The average value of a function over an interval is the signed integral divided by the interval’s width. For a continuous function f on [a,b], with a < b, its average value is (1/(b − a)) times the definite integral of f(x) from a to b. It is not generally the average of the two endpoint outputs, and it is not the value of the function at the midpoint.
For example, f(x) = x² + 2x + 1 over 0 ≤ x ≤ 2 has definite integral 26/3. The interval width is 2, so the average function value is 13/3. The midpoint value is 4 and the mean of the two endpoint values is 5—both different. This guide explains the calculation, signed averages, area interpretation, parameter recovery and why quadratic curvature creates a predictable difference from midpoint estimates.
What average value means for a changing function
The arithmetic mean of several recorded numbers is their sum divided by how many numbers are recorded. A function on a continuous interval has infinitely many possible input points, so averaging a finite selection of outputs is not the same mathematical operation.
A definite integral combines the continuously varying function values across an interval. Dividing that signed accumulation by the interval’s width gives the average function value per unit of input.
The result is one number with the same units as f(x), assuming the input variable has its stated physical units. It represents the height of a constant function that would produce the same signed integral over the same interval.
This interpretation connects integration to the familiar idea of average without replacing the continuous calculation by a few arbitrarily chosen sample values.
Write the definition with its correct interval width
For an integrable function f on [a,b] with a < b, write f_avg = (1/(b − a))∫_a^b f(x) dx.
The denominator b − a is the length of the input interval. It is not the upper limit b by itself, the number of terms in a polynomial or the number of ticks marked on an axis.
If the interval is [0,2], its width is 2. If it is [3,7], its width is 4. If the interval is [−2,2], its width is also 4.
The most reliable exam habit is to write the interval width separately before integrating. A correct antiderivative divided by the wrong width still produces an incorrect average.
Work the opening quadratic example
Let f(x) = x² + 2x + 1 over 0 ≤ x ≤ 2. Its antiderivative is F(x) = x³/3 + x² + x.
Evaluate the definite integral: F(2) − F(0) = 8/3 + 4 + 2 − 0 = 26/3.
The interval width is 2 − 0 = 2. Therefore the average value is (26/3)/2 = 13/3.
This exact value is approximately 4.333. The computation has two distinct steps: finding signed accumulation and dividing by the interval width. The definite integral itself, 26/3, is not the requested average.
Verify the answer with a constant-height rectangle
A constant function with height 13/3 over an interval of width 2 has area 2(13/3) = 26/3.
That is exactly the definite integral of f(x) = x² + 2x + 1 from 0 to 2. The constant-height rectangle therefore has the same accumulated signed area as the region under this positive curve.
This is the geometric meaning of the mean value. It does not say that the original curve is constant or that its actual graph is a rectangle.
The rectangle interpretation works directly as ordinary geometric area when the function is non-negative. For a function that changes sign, the same definition still gives a signed average, but the geometric picture must account for negative contributions.
Why the midpoint value is not the average
The midpoint of [0,2] is x = 1. At that input, f(1) = 1 + 2 + 1 = 4.
The actual average is 13/3, about 4.333. Thus evaluating at the midpoint gives a value that is too small for this upward-curving quadratic.
The discrepancy is not a numerical accident. A curved function does not generally spend equal amounts of input interval at equal distances in output from its midpoint height.
A student who writes f_avg = f((a+b)/2) without further justification is using a midpoint approximation rather than the exact average-value definition. For some special functions it works exactly, but the general formula is based on the integral.
Why the endpoint average is also different
The endpoint outputs are f(0) = 1 and f(2) = 9. Their arithmetic mean is (1 + 9)/2 = 5.
That is larger than the true continuous average 13/3. The endpoint mean pays attention only to the first and last outputs, while the integral accounts for the function throughout the interval.
For this upward-opening quadratic, the straight chord connecting the endpoints lies above the curve between them. Its average height equals 5, which explains why the endpoint mean overestimates the continuous mean.
The endpoint average and midpoint value can still be useful numerical approximations, but a school question asking for an exact function average requires the integral unless a proven special formula applies.
A linear function is a special case
Let f(x) = mx + c on an interval [a,b]. The graph is a straight line and the output changes uniformly with x.
Its exact average over the interval is f((a+b)/2), which also equals [f(a)+f(b)]/2. The midpoint and endpoint averages agree for a linear function.
For example, f(x) = 2x + 1 on [0,2] has endpoint outputs 1 and 5, with mean 3. Its midpoint output is f(1) = 3.
Integration confirms this: ∫_0^2 (2x + 1) dx = 6 and dividing by 2 gives 3. This is why the midpoint shortcut may appear to work in easier questions and then fail when the function becomes curved.
For a quadratic, there is an exact midpoint correction
Let f(x) = Ax² + Bx + C on [a,b], with midpoint m = (a+b)/2 and width L = b − a.
The exact average satisfies f_avg = f(m) + A L²/12. The correction depends on the quadratic coefficient and the square of the interval width.
For the opening example A = 1 and L = 2, so the correction is 1 × 4/12 = 1/3. Since f(1) = 4, the exact average is 4 + 1/3 = 13/3.
