Percentage Change vs. Difference vs. Percentage Points
Take two numbers: 40 and 50. Depending on which comparison you reach for, the gap between them is +25%, −20%, 22.22%, or — if they happen to be percentages themselves — 10 points. All four are correct answers to different questions, and most percentage arguments online are two people answering different questions at each other.
Percentage change: one number is the baseline
Question it answers: relative to where it started, how far did the value move?
Formula: (new − old) ÷ old × 100.
From 40 to 50 is +25%; from 50 back to 40 is −20%. Both statements describe the same pair of numbers, and they differ because the base differs — each change is measured against its own starting point. That base-dependence is not a flaw; it’s the meaning. It’s also why changes don’t cancel: +25% then −20% happens to return exactly to 40 (1.25 × 0.8 = 1), but +25% then −25% does not (1.25 × 0.75 = 0.9375).
Use percentage change whenever time or causality orders your numbers: last month’s rent vs. this month’s, before vs. after. The percentage change calculator computes it with the sign, and documents its convention for negative starting values (divide by |old|, so the sign always means direction).
Percentage difference: neither number is the baseline
Question it answers: how far apart are these two values, as a share of their typical size?
Formula: |a − b| ÷ ((a + b) ÷ 2) × 100.
For 40 and 50: 10 ÷ 45 × 100 = 22.22%, and the answer is the same in either order — that symmetry is the whole point. Science and engineering use it to compare two measurements of the same thing; shoppers can use it to compare two quotes when neither is “the original.”
The classic misuse is applying it where a baseline exists. Comparing this year’s revenue to last year’s with the symmetric formula quietly understates growth (the average base is inflated by the larger new value). Timeline → change. Peers → difference. The percentage difference calculator also explains why it refuses opposite-sign pairs instead of printing a meaningless number.
Percentage points: when the numbers already wear % signs
Question it answers: what is the absolute gap between two percentages?
Formula: subtraction. Nothing else.
A mortgage rate going from 4% to 5% rose 1 percentage point — and 25 percent, since (5 − 4) ÷ 4 × 100 = 25. Both are true; they are wildly different sizes; and news copy routinely writes “rates rose 1%” for the first meaning, which a literal reader must interpret as 4% → 4.04%. The ambiguity runs one level deeper with “basis points” (hundredths of a point, so 4% → 5% is 100 bps), which finance adopted precisely to end the confusion.
The percent-of-a-percent trap peaks in headlines about risk: “risk up 50%!” may mean 2 in 1,000 became 3 in 1,000 — a change of 0.1 percentage points. Neither framing is dishonest on its own; quoting the relative change without the base is what makes it misleading.
Why we tangle these
This isn’t merely sloppiness. The education literature’s deep survey of percent — Parker and Leinhardt (1995) — traces how percent language compresses a ratio, a base and an operator into one small symbol, with the base usually left implicit. Change, difference and points are three different choices of base (the old value, the average, and “none — use absolute units”), but ordinary sentences give you no slot to say which one you picked. The math is easy; the reference is what goes missing.
The two-question test
Before writing or trusting any “X%” comparison:
- Of what? Name the base. Old value → change. Average → difference. The numbers are already percentages → say points.
- Does the claim survive being restated as counts? “From 40% to 45% of 2,000 responses” is 800 → 900 people. If the restated version feels smaller than the headline, the headline was leaning on the ambiguity.
And when the numbers matter, skip the mental step: all three comparisons sit in the calculator as their own rows, computed exactly.