Score Normalisation in Examinations

Score normalisation is a statistical process that converts marks obtained in different examination sessions into a comparable form. It adjusts for unintended differences in question-paper difficulty so that candidates are assessed by their relative performance, rather than by raw marks alone.

Why normalisation is needed

Multi-session examinations use different question papers based on a common syllabus and blueprint, but the papers may still vary slightly in difficulty. Direct comparison of raw scores could therefore advantage candidates assigned an easier session.

• Scores are first analysed separately for each session and then mapped onto a common scale.
• A candidate's position within the session is used to estimate comparable performance across sessions.
• A monotonic transformation generally preserves the ordering of candidates within each session.

Common methods

The examination authority must disclose and consistently apply a predetermined method. No single formula is universally suitable for every examination.

• Under the percentile method, a score represents the percentage of candidates in that session who scored at or below it; a percentile is not the same as percentage marks.
• The standard-score method expresses performance in relation to the session mean and standard deviation, and may then map it onto a reference scale.
• In equipercentile equating, scores having equivalent percentile positions in comparable candidate groups are treated as corresponding scores.

Conditions and limitations

Normalisation improves fairness only when sessions measure the same abilities and candidate groups are reasonably comparable. It reduces session-related variation but cannot correct defects in the examination itself.

• Candidates should be allocated across sessions through a fair or random process, with adequate group sizes and uniform testing conditions.
• Question papers should follow a common syllabus, blueprint and level of intended competence.
• Normalisation cannot remedy malpractice, serious syllabus mismatch, defective questions or systematic clustering of stronger candidates in one session.
• The formula, rounding rules and treatment of ties should be transparent because they can affect final rankings.

How UPSC asks this