This identity is useful enrichment because it makes the effect of curvature precise. An upward-opening quadratic has A > 0 and an average above the midpoint value, while a downward-opening quadratic has its exact average below the midpoint value.
Derive the midpoint correction through symmetry
Write x = m + u, so the interval becomes −L/2 ≤ u ≤ L/2. Expand a quadratic around its midpoint.
The function becomes f(m+u) = f(m) + (2Am+B)u + Au². Over a symmetric interval, the odd linear term in u integrates to zero.
The average of the constant term remains f(m). The average of u² over this interval is L²/12, obtained by integrating u² from −L/2 to L/2 and dividing by L.
Therefore f_avg = f(m) + A L²/12. This derivation explains both why the B coefficient does not appear explicitly in the correction and why widening the interval increases the curvature effect quadratically.
A different quadratic can have a signed average
Consider g(x) = x² − 4x + 3 over 0 ≤ x ≤ 4. Its antiderivative is x³/3 − 2x² + 3x.
At x = 4, the antiderivative is 64/3 − 32 + 12 = 4/3. At x = 0 it is zero, so the definite integral is 4/3.
Divide by the width 4 to obtain an average value of 1/3. The midpoint output is g(2) = 4 − 8 + 3 = −1.
The exact midpoint correction gives −1 + (1)(4²)/12 = −1 + 4/3 = 1/3, confirming the integral. Notice that the midpoint output is negative even though the signed average is positive.
The signed average is not always an average height above the axis
The quadratic g(x) = x² − 4x + 3 factors as (x − 1)(x − 3). It is positive on 0 ≤ x < 1 and 3 < x ≤ 4, but negative between 1 and 3.
Its definite integral across [0,4] is 4/3 because the negative middle contribution partially cancels the positive outer contributions.
The average value 1/3 is therefore a signed average. It is not the total geometric area divided by four.
This distinction matters whenever a graph crosses the x-axis. A positive average does not mean that every function value was positive, and a zero average does not imply that the graph enclosed no geometric region.
Compute the geometric area separately
For g(x) = x² − 4x + 3 on [0,4], use its zeros x = 1 and 3 to split the region.
The signed integrals over [0,1], [1,3] and [3,4] are respectively 4/3, −4/3 and 4/3. Taking their magnitudes gives total geometric area 4.
Dividing that total by width 4 gives the average of |g(x)|, which is 1. But the signed average of g itself is 1/3.
Both quantities are mathematically valid; they answer different questions. A student should not take the absolute value of the final signed integral when the task asks for average function value, nor ignore absolute values when the task asks for average magnitude or total area.
Average velocity and average speed are different
Suppose a signed velocity is v(t) = t − 1 over 0 ≤ t ≤ 2, in suitable distance-per-time units. Its definite integral is [t²/2 − t] from 0 to 2, which equals zero.
The average velocity is therefore 0/2 = 0. But the object is not stationary throughout: the velocity is negative before t = 1 and positive afterwards.
The total distance travelled is ∫_0^2 |t − 1| dt = 1, so the average speed is 1/2 over the two-unit interval.
This contrast is a valuable application of the sign distinction. Average velocity uses signed displacement per time, while average speed uses total distance per time. They are not interchangeable when direction reverses.
Average value is not average rate of change
For a differentiable function f on [a,b], its average rate of change is [f(b) − f(a)]/(b − a). That is a secant slope, not an average of the function’s output heights.
In the opening example f(x) = x² + 2x + 1 on [0,2], the average rate of change is [9 − 1]/2 = 4.
The average function value is 13/3. The two numbers differ because they describe different mathematical quantities.
If x represents time and f represents position, average rate of change has units of position per time. The average value of f has units of position. The units themselves provide a useful clue about which calculation a question is requesting.
Recover an unknown coefficient from an average value
Let f(x) = x² + kx + 1 over 0 ≤ x ≤ 2, with unknown real k. Suppose its average value is 13/3.
Compute the integral: ∫_0^2 (x² + kx + 1) dx = 8/3 + 2k + 2 = 14/3 + 2k.
Divide by width 2. The average is 7/3 + k. Set it equal to 13/3, giving k = 2.
The reconstructed function is f(x) = x² + 2x + 1, exactly the opening model. Substitute k = 2 into the original average formula to verify the result independently.
A parameter can disappear from an average
Consider f(x) = kx + 3 on the symmetric interval −1 ≤ x ≤ 1. The integral of the odd term kx is zero over symmetric limits.
The constant contributes 3 × 2 = 6, so the average function value is 6/2 = 3 for every real k.
If a question says this average is 3, it cannot determine a unique k. Every real k fits. If it says the average is 4, there is no k that can satisfy it.
This special case demonstrates why a single average condition may or may not determine an unknown parameter. Evaluate the expression first and check whether the parameter actually remains in the equation.
A constant function confirms the definition
If f(x) = c for every x in [a,b], then its integral is c(b − a). Dividing by b − a gives average value c.
For example, f(x) = 7 on [2,6] has integral 7 × 4 = 28 and average 28/4 = 7.
