What “average rate of change” measures
When one quantity depends on another, we often want to describe how fast the output changes as the input changes. The average rate of change answers this question over a specific input interval. It compares two output values and spreads that change across the input change, producing a single number that summarizes the behavior on that interval.
If a function gives an output y for each input x, then between two input values x=a and x=b (with a≠b), the average rate of change is:
Average rate of change from a to b = (f(b) - f(a)) / (b - a)This is the same calculation you use for a slope between two points. The two points are (a, f(a)) and (b, f(b)). The average rate of change is the slope of the straight line that connects them.
Why slope is the right “speed” idea
Think of “change” as a ratio: output change per input change. If the input is time and the output is distance, then the ratio is distance per time, which is speed. If the input is hours studied and the output is test score, then the ratio is points per hour. If the input is number of items produced and the output is total cost, then the ratio is dollars per item.
In every case, the average rate of change tells you: “For each 1 unit increase in x, how many units does y change on average over this interval?”
Continue in our app.
You can listen to the audiobook with the screen off, receive a free certificate for this course, and also have access to 5,000 other free online courses.
Or continue reading below...Download the app
Interpreting the sign and size
Positive, negative, and zero rates
- Positive average rate of change: f(b) > f(a). Output increases as input increases on that interval. The secant line slopes upward left-to-right.
- Negative average rate of change: f(b) < f(a). Output decreases as input increases. The secant line slopes downward left-to-right.
- Zero average rate of change: f(b) = f(a). Output ends where it started; the secant line is horizontal.
Magnitude: how fast, not just which direction
The absolute value tells how strongly the output changes per unit input. A slope of 10 means “about 10 units of output per 1 unit of input,” while a slope of 0.2 means “only 0.2 units of output per 1 unit of input.”
Units matter
Average rate of change always has units of “output units per input unit.” This is not decoration; it is part of the meaning. If f(t) is measured in meters and t in seconds, then (f(b)−f(a))/(b−a) is meters per second.
Computing average rate of change: a step-by-step method
Use the same reliable procedure every time:
Step 1: Identify the interval
Decide the two input values a and b. Be careful about order; the formula works either way, but the sign will reflect direction.
Step 2: Evaluate the function at the endpoints
Compute f(a) and f(b). If you have a table, read them. If you have a formula, substitute. If you have a word problem, interpret what the function outputs.
Step 3: Compute the changes
Find the output change Δy = f(b) − f(a) and the input change Δx = b − a.
Step 4: Divide
Average rate of change = Δy/Δx.
Step 5: Attach units and interpret
State the result as “(output units) per (input unit)” and explain what it means in context.
Example 1: Distance traveled over time (average speed)
Suppose a car’s distance from home after t hours is given by f(t) = 50t + 20, where f(t) is in miles. Find the average rate of change from t=1 to t=4.
Step-by-step
- Interval: a=1, b=4
- Evaluate endpoints: f(1)=50(1)+20=70 miles; f(4)=50(4)+20=220 miles
- Changes: Δy = 220−70=150 miles; Δx = 4−1=3 hours
- Divide: 150/3 = 50
The average rate of change is 50 miles per hour. Because the function is linear, this is also the constant rate at every moment, but the key point here is that the average rate of change summarizes the interval.
Example 2: A non-linear situation (average rate depends on the interval)
Let f(x)=x^2 represent some quantity (for example, area of a square with side length x, measured in square units). Compute the average rate of change from x=1 to x=3, and then from x=3 to x=5.
From 1 to 3
- f(1)=1, f(3)=9
- Δy = 9−1=8, Δx = 3−1=2
- Average rate = 8/2 = 4
From 3 to 5
- f(3)=9, f(5)=25
- Δy = 25−9=16, Δx = 5−3=2
- Average rate = 16/2 = 8
The average rate of change is larger on the later interval. This captures an important idea: for many relationships, the “speed of change” itself changes as x changes.
Secant lines: slope between two points
The line through two points on a graph is called a secant line. Its slope is exactly the average rate of change between those two x-values. Even if you are not drawing, it helps to think: “I’m finding the slope of the line connecting (a,f(a)) and (b,f(b)).”
Given two points (x1, y1) and (x2, y2), the slope is:
m = (y2 - y1) / (x2 - x1)In function language, y1=f(x1) and y2=f(x2), so it becomes (f(x2)−f(x1))/(x2−x1).
Common pitfalls with slope calculations
- Mixing up the order: If you compute (y1−y2)/(x2−x1), you will get the negative of the correct slope. Keep the same order in numerator and denominator.
- Forgetting units: The number alone can be misleading; always interpret as “output per input.”
- Using points not on the function: If the values come from a table or a story, verify you are pairing the correct x with the correct f(x).
Average rate of change from a table
Often you are given data rather than a formula. The average rate of change is still computed the same way: pick two rows, subtract outputs, subtract inputs, divide.
Example table:
x: 0 2 5 7 10 (hours of operation) y: 100 130 190 230 310 (units produced)Find the average rate of change from x=2 to x=7.
- f(2)=130, f(7)=230
- Δy = 230−130=100 units
- Δx = 7−2=5 hours
- Average rate = 100/5 = 20 units per hour
Interpretation: between 2 and 7 hours, production increased at an average of 20 units per hour.
When x-steps are uneven
Notice the x-values in the table are not evenly spaced (0 to 2 is 2 hours, 2 to 5 is 3 hours, etc.). That does not matter. Average rate of change always uses the actual Δx of the chosen interval.
