_scores_gemma":[0.001172703,0.0001408239,0.001797341,0.0000486403,0.0001041884,0.000008037396,0.0008484778,0.6957911,8.829263e-7,0.04200962,0.2579222,0.0001559647],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0009814247,0.0004972299,0.9827126,0.01358713,0.0001758372,0.0005379662,0.0007531315,0.00005087415,0.0007037565],"genre_scores_gemma":[0.9093821,0.003648618,0.08063556,0.002851826,0.001737353,0.0001360006,0.001266983,0.0001080572,0.0002335019],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9084007,"threshold_uncertainty_score":0.9997808,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2891445448","doi":"10.1177/1536867x19830877","title":"Fast and wild: Bootstrap inference in Stata using boottest","year":2019,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Data Analysis with R","field":"Computer Science","cited_by":884,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University; Queen's University","funders":"","keywords":"Inference; Econometrics; Statistics; Computer science; Mathematics; Artificial intelligence","authors":[{"name":"David Roodman","is_ca":false},{"name":"Morten Ørregaard Nielsen","is_ca":true},{"name":"James G. MacKinnon","is_ca":true},{"name":"Matthew D. Webb","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06334283729434241,"gpt":0.3511636108219209,"spread":0.2878207735275785,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001959521,0.0002308226,0.0003172113,0.0002630412,0.0007136814,0.001139268,0.002805403,0.00003867674,0.00001605197],"category_scores_gemma":[0.0005480182,0.0001774549,0.00002442985,0.0004842229,0.0002828952,0.0008347277,0.001753386,0.0008887734,0.00001394432],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006524893,"about_ca_system_score_gemma":0.0002038323,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002543184,"about_ca_topic_score_gemma":0.0002653206,"domain_scores_codex":[0.9975875,0.00054481,0.0006649462,0.0003754748,0.0004156946,0.0004115687],"domain_scores_gemma":[0.994554,0.00201321,0.0004739452,0.00260377,0.0001820492,0.0001729937],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00008058441,0.0009111221,0.07460064,0.0002028084,0.0003527928,0.0001062096,0.01841531,0.002693555,0.0007026058,0.5584456,0.001658644,0.3418301],"study_design_scores_gemma":[0.0008486463,0.0003483066,0.01741587,0.0003212553,0.00005174001,0.0001621329,0.001004413,0.9475124,0.00001308508,0.02913528,0.002781559,0.0004052794],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.131603,0.001166612,0.8611724,0.003642075,0.0001333081,0.0005595923,0.001188513,0.00005027289,0.0004842386],"genre_scores_gemma":[0.8264369,0.001826464,0.1713376,0.0002277897,0.00001417356,0.00000413421,0.00007820636,0.00002003994,0.00005472962],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9448189,"threshold_uncertainty_score":0.9998977,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1566635326","doi":"10.1177/1536867x0900900202","title":"Updated Tests for Small-study Effects in Meta-analyses","year":2009,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":323,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Institute of Infection and Immunity","funders":"","keywords":"Funnel plot; Publication bias; Statistics; Meta-analysis; Test (biology); Statistical hypothesis testing; Mathematics; Econometrics; Medicine; Confidence interval; Internal medicine","authors":[{"name":"Roger Harbord","is_ca":false},{"name":"Ross Harris","is_ca":false},{"name":"Jonathan A C Sterne","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.8746962574195007,"gpt":0.6167479719745995,"spread":0.2579482854449012,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.09170786,0.0003622728,0.002589923,0.0004726717,0.001121889,0.002028911,0.004360756,0.00004015696,0.0002402187],"category_scores_gemma":[0.04480632,0.0001542838,0.0005628496,0.001092961,0.0001226602,0.0001864547,0.0003229472,0.0006028352,0.00005261872],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000370132,"about_ca_system_score_gemma":0.00009130833,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003179429,"about_ca_topic_score_gemma":0.0003212384,"domain_scores_codex":[0.9765521,0.01455799,0.006061588,0.0005826675,0.001883211,0.0003625046],"domain_scores_gemma":[0.9637128,0.02462295,0.003930758,0.006301098,0.001252598,0.0001797895],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002584726,0.006402273,0.01204066,0.0001747764,0.015849,0.0001199219,0.02633666,0.001778464,0.000406097,0.1603175,0.1972268,0.5790894],"study_design_scores_gemma":[0.003110711,0.003281508,0.074917,0.0001451232,0.01542086,0.0001623025,0.007763327,0.2253286,0.00002561226,0.6269671,0.04188331,0.0009944753],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1421778,0.009732588,0.8111012,0.02316651,0.0002697339,0.009112092,0.00223067,0.00002912415,0.002180279],"genre_scores_gemma":[0.8831125,0.0001876059,0.1154245,0.0004647684,0.00002431023,0.00006534654,0.00007464836,0.00001741846,0.0006288171],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7409347,"threshold_uncertainty_score":0.9990071,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2585958134","doi":"10.1177/1536867x1601600407","title":"Support Vector Machines","year":2016,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Face and Expression Recognition","field":"Computer Science","cited_by":218,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Categorical variable; Support vector machine; Multinomial distribution; Computer science; Multinomial logistic regression; Artificial intelligence; Binary classification; Statistical learning; Relevance vector machine; Machine learning; Binary number; Data mining; Mathematics; Statistics; Arithmetic","authors":[{"name":"Nick Guenther","is_ca":true},{"name":"Matthias Schonlau","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03801075982839052,"gpt":0.3092633223698961,"spread":0.2712525625415055,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009921005,0.0001288122,0.0001296385,0.0000783691,0.001020461,0.000386097,0.001780736,0.00002496387,0.00006362639],"category_scores_gemma":[0.0003901407,0.00006634058,0.00002636493,0.0001275525,0.0001690999,0.0003687345,0.0006743449,0.0002966243,0.00006855529],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002415514,"about_ca_system_score_gemma":0.00008071874,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001297771,"about_ca_topic_score_gemma":0.00001365291,"domain_scores_codex":[0.9986445,0.0003282465,0.0003442095,0.0001766106,0.0002666394,0.0002397622],"domain_scores_gemma":[0.9968542,0.001218744,0.000245265,0.001369626,0.000181441,0.0001306829],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001854596,0.0001368881,0.0002102801,0.00001384064,0.00004458524,0.00001400225,0.001903348,0.000001504354,0.001197007,0.149156,0.03211351,0.8151905],"study_design_scores_gemma":[0.003805822,0.002064206,0.0109506,0.001101728,0.0001200941,0.001295267,0.0007079629,0.09348311,0.00144363,0.6166655,0.2670088,0.001353232],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.004316144,0.0002493501,0.9624788,0.0306635,0.0002133248,0.0001972793,0.0006857838,0.00006978323,0.001126002],"genre_scores_gemma":[0.7529537,0.004404204,0.2407927,0.0008626495,0.000075007,0.00001928244,0.00005066071,0.00002699283,0.0008148155],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8138372,"threshold_uncertainty_score":0.7848666,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3088227231","doi":"10.1177/1536867x20953572","title":"Causal mediation analysis in instrumental-variables regressions","year":2020,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":186,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Instrumental variable; Mediation; Causal inference; Econometrics; Mediator; Psychology; Causal analysis; Causal model; Variables; Estimation; Statistics; Mathematics; Medicine; Sociology; Economics; Social science; Internal medicine","authors":[{"name":"Christian Dippel","is_ca":false},{"name":"Andreas Ferrara","is_ca":false},{"name":"Stephan