Rensberry.com
To normalize judge scores on a 1–10 scale and reduce the impact of extreme or biased scoring, z-score standardization is a widely used and effective method.
Z-score is preferred in judging contexts for its robustness.
It adjusts scores by removing judge-specific bias (such as consistently high or low scoring) by transforming each judge’s scores to have a mean of 0 and a standard deviation of 1. This ensures fairer comparisons across judges while preserving the relative ranking of performances.
For each judge, transform their scores so that their average becomes 0 and standard deviation becomes 1, using the formula:
Z = (X - u)/o
Where:
This adjusts for differences in judge leniency or strictness while preserving the relative ranking of performances.
After standardization, final scores are computed by averaging the z-scores across judges for each performance.
Alternative methods like min-max normalization (scaling to a fixed range) exist but are more sensitive to outliers.
Z = (X - u)/o