Why teachers curve grades
A curve exists to separate how hard the test was from how well the class understood the material. If an exam turns out to be poorly calibrated — a question was ambiguous, the pacing was too tight, a concept wasn't covered as thoroughly as the test assumed — the raw scores end up measuring the test's difficulty as much as anyone's actual knowledge. Curving adjusts for that, so a 62% on a brutal exam and a 62% on a fair one don't get treated identically.
It isn't the same as making a test easier after the fact, and it isn't grade inflation by default — most curves simply shift or reshape the scores that already exist, using a fixed, consistent rule applied to every student the same way.
The 6 curve methods, side by side
| Method | What it does | Can it lower a score? |
|---|---|---|
| Add points (flat) | Adds the same fixed points to every score | No |
| Highest score → 100% | Scales every score up so the top score reaches 100% | No |
| Percentage boost | Increases every score by the same percentage of itself | No |
| Reach a target mean | Shifts every score by whatever amount brings the class average to a chosen number | No (if the target mean is above the original) |
| Square root | Boosts low scores far more than high scores | No |
| Bell curve | Reshapes the whole distribution around a new mean and spread | Yes |
Add points (flat)
This is the curve most people picture when they hear the word. Every student gets the same number of points added to their raw score, so the gap between any two students never changes — a student who scored 10 points ahead of a classmate before the curve is still exactly 10 points ahead after it. Because it's easy to explain and impossible to argue with, it's the default choice for most teachers.
Highest score → 100%
Instead of picking a number of points, this method finds whatever the top score in the class actually was and calculates the exact multiplier needed to bring that score up to 100%. Every other score gets multiplied by the same factor. It's a good fit when the highest-performing student clearly should have gotten a perfect score, but didn't — usually because a question was flawed or worth partial credit that wasn't awarded.
Percentage boost
This increases every score by the same percentage of its own value, rather than the same flat number of points. A 10% boost turns an 80% into 88% (+8 points) but only turns a 60% into 66% (+6 points) — higher scorers gain more raw points than lower scorers, which is the opposite direction from the square-root method below.
Reach a target mean
Here the teacher picks a target class average instead of a points value, and the calculator works backward: find the difference between the current average and the target, then add that difference to every score. If the class averaged 68% and the target is 78%, every student gets +10. It's functionally identical to the flat-points method — it just starts from the average you want instead of the points you want.
Square root
Taking the square root of a score (expressed as a fraction of the maximum, then rescaled back up) has an effect that flat-points curves don't: it helps struggling students far more than it helps students who were already doing well. A raw 50% climbs to roughly 71% — a 21-point jump — while a raw 90% only reaches about 95%, a 5-point jump. It's the method to reach for when the goal is specifically to help the bottom of the class, not shift everyone by the same amount.
Bell curve
A bell curve is a different kind of adjustment altogether. Instead of shifting or scaling every score by the same rule, it converts each score into a z-score — how many standard deviations it sits from the class average — and then maps that same relative position onto a brand-new average and spread that the teacher chooses. Because it forces the whole class onto a specific distribution shape, a student who was comfortably above the original average can end up with a lower curved score than their raw one, if the new target spread compresses the top of the range. It's the method to use when standardizing results across multiple sections of the same course, not for an everyday "the test was hard" curve.
Worked example
Ten raw scores from the same quiz: 80, 80, 75, 60, 60, 55, 50, 50, 40, 25 — a class average of 57.5%.
A flat +15 curve (enough to bring the top score to 95) lifts every student by the same 15 points — the lowest score moves from 25 to 40, the highest from 80 to 95, and every gap in between stays exactly the same size.
Targeting a new mean of 75 with a standard deviation of 12 tells a different story. The 55 — close to the original average — moves to about 73.2, while the 80 jumps further, to about 91.1, because it was already well above where most of the class landed. The ranking of students doesn't change, but the size of each move now depends on where a student sat relative to everyone else, not on a single flat number.
Which method should I use?
Reach for Add points or Highest score → 100% when a test was simply harder than intended and every student's ranking should stay exactly where it was — these two cover the overwhelming majority of real curving situations. Use Square root specifically when the goal is to help the weaker half of the class more than the stronger half. Save the Bell curve for standardizing results across multiple sections, TAs, or exam versions of the same course — it's the one method genuinely capable of moving a score in the wrong direction, so it needs a real reason, not a default habit.
Common questions
What does it mean to grade on a curve?
It means adjusting a whole class's scores based on how the class performed as a group, instead of leaving every score exactly as raw-graded. It's usually applied when a test turns out harder than intended, so the grades reflect understanding relative to the exam's difficulty, not just the raw percentage.
What is the most common way to curve grades?
Adding a flat number of points to every score, often enough to bring the highest score up to 100%. It's the most common method because it's the easiest to explain and it never changes the ranking between students.
Is a bell curve the same thing as curving grades?
A bell curve is one specific curving method, not the only one. It forces the class's scores onto a normal (bell-shaped) distribution with a chosen mean and spread, which is a much stronger adjustment than simply adding points — and it's the one method that can lower an above-average score, not just raise weaker ones.
Does curving grades help everyone equally?
No — it depends on the method. A flat-points curve helps every student by the same fixed amount. A square-root curve deliberately helps lower scores more than higher ones. A bell curve can help some students and hurt others, since it reshapes the whole distribution around a target average instead of just shifting it.
How do you decide how many points to curve?
The simplest approach: find the highest score on the test and add enough points to bring it to 100%, then apply that same number of points to everyone. Other approaches target a specific class average instead of the top score — see the target-mean method above.