Heblich","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1940657655676138,"gpt":0.4336912814990895,"spread":0.2396255159314757,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001267243,0.0001783253,0.0003216113,0.0001861369,0.0006177296,0.0001794686,0.0008781159,0.00004780832,0.00006753988],"category_scores_gemma":[0.002773172,0.0001290113,0.00003787665,0.0006976724,0.0002173676,0.0002161373,0.0004633262,0.0008550542,0.000003170835],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007991783,"about_ca_system_score_gemma":0.00008836107,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004924916,"about_ca_topic_score_gemma":0.0002775393,"domain_scores_codex":[0.9979799,0.0005612726,0.0006865837,0.000195538,0.0003268168,0.0002499392],"domain_scores_gemma":[0.9956335,0.002618718,0.0005201558,0.0009178163,0.0001505502,0.0001592973],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00009708905,0.0004400637,0.008439199,0.00009685179,0.0005589162,0.000033917,0.01588289,0.0001834527,0.0006384958,0.9267418,0.006696343,0.04019093],"study_design_scores_gemma":[0.0007797037,0.0004163878,0.002981271,0.0002066255,0.0004022944,0.00003593552,0.003474199,0.1288637,0.0001576126,0.8602741,0.001996692,0.0004114998],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04410705,0.0003976573,0.9339352,0.01720043,0.00005551354,0.0007253061,0.00242127,0.0001482394,0.001009335],"genre_scores_gemma":[0.7301795,0.001694624,0.2675782,0.0002571064,0.00002519162,0.00001881272,0.0002035474,0.00002399501,0.00001896081],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6860725,"threshold_uncertainty_score":0.5260929,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3125007939","doi":"10.1177/1536867x1501500113","title":"A Robust Test for Weak Instruments in Stata","year":2015,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Monetary Policy and Economic Impact","field":"Economics, Econometrics and Finance","cited_by":168,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Heteroscedasticity; Econometrics; Estimator; Mathematics; Statistics; Homoscedasticity; Autocorrelation; Null hypothesis; Economics","authors":[{"name":"Carolin Pflueger","is_ca":true},{"name":"Su Wang","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2892050080069994,"gpt":0.3125236596126829,"spread":0.02331865160568347,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002906522,0.0001903712,0.0003597043,0.0002228486,0.0005453412,0.0003480962,0.001015862,0.00004817733,0.00003750018],"category_scores_gemma":[0.001585706,0.0001713452,0.00003997551,0.0001454326,0.0001838894,0.0003207768,0.0002923269,0.0005067845,0.00005390085],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001538698,"about_ca_system_score_gemma":0.00007581547,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005063177,"about_ca_topic_score_gemma":0.0002374518,"domain_scores_codex":[0.9982072,0.00009343737,0.0009585349,0.0002541809,0.00005954716,0.0004271318],"domain_scores_gemma":[0.9970446,0.0009463605,0.0006267775,0.00110657,0.00005283061,0.0002228674],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004039623,0.001688241,0.08923415,0.0001799026,0.0004294519,0.00001284444,0.02164557,0.004492391,0.00001006842,0.7365029,0.07658821,0.06881225],"study_design_scores_gemma":[0.0037442,0.0009462025,0.01091693,0.0001225691,0.0000303287,0.00009973767,0.002182927,0.4646135,0.000005429446,0.4031867,0.1135993,0.0005522048],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5390885,0.009767642,0.2854258,0.04896493,0.001457353,0.004527696,0.08365578,0.0001329511,0.0269793],"genre_scores_gemma":[0.9459536,0.002078393,0.05071521,0.0003745075,0.00006935281,0.00003437734,0.0002917158,0.00004140121,0.0004414505],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4601211,"threshold_uncertainty_score":0.6987255,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2896918891","doi":"10.1177/1536867x1801800209","title":"Qfactor: A Command for Q-methodology Analysis","year":2018,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Q Methodology Applications","field":"Decision Sciences","cited_by":51,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University","funders":"","keywords":"Computer science; Factor (programming language); Set (abstract data type); Quantitative analysis (chemistry); Statistical analysis; Qualitative analysis; Data mining; Statistics; Mathematics; Qualitative research; Programming language","authors":[{"name":"Noori Akhtar‐Danesh","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.5959575052896716,"gpt":0.5707345326609626,"spread":0.02522297262870898,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts"],"consensus_categories":[],"category_scores_codex":[0.02304197,0.0001981785,0.0005128447,0.0004957181,0.003310279,0.000573629,0.003407503,0.00007897025,0.0001685462],"category_scores_gemma":[0.02070577,0.0001267646,0.0001189877,0.001163932,0.001557186,0.0001549476,0.0007385669,0.0006557615,0.0000474813],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004036875,"about_ca_system_score_gemma":0.0001392746,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003993387,"about_ca_topic_score_gemma":0.0004790326,"domain_scores_codex":[0.9933745,0.004147692,0.001127514,0.0004082732,0.000553271,0.0003887373],"domain_scores_gemma":[0.942984,0.05121623,0.0009500835,0.003338656,0.001325926,0.000185059],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0003383726,0.0003929097,0.007889136,0.00001328192,0.001451652,0.000003213102,0.01533091,0.00007797179,0.0006535912,0.6199183,0.06590755,0.2880231],"study_design_scores_gemma":[0.0006476289,0.0005958945,0.01701711,0.00001330223,0.0006519508,0.00007822923,0.002373547,0.04990563,0.00008817646,0.8241676,0.1042054,0.0002555547],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01301015,0.0002537575,0.9737751,0.009500886,0.0001385161,0.0004501872,0.002179719,0.00002358053,0.0006681267],"genre_scores_gemma":[0.3354201,0.000310556,0.6630341,0.0005842384,0.0001003154,0.00004696618,0.00009381559,0.00002084849,0.0003889774],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.32241,"threshold_uncertainty_score":0.9979873,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2788623952","doi":"10.1177/1536867x1801700406","title":"Text Mining with n-gram Variables","year":2017,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":40,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"Social Sciences and Humanities Research Council of Canada; Universiteit van Tilburg; University of Waterloo","keywords":"n-gram; Computer science; Gram; Categorization; Process (computing); Text categorization; Natural language processing; Sequence (biology); Artificial intelligence; Language model; Programming language","authors":[{"name":"Matthias Schonlau","is_ca":true},{"name":"Nick Guenther","is_ca":true},{"name":"Ilia Sucholutsky","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0460079612838921,"gpt":0.321158498507218,"spread":0.2751505372233259,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001340424,0.0001653058,0.0001728829,0.00006359065,0.005596771,0.00297212,0.005000227,0.00002563623,0.000006441417],"category_scores_gemma":[0.0004306071,0.0001070665,0.00001804209,0.0001125478,0.0004475929,0.0005660434,0.00140044,0.000495941,0.000009972704],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002314581,"about_ca_system_score_gemma":0.0001382932,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009771233,"about_ca_topic_score_gemma":0.00005165714,"domain_scores_codex":[0.9986582,0.0001617026,0.0003338445,0.000258574,0.0002936679,0.0002939923],"domain_scores_gemma":[0.9939488,0.0009198689,0.0005760848,0.004188861,0.000218263,0.0001481651],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000009271261,0.0001566339,0.0003255789,0.00001353053,0.00008170773,0.0000143046,0.002541203,0.00002204602,0.00002309939,0.4389641,0.005948241,0.5519003],"study_design_scores_gemma":[0.001414859,0.0008440493,0.01327568,0.0004991471,0.0001186916,0.0007820473,0.001366394,0.8058097,0.00004400919,0.08010905,0.09501202,0.0007243751],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.001978051,0.0002012357,0.9847111,0.009860522,0.00007008961,0.000201636,0.0006078433,0.00004832127,0.002321274],"genre_scores_gemma":[0.2202826,0.0008589675,0.7783625,0.0001656641,0.00003987955,0.00002042162,0.00004889255,0.00001857632,0.0002026049],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.8057876,"threshold_uncertainty_score":0.9980629,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1525417239","doi":"10.1177/1536867x1501500111","title":"Estimating Net Survival using a Life-Table Approach","year":2015,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Colorectal Cancer Screening and Detection","field":"Medicine","cited_by":33,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Estimator; Statistics; Weighting; Survival analysis; Estimation; Inverse probability weighting; Relative survival; Table (database); Kaplan–Meier estimator; Biometrics; Life table; Mathematics; Computer science; Econometrics; Cancer registry; Medicine; Data