Interactional pluralism evaluation on contested-value pressure corpus
0.814Agreement ShiftModel B (GPT-4o)
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Model B (GPT-4o)Model=GPT-4o, Sample size (N)=100, Annotator setup=secondary independent-annotator replication on a stratified subset2026.05 | 0.814 | 0.185 | — | — | 0.112 | 0.623 | 0.14 | 0.674 | |
| Model A (Sonnet 4.5)Model=Claude Sonnet 4.5, Sample size (N)=198, Annotator setup=three independent annotators2026.05 | 0.732 | 0.247 | 0.116 | 0.302 | 0.184 | 0.491 | 0.21 | 0.522 |