School & Study FIELD NOTES · 4 MIN READ
Grade Curving: Flat, Linear and Top-Score Methods
The three standard ways to curve a set of scores, what each does to the class distribution, and how to defend the choice.
The exam is marked, the mean is 63, and half the class is below the pass line the syllabus promised would be reasonable. Now what? "Curving" is the family of fixes for that moment — and it is not one method but three quite different ones, each with its own arithmetic and its own fairness argument.
Why curves exist at all
A curve is an admission that the assessment missed its target. If an exam was meant to discriminate across the ability range but everyone scored 40-60%, the exam measured something other than intended — question difficulty, time pressure, a misjudged syllabus area. Curving re-centres the results without the cost and stress of re-examining anyone. It is a correction of the instrument, applied to the readings.
What a curve is not is a guarantee. Most institutional policies treat curving as an instructor judgement, and nearly all of them forbid curves that lower grades without warning. That constraint shapes which methods are popular in practice.
The flat add: the workhorse
The flat add gives every student the same N points, capped at the maximum. If the top score is 82 and the exam is out of 100, adding 18 points lifts the top to exactly 100 and everyone else by the same amount. Its virtues are transparency and fairness in the plainest sense: identical treatment, ranking preserved exactly, and the improvement is the same for every paper.
Its weakness is the mirror of its virtue: it helps strong and weak students equally, so it cannot fix a distribution that is too spread out — a class that is both too low on average and too wide stays that shape, just shifted upward.
The linear rescale: stretching the range
A linear rescale maps the current [lowest, highest] range onto a new target range — by default 0 to 100. Every score x becomes (x − min) × (targetMax − targetMin) ÷ (max − min) + targetMin. The lowest scorer lands on the target minimum, the highest on the target maximum, and everyone between scales proportionally.
Compared with the flat add, the rescale redistributes help: gains are largest at the bottom and smallest at the top. That is either its selling point (it rescues the struggling half of the class) or its unfairness (it compresses differences at the top, where a 95 and an 88 can both land near the ceiling). Choosing custom targets softens it — a 70-90 range, for example, lifts the floor without pretending everyone was close to perfect.
The degenerate case matters: if every student scored identically, there is no range to stretch, and the sensible output is the midpoint of the target range, not a division by zero.
The top-score adjust: the top sets the scale
The third method multiplies every score by target ÷ top score, so the best paper in the room becomes exactly 100 (or the chosen maximum) and everyone else scales up in proportion. An 80 in a class whose top is 92 becomes 87; a 70 becomes 76.1. This is the classic "the top scorer sets the scale" curve, beloved for its implicit fairness claim: the gap between any two students widens or narrows in exact proportion to their original gap.
It has a sharp edge, though. If the top score is already 100, the curve is a no-op and a struggling class gets nothing. And if the top score is low — 65 on a hard exam — the curve multiplies everyone by 1.54, which can push mediocre papers surprisingly high. Scaling amplifies, in every direction.
Choosing — and defending — a method
The honest comparison is made on before-and-after statistics: min, max and mean of the class before and after each candidate curve. A flat add that moves the mean from 63 to 68 is easy to defend in a faculty meeting; a linear rescale that lifts the bottom scorer from 30 to 70 needs a sentence of explanation. Recording the deltas per student — exactly who gained what — turns "I curved it" into an auditable decision, which is the difference between a curve and a mystery.
Try it yourself
The Grade Curve Calculator applies all three methods to the same list of scores side by side: paste the scores, pick flat, linear or top-score, and see every curved result, its delta, and the class stats before and after — with the linear method's slope and intercept shown so the transform is fully auditable. To place an individual curved score in the class distribution, use the Percentile Rank Calculator; to convert the curved percentage into a 4.0-scale grade, the GPA Scale Converter bridges the systems.