mining; Cancer; Artificial intelligence; Population; Engineering","authors":[{"name":"Enzo Coviello","is_ca":false},{"name":"Karri Seppä","is_ca":false},{"name":"Paul W. Dickman","is_ca":false},{"name":"Arun Pokhrel","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1584187295514557,"gpt":0.3714735736883055,"spread":0.2130548441368497,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002132678,0.0001585568,0.0002686848,0.00009865644,0.0009972048,0.0002705188,0.0003827496,0.00003992826,0.000007374151],"category_scores_gemma":[0.001644646,0.0001150465,0.00002907995,0.0002595212,0.0001689383,0.0001265724,0.0002907375,0.0008058895,0.000003123879],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001135858,"about_ca_system_score_gemma":0.0004243141,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003291145,"about_ca_topic_score_gemma":0.00002124742,"domain_scores_codex":[0.9983483,0.0003681717,0.0004277262,0.0001796505,0.0004034912,0.0002726709],"domain_scores_gemma":[0.9975773,0.0004960657,0.0003166268,0.000889728,0.0003817569,0.0003385489],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.01626455,0.005074032,0.02671887,0.001603595,0.003216215,0.0002478935,0.07559291,0.04578732,0.001422175,0.09654212,0.04736504,0.6801653],"study_design_scores_gemma":[0.001283727,0.001231082,0.001014894,0.0001727705,0.0001771065,0.000616312,0.002219466,0.9863201,0.000009537065,0.004636948,0.002143523,0.0001745496],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1052724,0.001745787,0.8853974,0.001827118,0.0003489187,0.000548761,0.000368491,0.00008825334,0.004402908],"genre_scores_gemma":[0.6394992,0.0001190494,0.3599289,0.000129484,0.0001394955,0.000007094584,0.00007526491,0.00002699383,0.00007452405],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9405327,"threshold_uncertainty_score":0.7669794,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2117460710","doi":"10.1177/1536867x0500500408","title":"Speaking Stata: Smoothing in Various Directions","year":2005,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Soil Geostatistics and Mapping","field":"Environmental Science","cited_by":25,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Smoothing; Bivariate analysis; Diagonal; Mathematics; Computer science; Econometrics; Applied mathematics; Statistics; Geometry","authors":[{"name":"Nicholas J. Cox","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02836177399764022,"gpt":0.2954626371214416,"spread":0.2671008631238013,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.001846708,0.0002054959,0.0002121865,0.0001126502,0.001588274,0.0003902349,0.0009573366,0.00003824283,0.0001902544],"category_scores_gemma":[0.0006462202,0.0001645661,0.00002821588,0.0003305922,0.0003706011,0.0003063122,0.0007869329,0.0009511759,0.00006081306],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002264313,"about_ca_system_score_gemma":0.00004870019,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001456936,"about_ca_topic_score_gemma":0.004643188,"domain_scores_codex":[0.9979139,0.0003696287,0.000620562,0.0002682377,0.0003796168,0.0004480701],"domain_scores_gemma":[0.9971933,0.001247729,0.0003391195,0.001016395,0.00004577303,0.0001576676],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003604629,0.0004304861,0.006209326,0.00002178085,0.00006388473,0.0000387765,0.01713417,0.002515217,0.0001759076,0.03218257,0.007621987,0.9335698],"study_design_scores_gemma":[0.002042661,0.0004319138,0.0704039,0.0004275347,0.00014349,0.000538606,0.005807874,0.3438374,0.00003733099,0.08770551,0.4875488,0.001074939],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2383749,0.004316835,0.6702838,0.03448442,0.0009178404,0.002336702,0.003745682,0.0002858562,0.04525398],"genre_scores_gemma":[0.8295948,0.003547538,0.1660025,0.0004754484,0.0000670828,0.00001715193,0.00007199254,0.00004001874,0.0001834411],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9324949,"threshold_uncertainty_score":0.9997115,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4390051293","doi":"10.1177/1536867x231212433","title":"Leverage, influence, and the jackknife in clustered regression models: Reliable inference using summclust","year":2023,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Statistical Methods and Bayesian Inference","field":"Mathematics","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University; Queen's University","funders":"Social Sciences and Humanities Research Council of Canada; Danmarks Grundforskningsfond; National Research Foundation; York University","keywords":"Leverage (statistics); Jackknife resampling; Estimator; Inference; Computer science; Regression; Econometrics; Variance (accounting); Data mining; Linear regression; Cluster (spacecraft); Statistics; Mathematics; Artificial intelligence; Machine learning; Accounting","authors":[{"name":"James G. MacKinnon","is_ca":true},{"name":"Morten Ørregaard Nielsen","is_ca":false},{"name":"Matthew D. Webb","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1913882021901753,"gpt":0.4374128927053959,"spread":0.2460246905152206,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.004993115,0.0002447612,0.0004128337,0.0001853813,0.001527644,0.0004823711,0.001020863,0.0000599269,0.00001194436],"category_scores_gemma":[0.006136741,0.0001383075,0.00003006225,0.0005451878,0.0008345997,0.0002591678,0.0009852637,0.001107719,0.000003804033],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005318852,"about_ca_system_score_gemma":0.000140907,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002427588,"about_ca_topic_score_gemma":0.0001271539,"domain_scores_codex":[0.9967666,0.001305683,0.0008418346,0.0002577785,0.0004087455,0.0004193599],"domain_scores_gemma":[0.9846845,0.0131739,0.0004731685,0.001275528,0.0002595264,0.0001334087],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001569557,0.00008909035,0.0003936402,0.0001337538,0.00004650048,0.0000231125,0.00585781,0.0004117537,0.00002876177,0.9515722,0.001288279,0.03999813],"study_design_scores_gemma":[0.0007149573,0.00005024939,0.0008026368,0.000420757,0.00003532408,0.00004066189,0.0007223964,0.3884981,0.000002576488,0.6084638,0.0001277801,0.0001207332],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09946884,0.0005368846,0.8930007,0.004142016,0.000115709,0.0009120302,0.0009037806,0.00007019037,0.00084987],"genre_scores_gemma":[0.516767,0.005101875,0.4777456,0.0001740569,0.00002477673,0.00002345345,0.00002714727,0.00003807601,0.00009804067],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.4172982,"threshold_uncertainty_score":0.9997723,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4233989131","doi":"10.1177/1536867x1701700406","title":"Text Mining with n-gram Variables","year":2017,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Data Mining Algorithms and Applications","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"n-gram; Gram; Computer science; Statistics; Natural language processing; Mathematics; Geology","authors":[{"name":"Matthias Schonlau","is_ca":true},{"name":"Nick Guenther","is_ca":true},{"name":"Ilia Sucholutsky","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0460079612838921,"gpt":0.321158498507218,"spread":0.2751505372233259,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001340424,0.0001653058,0.0001728829,0.00006359065,0.005596771,0.00297212,0.005000227,0.00002563623,0.000006441417],"category_scores_gemma":[0.0004306071,0.0001070665,0.00001804209,0.0001125478,0.0004475929,0.0005660434,0.00140044,0.000495941,0.000009972704],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002314581,"about_ca_system_score_gemma":0.0001382932,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009771233,"about_ca_topic_score_gemma":0.00005165714,"domain_scores_codex":[0.9986582,0.0001617026,0.0003338445,0.000258574,0.0002936679,0.0002939923],"domain_scores_gemma":[0.9939488,0.0009198689,0.0005760848,0.004188861,0.000218263,0.0001481651],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000009271261,0.0001566339,0.0003255789,0.00001353053,0.00008170773,0.0000143046,0.002541203,0.00002204602,0.00002309939,0.4389641,0.005948241,0.5519003],"study_design_scores_gemma":[0.001414859,0.0008440493,0.01327568,0.0004991471,0.0001186916,0.0007820473,0.001366394,0.8058097,0.00004400919,0.08010905,0.09501202,0.0007243751],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.001978051,0.0002012357,0.9847111,0.009860522,0.00007008961,0.000201636,0.0006078433,0.00004832127,0.002321274],"genre_scores_gemma":[0.2202826,0.0008589675,0.7783625,0.0001656641,0.00003987955,0.00002042162,0.00004889255,0.00001857632,0.0002026049],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.8057876,"threshold_uncertainty_score":0.9980629,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2606619291","doi":"10.1177/1536867x1701700111","title":"Biasplot: