# F.3 score (aggregate only; per-claim Opus answers are not shown here)

## Kept set (26, primary)
- compared: 26 (no Opus label: 0; more than one Opus label, first used: 8)
- verdicts agree: **24 of 26** (92%, 95% interval 76% to 98%); kappa -0.04
- table: founder fail / Opus pass 1; founder pass / Opus fail 1; founder pass / Opus pass 24
- exact flag agree: 18 of 26

## All 40 as labelled (secondary)
- compared: 40 (no Opus label: 0; more than one Opus label, first used: 10)
- verdicts agree: **34 of 40** (85%, 95% interval 71% to 93%); kappa -0.04
- table: founder fail / Opus pass 5; founder pass / Opus fail 1; founder pass / Opus pass 34
- exact flag agree: 22 of 40

## Sensitivity
- kept claims where the Opus reviewers split: 1
- kept agreement if a split counts as fail: 23 of 26

## Jev accepts in the kept set
- accepted by Jev: 9 of 26
- founder pass on those: 9; Opus pass: 9
- disagreements that were accepts: 0
