Appraisal Methods and Biases: Interview-Ready HR Revision Guide
In a calibration room, one manager insists her team member deserves the top rating because “the client loved last week’s save,” while another says, “I never give a 5 unless someone is exceptional.” Same company, same rating scale, two different mental yardsticks - and that is exactly where appraisal bias begins.
- Performance appraisal evaluates employee performance against standards; the method chosen decides what evidence gets noticed.
- Graphic rating scales are simple but invite halo, leniency, severity and central tendency bias.
- Ranking and forced distribution create differentiation but can distort ratings when teams are uneven in quality.
- BARS and BOS reduce ambiguity by rating observable behaviours, but still suffer from recency and documentation bias.
- MBO and OKR-style reviews improve goal alignment but can create outcome bias and sandbagging.
- 360-degree feedback broadens input but may introduce popularity, politics and rater-familiarity bias.
- The best answer is never “which method is best?” It is “which method fits the role, and what bias controls will we add?”
Big Picture: Appraisal Is a Filtering System, Not Just a Form
Every appraisal method turns messy human work into a neat decision - rating, bonus, promotion, development plan or exit risk. Bias enters because the system filters reality before it becomes a score.
Core Explanation: Appraisal Methods and the Bias Each One Introduces
The central idea is simple: no appraisal method is bias-free. Each method solves one problem and creates another. A strong HR answer names the method, explains what it measures, then names the specific bias risk and control.
Think of appraisal methods on two dimensions: how objective the evidence is and how broad the source of evidence is. The more subjective and narrow the method, the higher the need for calibration.
The Bias Map: Match Each Bias to the Moment It Enters
Bias is easier to fix when you know where it enters. Some biases enter during observation, some during memory, some during scoring and some during the final calibration discussion.
In large Indian IT services employers such as Infosys and TCS, appraisal quality is complicated by project-based work, onsite opportunities, billability, client feedback and changing reporting managers. A developer may be rated by a project manager who saw only one phase of work, while promotion decisions are compared across many delivery units. The strategic lesson is that calibration is not bureaucracy - it is needed because project context and manager standards differ sharply.
Definitions You Should Be Able to Say Cleanly
Performance appraisal: Gary Dessler defines it as “evaluating an employee’s current and/or past performance relative to his or her performance standards.”
Rater bias: A systematic error in evaluation caused by the evaluator’s perception, memory, preference or comparison point.
Calibration meeting: A review discussion where managers align rating standards across teams before final people decisions.
How to Audit Whether an Appraisal System Is Fair
If a company asks “is our appraisal system working?”, do not answer with feelings. Use a small set of diagnostic metrics. Benchmarks vary by role and industry, but these measures quickly reveal whether ratings are reliable, timely and fair.
Worked example - adverse impact check: Suppose 45 out of 150 men are promoted after appraisal, so the selection rate is 30%. Suppose 18 out of 90 women are promoted, so the selection rate is 20%. The adverse impact ratio is 20% ÷ 30% = 0.67. Because 0.67 is below 0.80, HR should investigate whether the appraisal criteria, manager ratings or promotion calibration created unintended bias.
Case Study: Deloitte Rebuilds Performance Management Around Check-ins
Deloitte replaced a heavy annual review process with more frequent check-ins and future-focused talent questions to reduce rating noise and improve coaching.
Situation: Deloitte’s traditional performance management process had become heavy, time-consuming and backward-looking. Publicly discussed in Harvard Business Review, the firm found that its annual process consumed nearly 2 million hours a year. More importantly, the firm questioned whether ratings captured the employee’s real performance or the rater’s personal standards.
The move: Deloitte shifted emphasis from annual rating rituals to frequent check-ins between team leaders and employees. Instead of asking managers to make broad personality judgments, it used more future-focused questions such as whether the leader would want the employee on the team again or whether the employee was ready for more responsibility.
The result or lesson: The primary driver was a move from retrospective judgment to frequent, work-adjacent coaching. Supporting drivers included simpler questions, stronger manager ownership, faster feedback and less dependence on one annual memory-based score. The lesson for appraisal design is powerful: reducing bias is not only about a better form - it is about changing the rhythm, evidence and decision rules of performance management.
How AI Changes Appraisal Methods and Bias
1. AI can improve evidence capture, but it can also scale old bias. Tools can summarise performance notes, project updates, sales calls or customer feedback into appraisal inputs. This reduces pure memory bias, but if historical ratings were biased, AI trained on them may reproduce the same patterns at scale.
2. AI can flag rating anomalies before calibration. HR teams can use analytics to detect managers who rate unusually harshly, teams with sudden rating inflation, or demographic groups with lower promotion conversion after similar performance evidence. The human decision still matters, but AI makes invisible patterns visible.
3. AI changes feedback writing. Generative AI can help managers turn vague comments like “needs ownership” into specific, behaviour-based feedback. The risk is generic feedback that sounds polished but lacks evidence. In India, companies must also be careful with employee data privacy, especially under the Digital Personal Data Protection Act, 2023.
Use NotebookLM or ChatGPT to practise like an HR analyst: upload a company HR policy, sample appraisal form and this lesson, then ask, “Identify the appraisal method, likely biases, fairness metrics and calibration questions for this company.”
Interview Relevance
“If you were designing an appraisal system for a sales team and a product engineering team, would you use the same appraisal method? Also, what biases would you watch for?”
Use the phrase “method-bias-control.” For every appraisal method you mention, immediately add the bias it creates and the control you would use. That sounds managerial, not textbookish.
Common Mistake
The most common mistake is saying “360-degree feedback is the fairest because it has more raters.” More raters can reduce single-manager bias, but they can also add popularity, politics and fear-based feedback. One-line fix: say “360 works only when rater selection, anonymity, question design and calibration are done properly.”
What to Revise Next
Next, revise the tools that make appraisal systems operational. Start with Rating Scales, Forced Distribution & the Calibration Meeting to understand how ratings are standardised, then move to Three-Hundred-and-Sixty-Degree Feedback Done Properly to learn how multi-source feedback can be designed without turning into a popularity contest.