A Package to Effective Plots to Assess Bias and Precision in Method Comparison Studies","year":2017,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Reliability and Agreement in Measurement","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Plot (graphics); Statistics; Computer science; Accuracy and precision; Data mining; Scatter plot; Mathematics","authors":[{"name":"Patrick Taffé","is_ca":false},{"name":"Mingkai Peng","is_ca":true},{"name":"Vicki Stagg","is_ca":false},{"name":"Tyler Williamson","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.6425845793989605,"gpt":0.5792506941073572,"spread":0.06333388529160333,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.02664722,0.0002212844,0.0005333433,0.0002904136,0.002931179,0.00184297,0.002724875,0.00003660729,0.000006908124],"category_scores_gemma":[0.03660919,0.0001334753,0.00003612901,0.0002947471,0.0003013283,0.0002950076,0.00221096,0.0006609109,0.00002440051],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001249129,"about_ca_system_score_gemma":0.00006048224,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001207723,"about_ca_topic_score_gemma":0.0007499783,"domain_scores_codex":[0.9940996,0.002808523,0.00108208,0.0004520993,0.001226511,0.0003312472],"domain_scores_gemma":[0.9796349,0.0157,0.0007480275,0.002899974,0.0007631395,0.0002539813],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0002737503,0.0004450355,0.02465296,0.00005018176,0.0001410786,0.00001523471,0.03486708,0.0004931497,0.000395698,0.01731669,0.008134353,0.9132148],"study_design_scores_gemma":[0.002562679,0.003372556,0.619732,0.002248355,0.0001824597,0.00007171032,0.05088929,0.07113797,0.0005865076,0.2258365,0.02245953,0.0009204246],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5223638,0.001136412,0.4499924,0.02234395,0.0003569507,0.002321517,0.0008434519,0.00002839104,0.0006130842],"genre_scores_gemma":[0.8300692,0.0006537357,0.1689374,0.0001930438,0.00002501576,0.00004676525,0.000004582099,0.00001483359,0.00005545561],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9122944,"threshold_uncertainty_score":0.9991932,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2810242367","doi":"10.1177/1536867x1801800109","title":"Fitting and Interpreting Correlated Random-coefficient Models Using Stata","year":2018,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Spatial and Panel Data Analysis","field":"Economics, Econometrics and Finance","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Estimator; Interpretation (philosophy); Random effects model; Estimation; Econometrics; Statistics; Computer science; Sample (material); Mathematics; Panel data; Economics","authors":[{"name":"Oscar Barriga Cabanillas","is_ca":false},{"name":"Jeffrey D. Michler","is_ca":true},{"name":"Aleksandr Michuda","is_ca":false},{"name":"Emilia Tjernström","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09588883221840087,"gpt":0.2989027927085462,"spread":0.2030139604901453,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.002741337,0.0002120451,0.0004039138,0.0002195452,0.001931823,0.0005963467,0.0008498408,0.00005149366,0.00007789349],"category_scores_gemma":[0.000843436,0.0001771537,0.00004645747,0.000294145,0.0004840099,0.0002951892,0.0006874999,0.0006060989,0.00001812043],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005930775,"about_ca_system_score_gemma":0.00004136054,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006007047,"about_ca_topic_score_gemma":0.00009436605,"domain_scores_codex":[0.9979904,0.0002234037,0.0009942313,0.0003291307,0.0001046325,0.0003582253],"domain_scores_gemma":[0.996666,0.00102677,0.0008665941,0.001097251,0.0001917042,0.0001516418],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001043422,0.0009878493,0.01255999,0.0003253539,0.002014086,0.0000504638,0.07739262,0.007168252,0.0003314317,0.7319311,0.005734382,0.1604611],"study_design_scores_gemma":[0.000750309,0.0001827489,0.0001880969,0.0001454202,0.00006697304,0.00006738531,0.0009414106,0.9444121,0.000007023882,0.05121971,0.00179408,0.0002247539],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0909538,0.003357092,0.8990952,0.001254002,0.0002002751,0.0003125165,0.003413536,0.00003106356,0.001382531],"genre_scores_gemma":[0.9618416,0.002464658,0.03514702,0.0002331765,0.00006068621,0.000004259507,0.0001505297,0.00003323721,0.00006484173],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9372438,"threshold_uncertainty_score":0.9993675,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1552664769","doi":"10.1177/1536867x1201200206","title":"Threshold Regression for Time-to-Event Analysis: The Stthreg Package","year":2012,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":12,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Centers for Disease Control and Prevention","keywords":"Regression analysis; Proportional hazards model; Computer science; Event (particle physics); Regression; R package; Path (computing); Statistics; Data mining; Mathematics; Machine learning; Programming language","authors":[{"name":"Tao Xiao","is_ca":false},{"name":"G. À. Whitmore","is_ca":true},{"name":"Xin He","is_ca":false},{"name":"Mei‐Ling Ting Lee","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1413015835246414,"gpt":0.4349213361843232,"spread":0.2936197526596817,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.005398488,0.0002446305,0.0003949827,0.0001287725,0.001885598,0.0003134318,0.001364456,0.00004751459,0.0001571503],"category_scores_gemma":[0.004215912,0.0001266077,0.00009922787,0.0004177802,0.0002701461,0.0001047351,0.000548673,0.0006579095,0.00003082458],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004547639,"about_ca_system_score_gemma":0.00005946316,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000009776473,"about_ca_topic_score_gemma":0.00001160038,"domain_scores_codex":[0.9973757,0.0007825071,0.0007129174,0.0001966985,0.0004221863,0.0005100436],"domain_scores_gemma":[0.986405,0.01054506,0.0005038035,0.001943175,0.0003189902,0.0002839517],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001135329,0.000507226,0.0002913969,0.00007563176,0.0008271782,0.000001814611,0.007307668,0.00001537226,0.0001718835,0.8373722,0.04054597,0.1127701],"study_design_scores_gemma":[0.0008513913,0.000748914,0.00502568,0.0003795972,0.002456806,0.0000544937,0.002051508,0.09470639,0.00009611926,0.861614,0.03140368,0.0006114215],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.005098415,0.0008096698,0.9824358,0.00716218,0.0001025773,0.0009212847,0.002542551,0.00003439945,0.000893174],"genre_scores_gemma":[0.2821216,0.000506649,0.715564,0.0005696774,0.0001440764,0.000100168,0.0001039309,0.00005703446,0.0008329261],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2770232,"threshold_uncertainty_score":0.9994138,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2154012562","doi":"10.1177/1536867x0600600407","title":"Mata Matters: Precision","year":2006,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Statistics Education and Methodologies","field":"Mathematics","cited_by":8,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Column (typography); Computer science; Point (geometry); Programming language; Base (topology); Quarter (Canadian coin); Matrix (chemical analysis); High-level programming language; Arithmetic; Programming paradigm; Mathematics; History; Telecommunications; Geometry","authors":[{"name":"William