Absolute Rating

Why absolute rating matters

The choice between absolute and relative rating is one of the highest-leverage decisions in performance management design. It shapes employee experience, manager workload, fairness perception, legal defensibility, and ultimately whether the appraisal system produces useful talent decisions or organisational theatre. Research published in the International Journal of Selection and Assessment (Roch et al., 2007) found that absolute rating formats are perceived as significantly more fair than relative rating formats by employees – a finding consistent with two decades of replication. Following well-documented failures of forced ranking at General Electric, Microsoft, and Yahoo!, most large US employers moved toward absolute-rating approaches in the 2010s, though hybrid models remain common.

Summarise this post with:

Absolute rating vs relative ranking

Dimension	Absolute rating	Relative ranking / Forced distribution
Comparison basis	Employee vs predefined standards	Employee vs peer group
Outcome distribution	Variable – any distribution possible	Forced – e.g. 20% top / 70% middle / 10% bottom
Multiple employees same rating?	Yes – everyone can be rated ‘exceeds’	No – quotas constrain
Fairness perception (research)	Higher	Lower
Legal defensibility	Stronger if criteria are job-related	Higher litigation risk; Ford, Microsoft, Conoco class actions
Team collaboration impact	Neutral to positive	Tends to be negative (zero-sum)
Best for	Stable workforces, capability building, modern PM	Aggressive performance cultures, talent rotation
Notable adopters	Adobe (post-2012), Deloitte, Microsoft (post-2013)	GE under Welch (since abandoned), Enron (cautionary)

In practice, most modern systems are absolute-rating with a soft calibration step – managers rate against criteria, then HR runs distribution analyses to surface inconsistency rather than enforce quotas.

The four main absolute rating methods

Graphic rating scale

The most widely used absolute rating method. Each performance dimension (quality of work, productivity, teamwork, communication) is rated on a numeric scale, typically 1-5 or 1-7, with anchored descriptions at each level. Simple to deploy at scale; consistent across raters when anchors are well-written; vulnerable to halo and central tendency biases when anchors are vague.

Behaviorally anchored rating scale (BARS)

An enhanced graphic rating scale where each numeric level is anchored to a specific observable behaviour rather than a generic adjective. BARS reduces rater subjectivity and improves construct validity but requires significant upfront work to develop role-specific behaviours. See BARS for the full treatment.

Critical incident method

The rater documents specific incidents – both successful and unsuccessful – observed during the review period. The appraisal narrative is built from these documented incidents. Highly defensible and developmental; impractically intensive for managers with large teams. Best when paired with a graphic scale rather than used alone.

Management by objectives (MBO) and modern okrs

Rather than rating behaviours, MBO evaluates whether the employee achieved jointly-set objectives during the period. Modern OKR (Objectives and Key Results) systems are direct descendants. MBO and OKR-based appraisal work well for goal-driven knowledge work but require quality goal-setting upstream.

Worked example: a graphic rating scale anchored with BARS principles

Performance dimension: Quality of Deliverables. Scale 1-5.

• 1 – Below standard: Work product frequently contains errors requiring rework. Senior colleagues spend material time correcting before delivery.
• 2 – Approaching standard: Work product meets baseline expectations on routine items but requires review and correction on complex or first-time tasks.
• 3 – Meets standard: Consistently delivers work product that meets quality expectations without significant rework. Catches own errors before final delivery.
• 4 – Exceeds standard: Quality is reliably above baseline. Peers and managers reference this employee’s work as an example.
• 5 – Distinguished: Sets the standard for the team. Materially improves quality processes, not just outputs. Coaches others to similar quality without diminishing own delivery.

Absolute rating: advantages and disadvantages

Advantages

• Higher perceived fairness. Per Roch 2007 and subsequent research, employees rate absolute systems as significantly more fair than relative systems.
• Better developmental feedback. Standards-based rating produces specific feedback (‘here is where you fall short of the standard’) rather than relational (‘you ranked below others’).
• Lower litigation exposure. EEOC and OFCCP scrutiny falls more heavily on relative systems that produce disparate impact.
• Compatible with team collaboration. No zero-sum incentive structure.

Disadvantages

• Rating inflation. Without calibration, manager-by-manager rating distributions drift – most managers eventually rate most employees at 4-5.
• Halo and central tendency biases. See first impression error for the cognitive bias context.
• Calibration discipline required. Absolute rating without cross-manager calibration is fragile. Calibration sessions are essential and not free.
• Limited differentiation for talent decisions. When most employees are rated 4, differentiating top performers for promotion requires additional signal beyond the rating itself.

Common rater biases – and how absolute methods address them

• Halo / horn effect. One strong (or weak) dimension dominates ratings on unrelated dimensions. BARS with role-specific behaviour anchors materially reduces this.
• Central tendency. Rater defaults all employees to the middle of the scale. The fix is rater training and calibration.
• Leniency and severity. Manager rates everyone high (leniency) or low (severity). Cross-manager calibration surfaces and corrects these distributions.
• Recency bias. Recent events dominate the rating. Critical incident method, where managers document throughout the period, mitigates this.
• Stereotype and demographic bias. Race, gender, age, and other protected characteristics can influence ratings. Periodic adverse-impact analysis on rating distributions surfaces patterns requiring intervention.

EEOC compliance: performance appraisal as a selection procedure

Per EEOC enforcement guidance and case law, performance appraisals are ‘selection procedures’ under the Uniform Guidelines on Employee Selection Procedures (29 CFR Part 1607) when used for consequential employment decisions – promotion, RIF selection, termination, compensation. They must therefore meet job-relatedness and adverse-impact standards. Practical implications:

• Rating criteria must be job-related – derived from documented job analysis.
• Rating distribution by protected class must not produce disparate impact (four-fifths rule analysis).
• Rater training on bias and consistency is part of the defensibility argument.
• Calibration meetings should be documented; the calibration decision rationale matters in litigation.

Per SIOP Principles for the Validation and Use of Personnel Selection Procedures, performance appraisals used for consequential decisions should meet the same validation standards as other selection tools.

How to design an absolute rating system: 7-step playbook

1. Start with job analysis. Identify the work behaviours, knowledge, and skills that drive job performance.

1. Choose the method. Graphic rating for scale and speed; BARS for defensibility and developmental power; critical incident for high-stakes individual roles; MBO/OKR for goal-driven work.
2. Develop behavioural anchors. Even graphic scales benefit from anchored descriptions. Write them per role family, not generically.
3. Train raters. Rater training covers the method, common biases, calibration, and documentation.
4. Run calibration sessions. Cross-manager calibration before ratings are finalised surfaces leniency, severity, and stereotype patterns.
5. Conduct adverse-impact analysis. Before ratings drive decisions (especially RIF and promotion), run four-fifths analysis by protected class.
6. Iterate annually. Rating distributions, fairness perception, and downstream outcomes feed back into anchor refinement and rater training for the next cycle.

Frequently asked questions