How this instrument works
Average rate of change asks a single question about a function: over the stretch from x₁ to x₂, how much did the output move for each unit the input moved? Divide the total change in f — that is f(x₂) minus f(x₁) — by the total change in the input, x₂ minus x₁, and what falls out is a slope: rate = (f(x₂) − f(x₁)) ⁄ (x₂ − x₁). The arithmetic matches the slope between two coordinate points, but the two points here are not arbitrary dots; they are readings of one function taken at two different inputs, which is what lets the answer be read as a property of that function rather than of two points on a page.
Geometrically, the pair (x₁, f(x₁)) and (x₂, f(x₂)) sit on the curve of f, and the line joining them is a secant — from the Latin for cutting, because it slices through the curve rather than grazing it. This rate is exactly that secant's slope. Shrink the gap between x₁ and x₂ toward zero and the secant swings until it lies flush against the curve at a single point; that limiting position is the instantaneous rate of change, the derivative. Average rate of change is the derivative's algebra-only ancestor — no limits required, just two readings and a subtraction.
One case is worth sitting with: the average can read zero even when the function was never flat. A ball thrown straight up leaves your hand and returns to the same height a few seconds later, so f(x₁) equals f(x₂) and the average rate of change over that flight is exactly 0, despite the ball climbing and falling the entire time. The figure describes the net change across the whole span, not the motion inside it. The Mean Value Theorem rescues some meaning here — for a smooth enough function it guarantees at least one instant inside the interval where the instantaneous rate matches the average exactly, though it stays silent about every other instant.
- Enter the earlier input into x₁ and the function's value there into f(x₁).
- Enter the later input into x₂ and its matching output into f(x₂); x₂ must exceed x₁ or the sheet flags it.
- Read Average rate of change: the net change in output per unit of input across that interval.
- A positive result means f rose overall on the interval, negative means it fell, and zero means the two endpoints matched exactly.
Worked example — from (1, 2) to (4, 11)
A cyclist's odometer reads f(x) kilometres at hour x. At x₁ = 1 hour the odometer reads f(x₁) = 2 km; by x₂ = 4 hours it reads f(x₂) = 11 km. Average rate of change: rate = (11 − 2) ⁄ (4 − 1) = 9 ⁄ 3 = 3 — the cyclist covered ground at an average of 3 kilometres per hour across those three hours, regardless of how the pace actually varied inside them.
These same two readings also happen to satisfy f(x) = 3x − 1 exactly — check it: 3(1) − 1 = 2 and 3(4) − 1 = 11. If that straight line were the whole story, 3 km/h would be the pace at every instant, not merely the average. But the calculation only ever certifies the net figure: a cyclist who sprinted to 6 km/h through the middle hour and coasted at the edges would produce the identical rate of 3, because average rate of change reports the trip's overall balance, not its interior wobble.
Questions
How is average rate of change different from slope?
Arithmetically they are the same division — change in output over change in input — but average rate of change is reserved for two points that come from evaluating one function, f(x₁) and f(x₂), rather than two arbitrary coordinates. Call it slope when you mean a straight line's steepness; call it average rate of change when the two points are readings of a function that might curve between them.
How does average rate of change relate to the derivative?
It is the derivative's algebraic ancestor. Shrink the interval by letting x₂ approach x₁, and the secant slope (f(x₂) − f(x₁)) ⁄ (x₂ − x₁) tends toward the instantaneous rate of change at that point — the derivative. Average rate of change needs only subtraction and division; the derivative needs a limit, the one extra idea calculus adds on top.
Can the average rate of change be zero even though the function moved?
Yes. Zero only certifies that f(x₁) and f(x₂) match — it says nothing about the path between them. A thrown ball returns to launch height with an average vertical rate of change of exactly zero over that flight, despite rising and falling the whole time. Equal endpoints, whatever the interior detour, always average to zero.
What does the Mean Value Theorem guarantee about this number?
For a function continuous on [x₁, x₂] and differentiable inside it, the theorem guarantees at least one point c strictly between x₁ and x₂ where the instantaneous rate equals this average exactly. It does not say where c sits, and there can be more than one — only that the average is never a value the function's own slope fails to hit somewhere inside the interval.
What's the most common mistake when computing this by hand?
Subtracting the pair in mismatched order — putting f(x₁) − f(x₂) on top while leaving x₂ − x₁ on the bottom, which silently flips the sign of the result. Keep the subtraction order identical in numerator and denominator: both 'later minus earlier,' or both 'earlier minus later,' never one of each.
Does it matter if the function is a straight line?
If f is linear, the average rate of change is the same number no matter which two points you pick — it equals the line's constant slope everywhere, and the Mean Value Theorem's promised point c could sit anywhere in the interval, since the instantaneous rate never varies. Curvature is what makes the choice of interval start to matter.