Gould","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2000500114887142,"gpt":0.4410708979479545,"spread":0.2410208864592403,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.002965704,0.0002218786,0.0002742959,0.0001372677,0.001312052,0.0004774175,0.001202716,0.00004846452,0.0001153996],"category_scores_gemma":[0.003017139,0.0001532952,0.00003981548,0.0001799481,0.0003775512,0.0001206337,0.0004391979,0.0006653617,0.00003147828],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006046748,"about_ca_system_score_gemma":0.0001045469,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006656929,"about_ca_topic_score_gemma":0.00005230536,"domain_scores_codex":[0.9973724,0.0009397231,0.0007451305,0.0002186063,0.0003889425,0.0003351465],"domain_scores_gemma":[0.9876533,0.009741194,0.0005571131,0.00164664,0.0002949695,0.0001068281],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00003629378,0.0002781042,0.0001842635,0.00005809498,0.00004103286,0.000006953481,0.001464699,0.000006524337,0.00008000005,0.6551277,0.3054844,0.03723195],"study_design_scores_gemma":[0.000435062,0.0001335962,0.00182819,0.0001333515,0.0000874674,0.0001881107,0.001270539,0.002791463,0.00005786088,0.9318357,0.06101059,0.0002280466],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.009058677,0.001015907,0.9583973,0.02286166,0.0006742434,0.000769155,0.003757598,0.0001421303,0.003323333],"genre_scores_gemma":[0.05870896,0.001599169,0.9368097,0.0006458158,0.000140755,0.00002841422,0.000253363,0.00006417366,0.001749635],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.276708,"threshold_uncertainty_score":0.9999881,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4313563060","doi":"10.1177/1536867x221141002","title":"qpair: A command for analyzing paired Q-sorts in Q-methodology","year":2022,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Q Methodology Applications","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University; Impact","funders":"","keywords":"Computer science; Factor (programming language); Statistical analysis; Data mining; Quantitative analysis (chemistry); Operations research; Statistics; Mathematics; Programming language","authors":[{"name":"Noori Akhtar‐Danesh","is_ca":true},{"name":"Stephen C. Wingreen","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.5272660379772833,"gpt":0.5212253227442247,"spread":0.006040715233058602,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.03787939,0.0001822138,0.0004528433,0.0004773774,0.003802014,0.0003133981,0.003247039,0.0000432504,0.0001719161],"category_scores_gemma":[0.01902633,0.0001242755,0.00006899323,0.0009449804,0.0005078492,0.0001395807,0.001498947,0.001229913,0.000005330709],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001068214,"about_ca_system_score_gemma":0.0002659592,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006424189,"about_ca_topic_score_gemma":0.0005369097,"domain_scores_codex":[0.988759,0.008477147,0.001322146,0.000406679,0.0006195728,0.0004154398],"domain_scores_gemma":[0.9365653,0.05975587,0.0008803953,0.002335934,0.0003381368,0.0001244087],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.001106066,0.0009095225,0.01310892,0.00003056428,0.0002295982,0.00004355797,0.02635845,0.002391822,0.000262294,0.3314149,0.05221507,0.5719292],"study_design_scores_gemma":[0.001441083,0.0007232292,0.009349731,0.00002646727,0.00007457596,0.0004637895,0.009445743,0.07048868,0.00001039477,0.8242432,0.08344225,0.0002908946],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02253089,0.0008920603,0.9499578,0.02252737,0.000205169,0.001053358,0.002545926,0.00003264405,0.0002547492],"genre_scores_gemma":[0.3673994,0.0005263629,0.630413,0.000707346,0.00003599783,0.0002957006,0.0001171281,0.00003344652,0.000471617],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.5716383,"threshold_uncertainty_score":0.9974949,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4362636314","doi":"10.1177/1536867x231161978","title":"Extended biasplot command to assess bias, precision, and agreement in method comparison studies","year":2023,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Reliability and Agreement in Measurement","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Statistics; Computer science; Limits of agreement; Accuracy and precision; Data mining; Mathematics; Nuclear medicine; Medicine","authors":[{"name":"Patrick Taffé","is_ca":false},{"name":"Mingkai Peng","is_ca":true},{"name":"Vicki Stagg","is_ca":false},{"name":"Tyler Williamson","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.6821565065636919,"gpt":0.5640156980598392,"spread":0.1181408085038527,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.03210633,0.0002326891,0.0005296866,0.0004745875,0.00158739,0.0008534907,0.001926187,0.00003679769,0.0000200374],"category_scores_gemma":[0.01159693,0.0001423396,0.00003851408,0.001092348,0.0003012477,0.0001915985,0.001696773,0.0006847375,0.00004360394],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001107587,"about_ca_system_score_gemma":0.00007938226,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005931568,"about_ca_topic_score_gemma":0.0003918984,"domain_scores_codex":[0.9929107,0.003031416,0.001574851,0.0004481004,0.001629408,0.0004054655],"domain_scores_gemma":[0.9803253,0.0162194,0.0005768546,0.001927694,0.0007368457,0.000213958],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002218873,0.0006836731,0.01766889,0.0000910969,0.0002499724,0.00002778284,0.03665916,0.002009287,0.0002446634,0.04566475,0.07490741,0.8215714],"study_design_scores_gemma":[0.00250304,0.001871654,0.1933587,0.001191663,0.0001643722,0.00007568775,0.1173106,0.2688134,0.0001709505,0.3535877,0.06012063,0.0008315455],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.511878,0.005431054,0.4156828,0.05961649,0.0008517563,0.003253567,0.001890371,0.0001412388,0.00125474],"genre_scores_gemma":[0.8281842,0.00560847,0.1653966,0.0003834131,0.00003846412,0.0000578263,0.00003810112,0.00002440761,0.0002685657],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8207399,"threshold_uncertainty_score":0.9997124,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4392957163","doi":"10.1177/1536867x241233671","title":"A Bayesian method for addressing multinomial misclassification with applications for alcohol epidemiological modeling","year":2024,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Statistical Methods and Bayesian Inference","field":"Mathematics","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Multinomial distribution; Computer science; Bayesian probability; Estimation; Statistics; Data mining; Econometrics; Data science; Machine learning; Artificial intelligence; Mathematics; Engineering","authors":[{"name":"William J. Parish","is_ca":false},{"name":"Arnie Aldridge","is_ca":false},{"name":"Martijn van Hasselt","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.3286245790278658,"gpt":0.4984497198034591,"spread":0.1698251407755933,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.004836833,0.0002756762,0.0004262522,0.0001186198,0.001750868,0.0006033144,0.000792834,0.00008512768,0.00001200824],"category_scores_gemma":[0.004211506,0.0001690565,0.00008250705,0.0001945478,0.0002730355,0.0001284208,0.0001445253,0.0006880577,0.00000144421],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007372252,"about_ca_system_score_gemma":0.0001845805,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001432318,"about_ca_topic_score_gemma":0.00001938402,"domain_scores_codex":[0.9973499,0.0007321535,0.0008611226,0.0004153616,0.0002276305,0.000413883],"domain_scores_gemma":[0.9714764,0.02659679,0.0003484675,0.0009862263,0.0004201518,0.0001719975],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00009242679,0.00009392743,0.000005647315,0.0002610988,0.000100986,0.000001279018,0.000653724,0.00006488447,0.00009508474,0.719839,0.00109225,0.2776997],"study_design_scores_gemma":[0.0003000744,0.0001784797,0.000007044931,0.0002120397,0.0001504579,0.00004821428,0.0002685721,0.5249945,0.000009825354,0.471069,0.002629155,0.0001326216],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00004577149,0.0004239711,0.9902198,0.004279708,0.00005659864,0.001752307,0.003001841,0.00007664217,0.0001433263],"genre_scores_gemma":[0.03402705,0.000288057,0.9644122,0.0001231021,0.0001432532,0.0007241725,0.0001791241,0.00006049196,0.00004254002],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.5249296,"threshold_uncertainty_score":0.9995487,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3194795045","doi":"10.1177/1536867x251341145","title":"netivreg: Estimation of peer effects in endogenous social networks","year":2025,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Game Theory and Applications","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Bank of Canada","funders":"","keywords":"Estimation; Endogeny; Computer science; Economics; Biology; Endocrinology","authors":[{"name":"Pablo Estrada","is_ca":false},{"name":"Juan Andrés Estrada","is_ca":false},{"name":"Kim P. Huynh","is_ca":true},{"name":"David T. Jacho‐Chávez","is_ca":false},{"name":"Leonardo