Updated 2026-09-16
Scoring calibration in hiring is a training exercise where interviewers independently score the same recorded answers or live scenarios, then discuss discrepancies against rubric anchors — aligning what 'meets bar' means before real candidates are evaluated. It reduces inter-rater variance, halo effect, and contrast effect distortion across panels and high-volume hiring teams.
Select representative samples
Recorded answers spanning strong, borderline, and weak — anonymized from past interviews or scripted actors.
Independent scoring
Each interviewer scores every competency with evidence notes — no discussion until all submit.
Reveal and discuss gaps
Compare scores — large spreads indicate vague anchors or training gaps, not 'wrong' candidates.
Revise anchors or guidance
Update rubric language, examples, or prohibited shortcuts — e.g. scoring charisma on technical criteria.
Re-calibrate periodically
After new question sets, new interviewers, or detected adverse impact shifts.
Panel interviews with multiple scorers — without calibration, the same answer receives 2 and 4 from different panelists.
High-volume hiring and campus hiring — dozens of interviewers must apply one standard.
AI-human hybrid flows — calibrate human reviewers on when to override model scores using evidence-cited review.
After changing score cutoffs — recalibrate before tightening thresholds to avoid false positive and false negative swings.
At onboarding for new interviewers and quarterly or when process changes — whichever comes first for high-volume teams.
AI can enforce rubric structure but human interviewers still need alignment on anchor interpretation for live stages.
It reduces scoring inconsistency — a major bias amplifier. It does not remove all bias sources; adverse impact monitoring continues.
Head back to Hiring glossary or start now.
· 3 free credits · pay per interview · nothing recurring