
Before
After

Reviewers calibrate against the group score, but only after they've recorded their own. When the two columns look identical, people read the wrong one and either anchor too early or double-enter. A small layout change here removes a real source of evaluator error.