Sánchez‐Aragón","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1103632264769011,"gpt":0.419005568832524,"spread":0.3086423423556229,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00628009,0.0001300479,0.0002766213,0.0002424208,0.001124223,0.0002956773,0.001647097,0.00004585353,0.00001822803],"category_scores_gemma":[0.00400061,0.00009024291,0.00004061179,0.0007536909,0.0004453701,0.0001224471,0.0004283479,0.0006098569,0.000006485106],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003669781,"about_ca_system_score_gemma":0.00011046,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003056684,"about_ca_topic_score_gemma":0.00004518866,"domain_scores_codex":[0.9970449,0.00115316,0.0008564302,0.000191295,0.0005342555,0.0002199567],"domain_scores_gemma":[0.9892048,0.008607615,0.0005359148,0.001139186,0.0004585361,0.00005400958],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00005740295,0.000251264,0.0003830216,0.00001918782,0.00003968778,0.000002871192,0.004430822,0.00350877,0.00008710708,0.6350558,0.003639811,0.3525243],"study_design_scores_gemma":[0.0006379387,0.0001283214,0.01045304,0.0001197459,0.00004970516,0.00002383551,0.001514365,0.2399759,0.00005158831,0.744028,0.002890508,0.0001270645],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04677633,0.0008218317,0.937114,0.009989002,0.0001354363,0.0006495833,0.0004257904,0.00001832346,0.004069721],"genre_scores_gemma":[0.9822839,0.0002287442,0.01699567,0.0001417458,0.00001685402,0.00002023102,0.00003869805,0.000008955754,0.0002651411],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9355077,"threshold_uncertainty_score":0.8646726,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1510648344","doi":"10.1177/1536867x1201200205","title":"Faster Estimation of a Discrete-Time Proportional Hazards Model with Gamma Frailty","year":2012,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of New Brunswick","funders":"","keywords":"Hessian matrix; Matrix (chemical analysis); Function (biology); Expression (computer science); Applied mathematics; Point (geometry); Algorithm; Gradient descent; Mathematics; Computer science; Mathematical optimization; Artificial intelligence; Geometry","authors":[{"name":"Michael G. Farnworth","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1220249435470352,"gpt":0.4053798658805197,"spread":0.2833549223334845,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002236601,0.0002061621,0.0003236555,0.00007968511,0.0006255211,0.0001231206,0.0006061138,0.00004327967,0.00008276314],"category_scores_gemma":[0.002183118,0.0001242379,0.00003306596,0.0001438619,0.0005034512,0.0002177035,0.0002655669,0.0005615265,0.000006123237],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003843048,"about_ca_system_score_gemma":0.0001591668,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000009633559,"about_ca_topic_score_gemma":0.000005039883,"domain_scores_codex":[0.9978246,0.0004658676,0.0007092897,0.0001452844,0.0005184564,0.0003365018],"domain_scores_gemma":[0.9949519,0.002828064,0.0006695907,0.001013171,0.0003518257,0.0001854394],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001341313,0.0005149379,0.0003206537,0.0001501711,0.0001514957,0.000001980051,0.004371083,0.0003661202,0.0001722275,0.9125851,0.00198858,0.07924347],"study_design_scores_gemma":[0.0004266139,0.0003395643,0.001146336,0.0002294927,0.0001464348,0.00007830376,0.0002939752,0.5847801,0.00006950057,0.4121266,0.0001625047,0.0002005332],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02198717,0.0001092457,0.9739979,0.000938116,0.00002559842,0.000336874,0.00163094,0.00001992484,0.0009542718],"genre_scores_gemma":[0.409324,0.00006455744,0.59036,0.00003639248,0.00001841099,0.00001309274,0.00005033014,0.00002196022,0.0001112899],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.584414,"threshold_uncertainty_score":0.5066274,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1508161314","doi":"10.1177/1536867x0800800406","title":"A Shortcut through Long Loops: An Illustration of Two Alternatives to Looping over Observations","year":2008,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Big Data Technologies and Applications","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Traffic Injury Research Foundation","funders":"","keywords":"Computer science; Search engine indexing; Key (lock); Mathematical optimization; Identifier; Data mining; Operations research; Information retrieval; Mathematics","authors":[{"name":"Ward Vanlaar","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.5569936812276851,"gpt":0.4802711286861797,"spread":0.07672255254150545,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.002183643,0.0001805117,0.0002844117,0.0001856535,0.001979442,0.0003814642,0.003018616,0.00004255505,0.00003980142],"category_scores_gemma":[0.002506578,0.0001238304,0.00004247481,0.0007911866,0.000667194,0.0007186542,0.0008382631,0.0005437444,0.00001099367],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003987458,"about_ca_system_score_gemma":0.0001437123,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001946813,"about_ca_topic_score_gemma":0.0004192762,"domain_scores_codex":[0.9970868,0.0004247653,0.00105718,0.0003385017,0.0008234432,0.0002693041],"domain_scores_gemma":[0.9933288,0.002327308,0.0007381004,0.002727944,0.0007450552,0.0001327731],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00006175563,0.0007577156,0.0128131,0.00001761279,0.0001393008,0.00001535611,0.01734987,0.001177134,0.001468638,0.7758566,0.01330376,0.1770392],"study_design_scores_gemma":[0.001613025,0.001397827,0.1396128,0.0003387981,0.0001376125,0.0004509469,0.01300691,0.1067881,0.0008237251,0.6995824,0.03535008,0.0008978127],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3396578,0.0002520564,0.6507815,0.006139236,0.00006390204,0.000461139,0.002304536,0.00003978638,0.0003000989],"genre_scores_gemma":[0.8022426,0.001562206,0.1955876,0.0002805714,0.00003579612,0.00002345279,0.0001480473,0.000016992,0.0001028154],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.4625848,"threshold_uncertainty_score":0.9993199,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1519629121","doi":"10.1177/1536867x0600600308","title":"Mata Matters: Interactive use","year":2006,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Probability and Statistical Research","field":"Mathematics","cited_by":1,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Column (typography); Computer science; Quarter (Canadian coin); Matrix (chemical analysis); Programming language; History; Archaeology; Telecommunications","authors":[{"name":"William Gould","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1496649167491727,"gpt":0.4148212657270012,"spread":0.2651563489778285,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.002295205,0.0002223402,0.0002884038,0.0001155645,0.001379968,0.0009154211,0.001295888,0.00004769321,0.0001049412],"category_scores_gemma":[0.004487814,0.0001542394,0.00004617005,0.0001894091,0.0006454626,0.0003520157,0.0007413717,0.001222963,0.00003103269],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001051809,"about_ca_system_score_gemma":0.0001066034,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002556878,"about_ca_topic_score_gemma":0.0002472168,"domain_scores_codex":[0.9971019,0.0009391921,0.00072895,0.0002484702,0.00053119,0.0004502769],"domain_scores_gemma":[0.9841202,0.0132909,0.000351437,0.001685819,0.0003906307,0.0001610465],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001088018,0.0005053935,0.0002863603,0.00008784908,0.00007371087,0.00002826669,0.001282975,0.000005303439,0.00005436969,0.9241446,0.05412986,0.0192925],"study_design_scores_gemma":[0.0004716656,0.0002211081,0.002346787,0.0001752536,0.00006428381,0.0001416631,0.0005371607,0.0127106,0.00002413979,0.968564,0.01452372,0.0002196678],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03726432,0.0004729671,0.9154344,0.03322833,0.0001910021,0.001392859,0.008846285,0.0001386997,0.00303117],"genre_scores_gemma":[0.5602168,0.0008648119,0.4367883,0.0006307228,0.00009024134,0.00003833883,0.0002667262,0.00007454553,0.001029496],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5229525,"threshold_uncertainty_score":0.9999201,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4392957219","doi":"10.1177/1536867x241233672","title":"sendemails: An automated email package with multiple applications","year":2024,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Names, Identity, and Discrimination Research","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Group for Research in Decision Analysis","funders":"","keywords":"Computer science; Audit; Audit trail; World Wide Web; Human–computer interaction","authors":[{"name":"Luca Fumarco","is_ca":false},{"name":"S. Michael Gaddis","is_ca":false},{"name":"Francesco