Average rate of change from a word problem
In applications, you may not be told “here is f(x).” Instead, you are told two measurements and asked for an average rate. The structure is the same: change in the measured quantity divided by change in the input variable.
Example: Temperature change
A cup of coffee cools from 90°C to 70°C over 10 minutes. What is the average rate of change of temperature with respect to time?
- ΔT = 70−90 = −20°C
- Δt = 10 minutes
- Average rate = −20/10 = −2°C per minute
The negative sign is meaningful: temperature is decreasing. The coffee cools at an average of 2°C per minute over that time period.
Comparing average rates on different intervals
Average rate of change is interval-based, so two different intervals can produce different averages even for the same function or the same real-world process. This is especially important when the process speeds up or slows down.
Example: Revenue vs. price (a simple model)
Suppose revenue (in dollars) from selling a product at price p (in dollars) is modeled by R(p)=200p−5p^2. Compute the average rate of change from p=10 to p=12.
- R(10)=200(10)−5(100)=2000−500=1500
- R(12)=200(12)−5(144)=2400−720=1680
- ΔR = 1680−1500=180
- Δp = 12−10=2
- Average rate = 180/2 = 90 dollars per dollar
Interpretation: increasing price from $10 to $12 increases revenue by an average of $90 for each $1 increase in price over that interval. This does not mean revenue increases by $90 for every $1 at all prices; it is specific to that interval.
Average rate of change as a “per-unit” comparison
Sometimes the most useful interpretation is as a per-unit conversion factor. For example:
- If cost changes by $60 when quantity changes by 15 items, then average cost increase is $4 per item.
- If altitude changes by −300 meters over 2 kilometers of horizontal distance, then average slope is −150 meters per kilometer.
This “per-unit” view helps you compare situations with different interval lengths. A total change of 100 might be large or small depending on whether it happened over 1 unit of input or 100 units of input.
Connecting average rate of change to linear models
Average rate of change is the key ingredient in building a simple linear approximation between two data points. If you know the function values at a and b, you can create a line that matches the function at those endpoints. That line’s slope is the average rate of change.
The point-slope form of the secant line through (a, f(a)) and (b, f(b)) is:
y - f(a) = [(f(b) - f(a)) / (b - a)] (x - a)This line can be used to estimate intermediate values when the function is not too curved on the interval. The estimate will not be perfect for a nonlinear function, but it is often a useful first approximation.
Example: Estimating with a secant line
Let f(x)=x^2. Use the secant line from x=2 to x=4 to estimate f(3).
- f(2)=4, f(4)=16
- Slope m = (16−4)/(4−2)=12/2=6
- Secant line through (2,4): y−4 = 6(x−2)
- At x=3: y−4 = 6(1)=6, so y=10
The true value is f(3)=9, so the secant-line estimate gives 10. The difference reflects the curvature of x^2 on that interval.
Average rate of change and “steepness” in context
In everyday language, people say a graph is “steeper” in some region. Average rate of change gives a precise way to quantify steepness over an interval. A larger positive slope means steeper upward change; a more negative slope means steeper downward change.
Example: Road grade
Road grade is often described as a percent: (rise/run)×100%. If a road rises 30 meters over a horizontal run of 600 meters, then:
- Average rate of change = 30/600 = 0.05 meters per meter
- Grade = 0.05×100% = 5%
This is an average over that segment of road. Different segments can have different grades.
Choosing intervals wisely
Because average rate of change depends on the interval, the choice of a and b should match the question you are trying to answer.
- To describe a trend over a long period, choose endpoints far apart (for example, average population change per year over a decade).
- To describe local behavior, choose endpoints close together (for example, average speed over a short time window).
- To compare two phases, compute averages on separate intervals (for example, cooling rate in the first 5 minutes vs. the next 5 minutes).
In data analysis, it is common to compute several average rates over consecutive intervals to see whether the process is accelerating or decelerating.
Practice pattern: compute and interpret
Use this pattern to build skill and avoid “formula-only” answers.
Prompt A
A tank contains 500 liters of water at noon and 380 liters at 3 pm. Compute the average rate of change of water volume with respect to time.
Work it out
- ΔV = 380−500 = −120 liters
- Δt = 3 hours
- Average rate = −120/3 = −40 liters per hour
Interpretation: the tank is losing water at an average of 40 liters per hour over that time.
Prompt B
For f(x)=3x^2−2x, compute the average rate of change from x=1 to x=4.
Work it out
- f(1)=3(1)−2(1)=1
- f(4)=3(16)−2(4)=48−8=40
- Δy = 40−1=39
- Δx = 4−1=3
- Average rate = 39/3 = 13
Interpretation: over x from 1 to 4, the output increases by an average of 13 units for each 1 unit increase in x.
Reading average rate of change from two points on a graph
If you are given a graph and can identify two points on the curve, you can compute the average rate of change without knowing the formula. The steps are the same: read coordinates, compute (y2−y1)/(x2−x1). In practice, you should choose points that are clearly marked or that lie on grid intersections to reduce reading error.
Handling approximate readings
Graph readings are often approximate, so your average rate will be approximate too. You can reflect that by rounding reasonably and stating “approximately.” For example, if you read f(2)≈5.1 and f(6)≈12.9, then:
Average rate ≈ (12.9 - 5.1) / (6 - 2) = 7.8 / 4 = 1.95The interpretation remains the same, but you should not claim more precision than the graph supports.
What average rate of change prepares you for next
Average rate of change is the bridge between “change over an interval” and “change at an instant.” By getting comfortable with slopes of secant lines and with interpreting units, you build the intuition needed to talk about how fast something is changing at a specific input value, not just across a span.