This is exactly what an average should do: a constant quantity’s mean equals that constant, regardless of the interval width.
The example also provides a quick dimension check. The integral has units of function output multiplied by input units; dividing by interval width returns the output units.
The mean value theorem for integrals gives a useful guarantee
If f is continuous on [a,b], it attains minimum and maximum values there. Its average value lies between those bounds.
A related result says that at least one c in [a,b] satisfies f(c) = f_avg. In other words, the continuous function reaches its own average output somewhere within the interval.
For the opening quadratic, f_avg = 13/3. Solving (x + 1)² = 13/3 gives x = −1 + √(13/3) within [0,2], approximately 1.082.
Thus the curve’s output equals its exact average at an input slightly to the right of the midpoint. That input need not be the midpoint and need not be unique for other functions.
An average stays between the function’s extremes
For f(x) = (x + 1)² on [0,2], the minimum value is f(0) = 1 and the maximum is f(2) = 9.
The calculated average 13/3 lies strictly between 1 and 9. This is a necessary consistency check for a continuous real-valued function averaged over a finite interval in the usual way.
If a student obtains an average of 12 for this function on [0,2], it is impossible because the function never reaches 12 on that interval.
Such range checks can expose an incorrect interval width, a missing power-rule divisor or a sign mistake before the answer is submitted. They are especially valuable in multi-step calculus problems.
An average can be zero even with a nonzero function
Take f(x) = x on [−1,1]. Its antiderivative is x²/2, so the integral from −1 to 1 is 1/2 − 1/2 = 0.
The average is therefore zero divided by width 2, which is zero. Yet f is nonzero at every point except x = 0.
The positive contributions to the right cancel the negative contributions to the left. More generally, an odd integrable function averaged over symmetric limits has average zero.
This does not mean the total geometric area is zero. It means the signed average vanishes. A good student can explain why both statements are simultaneously true.
Average value over a full sine period
For f(x) = sin x over 0 ≤ x ≤ 2π, the definite integral is [−cos x] from 0 to 2π, giving −1 − (−1) = 0.
The interval width is 2π, so the average value is zero.
The positive sine values over the first half-cycle cancel the negative values over the second half-cycle. The result reflects symmetry across the complete period.
If the function is instead f(x) = 3 + 2sin x, the sine component still averages to zero, so the mean is 3. This is an example of a baseline that remains after an oscillating component cancels.
Interval width matters even with the same integrand
Consider f(x) = x². On [0,1], its integral is 1/3 and its average is 1/3.
On [0,2], its integral is 8/3 and its average is (8/3)/2 = 4/3. On [0,3], its integral is 9 and its average is 9/3 = 3.
The function formula has not changed, but the averaging interval has. Larger allowed x-values contribute increasingly large outputs because x² grows.
Students should not carry an earlier average into a new interval without recomputing the bounds and width. The average is a property of both the function and the specified domain interval.
A short examination solution format
Start by writing f_avg = [∫_a^b f(x) dx]/(b − a), with a < b. Calculate the antiderivative accurately and evaluate upper minus lower.
Divide the exact integral by the interval width. Keep fractions exact unless the question requests a rounded value.
If the function changes sign, remember that this is a signed average, not total geometric area divided by width. If the task asks for average magnitude, speed or area, handle absolute values or regions as required.
Finally check that the average lies between the function’s minimum and maximum on the interval when such bounds are known. That simple check often catches a numerical result that cannot be correct.
Practice questions with checked answers
Question one: f(x) = x² + 2x + 1 on [0,2]. Average value: 13/3.
Question two: f(x) = 2x + 1 on [0,2]. Average value: 3.
Question three: g(x) = x² − 4x + 3 on [0,4]. Signed average: 1/3.
Question four: v(t) = t − 1 on [0,2]. Average velocity: 0; average speed: 1/2.
Question five: f(x) = x² + kx + 1 on [0,2] has average value 13/3. Answer: k = 2.
Question six: f(x) = sin x on [0,2π]. Average value: 0.
Question seven: f(x) = (x + 1)² on [0,2]. Find an input at which f equals its average. Answer: x = −1 + √(13/3), within the interval.
For each, identify whether the requested quantity is an average output, average rate, average magnitude or total area before calculating.
Syllabus suitability and further study
Average values from integrals connect signed accumulation, function ranges, rates of change and mathematical modelling. The mean-value-theorem and quadratic midpoint-correction discussions are extensions that should be selected according to the learner’s taught syllabus.
For Singapore, consult SEAB’s relevant 2026 O-Level Additional Mathematics, 2027 SEC G3 or 2027 SEC G2 syllabus lists.
Related guides include How to Find an Unknown Parameter from a Definite Integral, How to Find an Unknown Upper Limit of a Definite Integral and How to Check the Meaning of a Calculus Answer. Explore the Additional Mathematics archive or enquire about eduKate Singapore small-group tutorials near Sixth Avenue MRT.
The important skill is deciding what is being averaged and then computing the exact continuous mean over the specified interval, rather than relying on a single convenient point. Properly taught kids shine a bright light into the future.