Sarracino","is_ca":false},{"name":"Iain Snoddy","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07348863440698308,"gpt":0.4315671026476287,"spread":0.3580784682406456,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.002886861,0.0001380322,0.0001451448,0.0001831719,0.004309521,0.001937018,0.00114861,0.00004337197,0.00009000677],"category_scores_gemma":[0.0006482184,0.000096637,0.00002672814,0.000496891,0.0009295535,0.0004627164,0.0002156432,0.0006629658,0.00003352003],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000988155,"about_ca_system_score_gemma":0.0004847874,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008582977,"about_ca_topic_score_gemma":0.004962003,"domain_scores_codex":[0.9974329,0.0009843955,0.0003534753,0.0002303208,0.00064413,0.0003547584],"domain_scores_gemma":[0.9963098,0.001882821,0.0001469953,0.000972578,0.0004237528,0.0002641214],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004342452,0.0005729066,0.001959086,0.0001766222,0.0001839081,0.00004416984,0.0402815,0.00003298448,0.00007471433,0.8025245,0.02155913,0.132547],"study_design_scores_gemma":[0.001628747,0.0009865495,0.009541198,0.0006500832,0.0004076039,0.000209722,0.08406214,0.2714365,0.00003084146,0.2039333,0.4260475,0.001065837],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0453408,0.01030905,0.8332053,0.05978055,0.000532087,0.00496652,0.008941565,0.002063147,0.03486098],"genre_scores_gemma":[0.964775,0.00556583,0.02797372,0.0001180283,0.0001213382,0.0000793994,0.000267208,0.00003799721,0.001061453],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9194342,"threshold_uncertainty_score":0.9990991,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W15433897","doi":"10.1177/1536867x1401400310","title":"Csvconvert: A Simple Command to Gather Comma-Separated Value Files into Stata","year":2014,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Probability and Statistical Research","field":"Mathematics","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Variable (mathematics); Value (mathematics); Quarter (Canadian coin); Computer science; Simple (philosophy); Data file; Database; Arithmetic; Mathematics; Geography","authors":[{"name":"Alberto A. Gaggero","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09890068580412437,"gpt":0.4163357268686034,"spread":0.317435041064479,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.005779407,0.0003660789,0.0005539134,0.0001747995,0.002441307,0.0007062223,0.002377286,0.00008501409,0.0002004672],"category_scores_gemma":[0.01175581,0.0002642993,0.0000645517,0.0003911386,0.0007521101,0.0001850195,0.001348359,0.001412523,0.00008086508],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001301406,"about_ca_system_score_gemma":0.0001989697,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003067268,"about_ca_topic_score_gemma":0.0005586217,"domain_scores_codex":[0.9948403,0.002222416,0.001035856,0.0003998509,0.0007925684,0.0007089746],"domain_scores_gemma":[0.9812312,0.01445224,0.0003754089,0.002805407,0.0005787839,0.0005569765],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000202237,0.0005648634,0.0002096846,0.0002116463,0.0001676884,0.00001102489,0.007833977,0.00004912116,0.0001129377,0.8793288,0.05038156,0.06092648],"study_design_scores_gemma":[0.0008016562,0.0007496619,0.0005639687,0.0001818267,0.00008105917,0.00004814571,0.0006782048,0.08748529,0.00004514304,0.8581472,0.05086362,0.0003541899],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03960457,0.0003572432,0.9339144,0.01606719,0.0001224617,0.001637192,0.004889194,0.0001323728,0.003275349],"genre_scores_gemma":[0.688404,0.000719399,0.3087695,0.001033286,0.00007700652,0.00007456317,0.0003620943,0.000102074,0.000458025],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6487995,"threshold_uncertainty_score":0.9999809,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1492264985","doi":"10.1177/1536867x1201200106","title":"Respondent Driven Sampling","year":2010,"lang":"en","type":"preprint","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"HIV, Drug Use, Sexual Risk","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Snowball sampling; Sampling (signal processing); Respondent; Markov chain; Population; Markov chain Monte Carlo; Statistics; Sampling design; Computer science; Psychology; Econometrics; Demography; Mathematics; Sociology; Monte Carlo method; Telecommunications; Political science","authors":[{"name":"Matthias Schonlau","is_ca":true},{"name":"Elisabeth Liebau","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1608193926967826,"gpt":0.4394965898380185,"spread":0.2786771971412359,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.002880178,0.0004578268,0.0006742134,0.0003027176,0.001622013,0.0006332314,0.002133585,0.0002441399,0.00008290919],"category_scores_gemma":[0.002776262,0.0003349179,0.0001014665,0.0001765471,0.0006318624,0.00008125675,0.00311325,0.007323102,0.00003906292],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001422965,"about_ca_system_score_gemma":0.0007376825,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001035066,"about_ca_topic_score_gemma":0.0001804593,"domain_scores_codex":[0.9963537,0.0007476635,0.001123264,0.0004705538,0.0007870792,0.0005177216],"domain_scores_gemma":[0.9907259,0.002629424,0.001036506,0.004475637,0.0006961462,0.0004363714],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.001595193,0.003053173,0.003008222,0.001732934,0.003103298,0.0004531717,0.1264405,0.0008413168,0.004274883,0.07992879,0.03705204,0.7385165],"study_design_scores_gemma":[0.009151421,0.004518356,0.02566921,0.01072908,0.006130939,0.005188991,0.05319082,0.2952271,0.0003667039,0.3458898,0.2398447,0.004092894],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4717604,0.01032926,0.3967344,0.06956417,0.002888726,0.007161187,0.03054708,0.0005450895,0.01046968],"genre_scores_gemma":[0.6398749,0.01289253,0.3435747,0.0005910533,0.0003822734,0.00005519465,0.001652544,0.0001757058,0.0008010709],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7344236,"threshold_uncertainty_score":0.9999103,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2897092992","doi":"10.1177/1536867x1801800206","title":"Attrition Diagrams for Clinical Trials and Meta-analyses in Stata","year":2018,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Manitoba","funders":"","keywords":"Attrition; Documentation; Computer science; Diagram; Set (abstract data type); Programming language; Database; Medicine","authors":[{"name":"Christiaan H. Righolt","is_ca":true},{"name":"Salaheddin M. Mahmud","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.9846130439606658,"gpt":0.7497721060674238,"spread":0.234840937893242,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3185522,0.0003437929,0.004174404,0.0003768518,0.001294744,0.002398841,0.00296225,0.0000768195,0.0006778249],"category_scores_gemma":[0.1864691,0.000148453,0.0008294189,0.0006940832,0.0007124861,0.0002991328,0.0005853386,0.0006547998,0.00005811589],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002705485,"about_ca_system_score_gemma":0.0001209065,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003837054,"about_ca_topic_score_gemma":0.0003762013,"domain_scores_codex":[0.940099,0.04361108,0.01310247,0.0007574575,0.002043268,0.0003867497],"domain_scores_gemma":[0.8880188,0.09551094,0.00821835,0.005854626,0.0021007,0.0002966104],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000341928,0.001166124,0.006449492,0.0001783818,0.01184104,0.00001481587,0.007822931,0.00002835247,0.00009704492,0.1180176,0.3016831,0.5523592],"study_design_scores_gemma":[0.00277112,0.001656874,0.01301574,0.0001257963,0.01937554,0.0001376017,0.006495683,0.1465849,0.00002151309,0.557754,0.2513042,0.0007571387],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02788903,0.01683874,0.917348,0.02421967,0.0003558996,0.004041232,0.008501693,0.0000136911,0.0007920002],"genre_scores_gemma":[0.7227942,0.009625616,0.2647119,0.0009799713,0.0002383111,0.0001478272,0.0002579495,0.00004720627,0.001196982],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6949052,"threshold_uncertainty_score":0.9986368,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4205090414","doi":"10.1177/1536867x211063410","title":"Review of Michael N. Mitchell’s Interpreting and Visualizing Regression Models Using Stata, Second Edition","year":2021,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Insurance, Mortality, Demography, Risk Management","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"NOSM University; Lakehead