Calculate Sensitivity: Formula and Worked Example

Calculate sensitivity by dividing the change in an output by the change in the input that caused it: sensitivity = Δoutput / Δinput. Enter the two output values and the two corresponding input values to get the average rate of change, with units that show how much the output changes per unit of input.

Formula

The standard finite-change formula is:

Sensitivity = Δy / Δx = (y₂ − y₁) / (x₂ − x₁)

• x₁ and x₂ are the initial and final input values, respectively, in the same input unit.
• y₁ and y₂ are the output values corresponding to those inputs, in the same output unit.
• Δx = x₂ − x₁ is the input change.
• Δy = y₂ − y₁ is the output change.

The result is an average sensitivity over the interval from x₁ to x₂. Its unit is output units per input unit, such as dollars per item, meters per second of time, or degrees Celsius per unit of a control setting. The calculation is valid when the two output observations correspond to the two stated input values.

Worked Example

Use the calculator’s defaults: output changes from 10 to 16 units while input changes from 2 to 5 units.

1. Find the output change: Δy = 16 − 10 = 6 output units.
2. Find the input change: Δx = 5 − 2 = 3 input units.
3. Divide: sensitivity = 6 / 3 = 2 output units per input unit.

The helper results are Δy = 6 and Δx = 3. The calculator rounds displayed values to two decimal places, so the headline is 2.00 output units per input unit. In this example, the output increased by an average of two units for each one-unit increase in input across the measured interval.

What Sensitivity Measures

Sensitivity describes how responsive one quantity is to changes in another. It is a slope: the rise in output divided by the run in input. A positive result means the output rose as the input rose over the interval; a negative result means it fell. A result of zero means the two observed output values are equal, though it does not prove the output stayed unchanged between them.

This basic calculation is useful when comparing measurements, interpreting an experiment, or describing how a model responds to a parameter. For example, if a device’s measured output changes from 3 volts to 4 volts as an input changes from 1 to 3 units, average sensitivity is (4 − 3) / (3 − 1) = 0.5 volts per input unit.

It is important to distinguish this slope from percentage or elastic sensitivity. The slope has physical units; percentage sensitivity is usually defined as a percentage change in output divided by a percentage change in input. Those are different measures and can produce different numerical values.

How to Use and Interpret the Result

Choose two input values and the output measured at each one. Keep the pairing clear: y₁ must belong to x₁, and y₂ to x₂. Enter the values in the same units within each category. The input and output units may differ, as they normally do, but the final result should name both.

• Positive sensitivity: the output is higher at the second point.
• Negative sensitivity: the output is lower at the second point.
• Larger absolute value: the output changes more per input unit over the chosen interval.
• Near zero: the observed output changes little relative to the input change, subject to measurement precision.

For comparisons, use consistent units and comparable intervals. A slope measured in centimeters per second cannot be directly compared with one measured in inches per minute without conversion. Convert one result first, or convert the original measurements into common units before calculating.

Local Sensitivity and Limits

The two-point calculation gives an average slope, not necessarily the sensitivity at every point. If the relationship is curved, the slope can vary with the input. For a differentiable function y = f(x), the local sensitivity at a particular input is the derivative dy/dx at that point. A two-point estimate over a narrow interval can approximate that local slope, but only when the measurements are sufficiently reliable and the interval is appropriate.

When more than two observations are available, a scatter plot can reveal whether a straight-line description is reasonable. A fitted linear model estimates an overall slope across the data; it is not identical to calculating the slope between any chosen pair. In experimental work, repeated readings, instrument precision, and uncertainty in both input and output can affect the interpretation. The calculator reports the arithmetic result, not an uncertainty estimate or proof of cause and effect.

Common Calculation Problems

• Reversing only one difference: if you calculate y₁ − y₂, calculate x₁ − x₂ too. Reversing both preserves the slope; reversing just one changes its sign.
• Using mismatched observations: pair each output with the input at which it was measured.
• Dividing by zero: when x₂ = x₁, the input change is zero and this formula cannot produce a finite sensitivity. Use observations with different input values or a suitable model.
• Confusing slope with percentage change: Δy/Δx is not a percentage unless an additional, explicitly defined normalization is applied.
• Over-rounding: retain precision in intermediate calculations and round the displayed result at the end. Extra decimal places do not make imprecise measurements more accurate.

Frequently Asked Questions

How do you calculate sensitivity?
Subtract the initial output from the final output, then divide by the final input minus the initial input: (y₂ − y₁)/(x₂ − x₁).

Can sensitivity be negative?
Yes. A negative result means the output decreased as the input increased over the selected interval. Check that the input-output pairs and subtraction order are correct before interpreting the sign.

What does a sensitivity of 2 mean?
It means the output changed by an average of two output units for every one input unit across the interval. The unit label matters; “2” alone is incomplete.

What is the difference between sensitivity and percentage sensitivity?
Ordinary sensitivity is the output change divided by the input change and has units. Percentage sensitivity compares relative or percentage changes, so its definition and value differ.

What if the input change is zero?
The denominator is zero, so the two-point slope is undefined. You need different input values or a method that models the relationship using other information.

Related

• Confusion Matrix Accuracy: Formula and Example
• Box Dimensions Calculator: Find Volume and Fit
• Box Plot Maker Calculator: Build and Read a Box Plot

in y1 | Output before | units | 10 | 0
in y2 | Output after | units | 16 | 0
in x1 | Input before | units | 2 | 0
in x2 | Input after | units | 5 | 0
out sensitivity | Average sensitivity | output units/input unit | (y2 - y1) / (x2 - x1) | 2
out dy | Output change | output units | y2 - y1 | 2
out dx | Input change | input units | x2 - x1 | 2