How to Build Behaviourally Anchored Rating Scales (BARS) for Candidate Assessments

Two interviewers hear the same candidate answer. One awards five out of five because the candidate sounds confident. The other awards two because the answer lacks a specific action plan. Both may believe they are being fair, but their scores are based on different, largely unstated standards.

Behaviourally anchored rating scales (BARS) make scoring expectations more explicit. Rather than asking assessors to judge vague traits such as “excellent communication”, a BARS rubric describes observable behaviours associated with different levels of performance.

For recruitment, the objective is to connect a job-relevant competency to clear evidence that an interviewer, work-sample assessor or assessment-centre observer can recognise and record. Well-designed anchors can improve scoring transparency and assessor training, although the mere presence of a BARS rubric does not establish reliability, fairness or validity.

This guide explains how to develop a practical BARS framework, provides an original worked example and identifies the controls employers should apply before relying on scores.

What is a behaviourally anchored rating scale?

A BARS instrument typically combines a numerical rating scale with descriptions of behaviour representing different performance levels. The descriptions are intended to reduce ambiguity about what a score means.

Consider a customer service assessment. “Good problem-solving” is an abstract judgement. “Identifies the customer's actual issue, verifies the relevant policy and proposes a feasible next step” describes observable evidence.

The middle scores must also be defined or accompanied by a consistent rule for judging responses between anchors. Otherwise, reviewers may still apply different assumptions.

When should employers use BARS?

BARS can be useful when people must evaluate observable behaviour in structured interviews, role plays, presentations, practical simulations and some work samples.

It is less appropriate to replace objective marking for questions with a single correct numerical or factual answer. A spreadsheet formula that is demonstrably correct should generally be marked against a technical answer key, not an impressionistic behaviour scale.

Step 1: Define the job competency precisely

Begin with job analysis. Identify important tasks, then define the behaviours that contribute to successful performance.

For a customer service representative, “customer focus” is too broad to score consistently. Break it into observable components: recognising the issue, confirming constraints, choosing a permitted resolution and communicating the next action.

Write a competency definition that is narrow enough to guide scoring.

Example competency: Complaint resolution judgement. The ability to identify a customer's substantive problem, apply stated service rules, propose a feasible response and explain what happens next.

Avoid bundling too many unrelated skills into one scale. If an assessor must judge accuracy, empathy, writing mechanics, legal compliance and numerical reasoning simultaneously, the final score becomes difficult to interpret.

Step 2: Collect critical incidents and observable examples

A traditional BARS development process uses examples of effective and ineffective job behaviour, often called critical incidents. These can be gathered from knowledgeable employees, supervisors and records of realistic work situations, subject to confidentiality and privacy controls.

Ask contributors to describe what happened, what the person did and what made the behaviour effective or ineffective. Do not ask only for adjectives such as “professional” or “confident”.

For a procurement analyst, examples might include checking delivery performance before recommending a supplier, documenting assumptions and escalating a conflict of interest. An ineffective example might be recommending the cheapest quote while ignoring a stated quality requirement.

Review incidents for relevance, clarity, redundancy and potential bias. The examples should reflect job performance rather than personal style preferences.

Step 3: Group incidents into coherent dimensions

Sort the incidents by the competency they demonstrate. Separate genuinely different dimensions instead of forcing every example into one overall score.

For a fictional complaint-handling assessment, the dimensions could be:

Subject-matter experts should review whether each dimension is necessary for the role and whether important competencies are missing.

Step 4: Write behaviourally specific anchors

Anchors should describe evidence, not label the candidate as a type of person. Avoid wording such as “brilliant”, “weak attitude” or “natural leader”.

Use parallel language across performance levels, and avoid defining higher scores solely by writing length or excessive detail.

Worked example: Five-point complaint-resolution scale

Candidates receive this fictional scenario:

A customer says an order arrived three days late and requests an immediate full refund. The company policy allows a delivery-fee refund for confirmed late delivery, but a full product refund requires a separate eligibility review. Explain how you would respond.

These are illustrative anchors, not a validated instrument. In actual development, subject-matter experts should review whether the differences between levels are meaningful, observable and sufficiently distinguishable.

A candidate should not receive extra credit for inventing a remedy that violates the policy, even if the answer sounds customer-friendly.

Step 5: Create scoring instructions for assessors

Anchors alone do not resolve every judgement call. A scoring manual should specify how to handle partial evidence, conflicting behaviours and responses that fall between anchor descriptions.

For example:

• Score the evidence provided, not assumptions about the candidate's intentions.
• Record a short supporting quotation or behavioural observation.
• Do not infer a competency from confidence, accent or presentation style unless directly relevant to the role.
• Apply the same instructions and information to all candidates, with appropriate accommodations.
• Where multiple dimensions are used, score each separately before calculating any composite.
• Escalate unclear or disputed cases for documented moderation.

If a candidate proposes the correct remedy but promises an unauthorised deadline, assessors should know how that affects the relevant dimension. The scoring manual should resolve such cases before live use where possible.