University","funders":"","keywords":"Computer science; Regression; Statistics; Econometrics; Mathematics","authors":[{"name":"Angela MacIsaac","is_ca":true},{"name":"Bruce Weaver","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07237149430034216,"gpt":0.4009634019579996,"spread":0.3285919076576574,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.004528692,0.0001672317,0.0003257236,0.0001128195,0.00195661,0.0002796154,0.0006418124,0.00004333474,0.00004678031],"category_scores_gemma":[0.0008258488,0.0001411593,0.00005393432,0.000307916,0.0006089736,0.0003772734,0.0005726875,0.0005553838,6.4438e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006818283,"about_ca_system_score_gemma":0.0002077591,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002568459,"about_ca_topic_score_gemma":0.0005994762,"domain_scores_codex":[0.9966077,0.001598464,0.0007686877,0.0002355996,0.0004842111,0.0003053975],"domain_scores_gemma":[0.9965339,0.001043471,0.0008300171,0.0008530932,0.0006113034,0.0001282062],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001446168,0.001020404,0.002561169,0.01150351,0.0008674277,0.00008676964,0.09978311,0.0001283295,0.001538715,0.5457281,0.02144068,0.3151971],"study_design_scores_gemma":[0.002827122,0.0008405698,0.002288951,0.09614595,0.001509892,0.0003359349,0.1002566,0.2513394,0.0004826472,0.4392678,0.102566,0.002139131],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1560954,0.316643,0.4919347,0.01690049,0.001141433,0.002774285,0.005332054,0.0001318659,0.009046749],"genre_scores_gemma":[0.6387616,0.2877136,0.07235573,0.0008194218,0.00008082588,0.000009757118,0.0001486395,0.00003606762,0.00007432031],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.4826662,"threshold_uncertainty_score":0.9993427,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3143332811","doi":"10.1177/1536867x211000008","title":"msreg: A command for consistent estimation of linear regression models using matched data","year":2021,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Estimator; Ordinary least squares; Matching (statistics); Statistics; Econometrics; Sample (material); Parametric statistics; Linear regression; Least-squares function approximation; Regression; Mathematics; Computer science","authors":[{"name":"Masayuki Hirukawa","is_ca":false},{"name":"Di Liu","is_ca":false},{"name":"Artem Prokhorov","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.4693799478599424,"gpt":0.4955069563491542,"spread":0.02612700848921179,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002668425,0.000185356,0.0004226289,0.00006197044,0.0009791668,0.00015874,0.001034349,0.00005151335,0.00001801407],"category_scores_gemma":[0.006807721,0.0001294449,0.00003867209,0.000155246,0.0003264968,0.0001612657,0.0008790537,0.0004445093,4.25339e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003510805,"about_ca_system_score_gemma":0.0002461029,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002624296,"about_ca_topic_score_gemma":0.00002010115,"domain_scores_codex":[0.9975716,0.0007350937,0.0008795047,0.0002434455,0.0003307811,0.0002394998],"domain_scores_gemma":[0.9872379,0.00884199,0.0007473842,0.002363795,0.0006938027,0.000115096],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001284091,0.0005151091,0.00002494053,0.0005176136,0.000223703,0.000009723805,0.002993932,0.0007665122,0.0004911138,0.8981333,0.002968539,0.09322713],"study_design_scores_gemma":[0.0003755892,0.0000852958,0.00001497894,0.0003512914,0.0001336273,0.00005598179,0.0005757016,0.5731102,0.00009180122,0.4249641,0.000151432,0.00009002851],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004999455,0.0005427432,0.9870858,0.001154353,0.00006694265,0.0003668195,0.005653911,0.00001394387,0.0001159901],"genre_scores_gemma":[0.1121787,0.0008059955,0.8865254,0.00006127793,0.00002182414,0.00000776812,0.0003391742,0.00002887017,0.00003101499],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.5723437,"threshold_uncertainty_score":0.814997,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1512757258","doi":"10.1177/1536867x0600600108","title":"Mata Matters: Creating New Variables—sounds Boring, isn't","year":2006,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"demographic modeling and climate adaptation","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Column (typography); Quarter (Canadian coin); Focus (optics); Computer science; Programming language; History; Telecommunications; Archaeology","authors":[{"name":"William Gould","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1189247320914132,"gpt":0.3869377159663415,"spread":0.2680129838749282,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.006576093,0.0002455064,0.0003220567,0.0002871503,0.002275492,0.002326309,0.002346736,0.00005595526,0.0002119397],"category_scores_gemma":[0.001830358,0.000166166,0.00006647826,0.0006078763,0.000287767,0.0004141391,0.0005504918,0.0007509094,0.00006264356],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004800934,"about_ca_system_score_gemma":0.0001765511,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007693754,"about_ca_topic_score_gemma":0.0002282559,"domain_scores_codex":[0.9959599,0.0007206713,0.001328905,0.000378406,0.001178753,0.0004333257],"domain_scores_gemma":[0.9913366,0.004746426,0.0009525277,0.002251747,0.0005163277,0.0001963646],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000112877,0.0004719651,0.005396418,0.00003787539,0.0001361359,0.00002630375,0.00745357,0.00321122,0.0003140269,0.4549858,0.3217094,0.2061444],"study_design_scores_gemma":[0.0009323211,0.0002485091,0.00557558,0.0002233343,0.0001254986,0.0002382534,0.004352252,0.2139231,0.00001503459,0.68072,0.0932146,0.0004315191],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01702211,0.001366567,0.9502385,0.01643772,0.0003151161,0.000384272,0.001157964,0.00008969499,0.01298808],"genre_scores_gemma":[0.6939763,0.001900077,0.2958028,0.001175261,0.0002915637,0.00001180799,0.0003224737,0.0000762554,0.006443463],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6769542,"threshold_uncertainty_score":0.9990234,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4390051166","doi":"10.1177/1536867x231212432","title":"csa2sls: A complete subset approach for many instruments using Stata","year":2023,"lang":"en","type":"article","venue":"The Stata Journal Promoting communications on statistics and Stata","topic":"Consumer Market Behavior and Pricing","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University","funders":"","keywords":"Estimator; Mean squared error; Statistics; Monte Carlo method; Econometrics; Instrumental variable; Least-squares function approximation; Mathematics; Bias of an estimator; Computer science; Minimum-variance unbiased estimator","authors":[{"name":"Seojeong Lee","is_ca":true},{"name":"Siha Lee","is_ca":true},{"name":"Julius Owusu","is_ca":true},{"name":"Youngki Shin","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1563916626367033,"gpt":0.3397580369528209,"spread":0.1833663743161176,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.002649477,0.0002719771,0.0003112009,0.0003495943,0.002723062,0.0012353,0.001369654,0.00004251341,0.00003305243],"category_scores_gemma":[0.0004473063,0.0002165888,0.00006528777,0.0006150785,0.0002511419,0.0005421758,0.001077182,0.0005758452,0.00001948995],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004429427,"about_ca_system_score_gemma":0.00006996967,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000274295,"about_ca_topic_score_gemma":0.0000538549,"domain_scores_codex":[0.997982,0.0001412337,0.0006573094,0.0002990707,0.0003864363,0.0005339051],"domain_scores_gemma":[0.9969979,0.0008201262,0.0005777236,0.001232857,0.0003199279,0.00005146458],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0009843999,0.001299298,0.04360315,0.002306318,0.00100352,0.00005774677,0.005922775,0.0006863764,0.001932086,0.2370798,0.0468451,0.6582794],"study_design_scores_gemma":[0.001924457,0.0000916412,0.01176082,0.0002710821,0.0004563939,0.00007016514,0.003249172,0.8981785,0.000005266465,0.0262837,0.05708025,0.000628506],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5533379,0.0009832818,0.4215402,0.006445488,0.0009091009,0.003975159,0.00819676,0.0004997303,0.004112336],"genre_scores_gemma":[0.9330062,0.0005830327,0.06368809,0.0005304419,0.0002054312,0.00005364835,0.001694694,0.00009569478,0.000142741],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8974922,"threshold_uncertainty_score":0.9998015,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}