Percentage Error & Measurement Error
Every measurement made in science, engineering, or everyday life carries some degree of uncertainty. Whether you are measuring the length of a table with a ruler or timing a race with a stopwatch, the result is never perfectly exact. Error analysis is the branch of mathematics that gives us precise tools to quantify, communicate, and work with these uncertainties in a rigorous way. Percentage error is the most common way to express how large an error is relative to the true or accepted value. A measurement error of 2 cm means very little on its own — it is devastating if you are measuring a 5 cm insect, but negligible if you are measuring a 500 m road. Percentage error rescales the raw error so that it is meaningful regardless of the size of the quantity being measured, making it possible to compare accuracy across completely different contexts. When a quantity of interest is calculated from several measured quantities — for example, finding the area of a rectangle from two measured lengths, or the speed of an object from measured distance and time — errors in the individual measurements combine and propagate into the final result. Understanding how errors propagate through mathematical operations is essential for scientists, engineers, and anyone who relies on quantitative reasoning to make sound decisions.
- Absolute Error
- Percentage Error
- Percentage Uncertainty in a Single Measurement
- Propagation of Errors — Addition and Subtraction
- Propagation of Errors — Multiplication and Division
- Propagation of Errors — Powers
Absolute Error
The absolute error (also called the absolute uncertainty) is simply the difference between a measured value and the true or accepted value: |measured − true|. It has the same units as the quantity being measured. For example, if the true length is 50.0 cm and you measure 49.3 cm, the absolute error is |49.3 − 50.0| = 0.7 cm. When a single measuring instrument is used repeatedly, the absolute error is often taken as half the smallest scale division (the resolution) of the instrument.
Percentage Error
Percentage error expresses the absolute error as a fraction of the true (or accepted) value, multiplied by 100. The formula is: percentage error = (|measured − true| / true) × 100%. This dimensionless quantity allows you to compare the accuracy of measurements of completely different physical quantities. A 2% error on a mass measurement is directly comparable to a 2% error on a time measurement, whereas their absolute errors (in kg and s respectively) cannot be compared.
Percentage Uncertainty in a Single Measurement
Even when you do not know the true value, you can still quote a percentage uncertainty. If a ruler has a resolution of 1 mm, you might quote the absolute uncertainty as ±0.5 mm. If you measure a length L = 24.5 cm = 245 mm, the percentage uncertainty is (0.5 / 245) × 100% ≈ 0.20%. A vernier calliper with resolution 0.1 mm would reduce this to (0.05 / 245) × 100% ≈ 0.02%, demonstrating that finer instruments produce smaller percentage uncertainties.
Propagation of Errors — Addition and Subtraction
When two measured quantities are added or subtracted to obtain a derived quantity, absolute errors add. If z = x + y or z = x − y, then Δz = Δx + Δy. This is the most pessimistic (worst-case) estimate: we assume errors could both be in the same direction. For example, if x = 12.0 ± 0.2 cm and y = 7.5 ± 0.3 cm, then z = x − y = 4.5 cm and Δz = 0.2 + 0.3 = 0.5 cm, giving z = 4.5 ± 0.5 cm.
Propagation of Errors — Multiplication and Division
When measured quantities are multiplied or divided, percentage uncertainties add. If z = x × y or z = x / y, then (Δz / z) × 100% = (Δx / x) × 100% + (Δy / y) × 100%. For example, speed v = d / t where d = 80.0 ± 0.5 m and t = 10.0 ± 0.2 s. Percentage uncertainty in d: (0.5/80.0) × 100% = 0.625%. Percentage uncertainty in t: (0.2/10.0) × 100% = 2.00%. Percentage uncertainty in v: 0.625% + 2.00% = 2.625%. So v = 8.00 m/s with a percentage uncertainty of about 2.6%, meaning Δv ≈ 0.21 m/s and v = 8.00 ± 0.21 m/s.
Propagation of Errors — Powers
When a measured quantity is raised to a power n, the percentage uncertainty is multiplied by |n|. If z = x^n, then (Δz / z) × 100% = |n| × (Δx / x) × 100%. This is critically important: squaring a measurement doubles its percentage uncertainty, and cubing it triples it. For example, the volume of a cube with side L = 5.0 ± 0.1 cm: percentage uncertainty in L = (0.1/5.0) × 100% = 2.0%. Percentage uncertainty in V = L³ is 3 × 2.0% = 6.0%. So V = 125 cm³ ± 6.0%, giving ΔV = 7.5 cm³ and V = 125 ± 8 cm³ (rounded to 1 significant figure in the uncertainty).
The Key Equation
- x_{\text{measured}} — Measured value (same as quantity): The value obtained from the measurement instrument or experiment.
- x_{\text{true}} — True (accepted) value (same as quantity): The accepted, theoretical, or reference value against which the measurement is compared.
- |x_{\text{measured}} - x_{\text{true}}| — Absolute error (same as quantity): The magnitude of the difference between measured and true values; always non-negative.
- \%E — Percentage error (%): A dimensionless measure of accuracy expressed as a percentage, allowing comparison across different quantities.
Worked Example
Problem. A student measures the period of a pendulum by timing 20 complete oscillations, obtaining a total time of 29.6 s. The stopwatch has a resolution of 0.1 s and a reaction-time uncertainty of ±0.2 s for the whole 20-oscillation count. The accepted period from theory (using L = 0.546 m, g = 9.81 m/s²) is T = 2π√(L/g) = 1.481 s. (a) Find the measured period and its absolute uncertainty. (b) Find the percentage uncertainty in the measured period. (c) Find the percentage error compared to the accepted value.
Answer. Measured period: 1.48 ± 0.01 s; Percentage uncertainty: 0.68%; Percentage error: 0.07%.
Origins · 1 minCarl Friedrich Gauss and how this was discovered
Gauss made foundational contributions to error analysis through his development of the method of least squares (published 1809) and his formalization of the normal distribution (Gaussian distribution), which describes how random measurement errors are typically distributed around a true value. His work gave mathematicians and scientists the theoretical framework for understanding and minimizing measurement uncertainty, forming the basis of modern statistical error analysis. Gauss reportedly discovered the method of least squares at age 18, years before he published it. He used it to successfully predict the reappearance of the dwarf planet Ceres after astronomers had lost track of it — an achievement that made him famous across Europe overnight.
Related Concepts
- Significant Figures — Significant figures are the primary tool for expressing the precision of a measurement, and understanding them is a prerequisite for correctly reporting measurements with their uncertainties.
- Rounding and Estimation — Rounding rules determine how calculated results and uncertainties should be expressed to an appropriate number of significant figures, which is essential when reporting percentage errors.
- Mean and Standard Deviation — When multiple measurements are taken, the standard deviation quantifies the spread of results and serves as a statistical measure of uncertainty, complementing the percentage error approach.
- Ratio and Proportion — Percentage error is itself a ratio — the error divided by the true value — so a solid understanding of ratios and proportional reasoning is necessary to interpret and calculate percentage errors correctly.