Step 6: Calibrate assessors using common examples

Training should include independently scored sample responses, followed by discussion of the evidence supporting each score. Merely reading the rubric is rarely enough.

Illustrative calibration exercise

Three assessors independently score the same fictional response on the five-point scale.

The reference score is a training judgement, not a statistical truth. The aim is to identify why scores differed and revise anchors or instructions where necessary.

After calibration, use an appropriate inter-rater reliability or agreement method on a sufficiently representative sample. A single moderated example does not establish reliable scoring.

Step 7: Pilot the scale and review score quality

Pilot the assessment with people relevant to the intended population, where practical. Examine whether candidates understand the task and whether assessors can distinguish performance levels.

Useful questions include:

1. Are some anchors never used?
2. Do reviewers consistently confuse adjacent scores?
3. Are certain dimensions redundant?
4. Does the rubric reward job-related behaviour rather than answer length?
5. Can assessors explain scores with observable evidence?
6. Are there accessibility or language barriers unrelated to the intended competency?
7. Do scoring differences materially affect hiring decisions?

A small pilot can reveal obvious design flaws. Strong claims about score reliability, subgroup differences or predictive validity require an appropriate study design and sample.

Step 8: Combine BARS scores without hiding essential weaknesses

Suppose a supervisory role uses three behavioural dimensions. The following fictional weights and scores illustrate a composite calculation:

The arithmetic is (4 × 0.40) + (3 × 0.35) + (5 × 0.25) = 3.90. This does not prove the weights are appropriate.

If escalation judgement is safety-critical, the employer may need a separate minimum requirement rather than allowing a high delegation score to compensate for a serious weakness. Such rules should be justified through job analysis and documented before candidate scores are reviewed.

BARS versus other scoring approaches

These approaches are not mutually exclusive. An assessment programme may use objective answer keys for numerical questions and behavioural anchors for judgement exercises.

Common BARS design mistakes

Using personality labels instead of behaviours. “Confident leader” is not a sufficiently observable scoring anchor.

Writing only extreme anchors. Without meaningful intermediate levels or clear decision rules, reviewers may still disagree.

Rewarding lengthy answers. A concise, accurate response can be stronger than a long response containing irrelevant detail.

Combining unrelated competencies. One score may hide whether a candidate failed on accuracy, communication or judgement.

Copying anchors between unrelated roles. The same behaviour can have different importance in a contact centre, hospital or engineering environment.

Claiming validation because the rubric looks detailed. Job relevance, score precision, fairness and intended use still need evidence.

Changing anchors during live assessment. Uncontrolled changes make candidate scores difficult to compare.

Practical BARS implementation checklist

• Document job tasks and the competency being assessed.
• Collect and review relevant behavioural incidents.
• Group examples into coherent, non-overlapping dimensions.
• Write observable anchors at meaningful performance levels.
• Check for irrelevant style, cultural and language preferences.
• Define scoring instructions and rules for partial evidence.
• Calibrate assessors with independent sample scoring.
• Pilot with suitable participants and inspect disagreements.
• Justify any weights or minimum standards.
• Version-control the rubric and monitor scoring quality.

Applying behavioural scoring in Talentrenza Assess

Talentrenza Assess supports structured skills-assessment workflows. Employers can use a documented competency blueprint alongside suitable assessment tasks and review results as part of a broader selection process.

A BARS rubric should not be described as an automatically available platform feature unless the relevant scoring workflow has been verified. Where human judgement is required, employers should maintain the scoring guide, assessor records and moderation controls appropriate to their process.

For complementary reading, see Competency-Based Assessments for Employee Skills, Design Pre-Employment Assessment Tests by Job Role, and Fair and Objective Employee Promotion Assessments.

Frequently asked questions

What does BARS stand for in recruitment?

BARS stands for Behaviourally Anchored Rating Scales. The approach links numerical scores to descriptions of observable performance relevant to a defined competency.

Can BARS be used for structured interviews?

Yes. Interviewers can score job-related evidence using predefined behavioural anchors, provided the questions, scoring rules and assessor training are appropriate.

Is a five-point scale always better than a three-point scale?

No. The appropriate number of levels depends on whether meaningful behavioural distinctions can be defined and applied consistently. More points do not automatically create greater precision.

Does BARS eliminate interviewer bias?

No. Clear anchors and training may reduce some sources of inconsistency, but fairness and bias still require deliberate review and evidence.

Can BARS scores be combined with numerical reasoning results?

They can contribute to a documented multi-method selection process, but weights and decision rules should reflect job requirements and the quality of available evidence.

Is a BARS rubric automatically validated once experts approve it?

No. Expert review can contribute content evidence, but intended score use, scoring consistency, fairness and other relevant validity evidence still need consideration.

Conclusion

Behaviourally anchored rating scales help employers replace vague impressions with explicit descriptions of job-relevant performance. Their value depends on the quality of the job analysis, the specificity of the anchors, assessor calibration and the evidence supporting how scores are used.

A useful BARS rubric should help a reviewer answer a simple question: What did the candidate actually demonstrate, and why does that evidence justify this score?