Behavioral Rating Scales (BARS)
Quick definition
Behavioral Rating Scales (BARS) combine numeric ratings (e.g., 1–5) with anchored behavioral descriptions per level, so two interviewers score the same answer the same way. Each level is paired with a sentence describing the observable behavior that maps to it.
How it works
The structure is typically competency-based. For each competency (say, "navigates ambiguity") you define a 1–5 scale and write one anchor per level: 1 = "avoids ambiguity, asks for repeated clarification"; 3 = "moves forward with a few unknowns but doesn't surface risks explicitly"; 5 = "takes MVP-sized action quickly, flags risks visibly, adds them to the plan." The interviewer matches the candidate's answer to the closest anchor. Per-question scores are then aggregated using role-specific weights — instead of one overall gut-feel note, you get a concrete record.
Why it matters
Concrete anchors make it likelier that two raters give the same answer the same score — inter-rater reliability goes up, which directly raises the ceiling on predictive validity. On the candidate side, feedback gets specific: "to land at a 4 instead of a 3, surface risks explicitly." On the regulatory side, BARS leaves a defensible record against adverse-impact challenges.
Related terms
- Structured Interview — BARS is the scoring component of a structured interview.
- Agentic Interview — the output of an agentic interview is typically scored against BARS.
- Voice AI Interviewer
- How AI interviews work
- EU AI Act compliance