Forced Ranking: The Confusion Between Absolute and Relative Performance
Forced ranking can confuse performance relative to teammates with performance against a standard. Requiring a fixed distribution can also impose distinctions that a small or uniformly performing team does not support.
THE IDEA
Performance ratings can compare work with an absolute standard or compare employees with one another. Ranking on separate dimensions can make comparison more structured, but relative rank does not show whether anyone meets an external standard. Requiring every team to fit a fixed distribution creates further problems.
Two Issues with Forced Ranking
-
Absolute and relative performance answer different questions. A fixed share rated as "failing to meet expectations" can misrepresent a team in which everyone meets the stated expectations. Conversely, being best in a weak team need not mean meeting the standard.
-
Small teams need not mirror a population distribution. Even if performance is normally distributed in a larger population, a small team may not have that shape. Teams are also selected rather than randomly assembled. Forcing distinctions among indistinguishable performances can add arbitrary variation.
Reference cases and frame-of-reference training offer ways to clarify what a rating means. Comparative judgment is useful when the comparison matches the decision; it does not justify imposing the same distribution on every group.
When to use it
Performance appraisals in large organizations where forced ranking is currently used.
Evaluating employees for promotion when only a fixed number of slots exist (e.g., colonels to generals).
Designing a noise-reduced evaluation system that combines relative comparisons across dimensions.
Training raters to use case scales for more reliable judgments in professional services.
FROM THEORY TO EVERYDAY LIFE
In practice
A High-Performing Mission Team
Suppose all astronauts on a successful mission meet the stated performance standards. A policy requiring 10% to be rated unsatisfactory would confuse a relative position with failure to meet those standards.
Undifferentiated Team
A supervisor has five employees whose performance is indistinguishable. Forced ranking requires a differentiated distribution, but there is no meaningful difference, so the ranking arbitrarily adds noise and unfairness.
A MOMENT TO REFLECT
Put it to work.
How might forcing a distribution on a small team lead to erroneous conclusions about individual performance?
In what situations is relative ranking more appropriate than absolute rating?
Why does forcing a rank order on indistinguishable individuals increase noise rather than reduce it?
Often confused with
References
- Noise: A Flaw in Human Judgment — Daniel Kahneman, Olivier Sibony, Cass R. Sunstein (2021)
CONTINUE THE CONVERSATION
Source texts
- Noise A Flaw in Human Judgment (Daniel Kahneman, Olivier Sibony etc.)