What Uncertainty Means and Why You Calculate It

Uncertainty is the range of values that could reasonably contain the true value of something you measured or calculated. When you measure temperature with a thermometer, weigh an object on a scale, or time an event with a stopwatch, your measurement is never perfectly exact — it always has some margin of error. Computing uncertainty tells you how much that margin might be, so you know whether your result is trustworthy enough for your purpose.

If you measure a board as 10 centimeters long but your ruler's smallest marking is 1 centimeter, you might report the length as 10 ± 0.5 cm. That ± symbol means the true length could be anywhere from 9.5 to 10.5 cm. Without calculating and stating that range, someone reading your result would not know whether you measured to the nearest millimeter or the nearest meter.

Uncertainty matters in science, engineering, cooking, medicine, and any field where precision affects the outcome. A pharmaceutical company needs to know the uncertainty in drug dosages. A carpenter needs to know the uncertainty in cut lengths. A weather station needs to know the uncertainty in temperature readings. Computing it forces you to think about what could go wrong with your measurement and how much that matters.

Key Takeaways

  • Absolute uncertainty is the actual range (like ± 0.5 cm), while relative uncertainty is that range as a percentage of your measurement (like 5%).
  • For a single measurement, uncertainty usually equals half the smallest division on your instrument, or the manufacturer's stated precision.
  • When you combine measurements (add, subtract, multiply, or divide), you combine their uncertainties using specific rules that depend on the operation.
  • Repeated measurements let you calculate uncertainty using standard deviation, which shows how much your results scattered around the average.
  • Always report your final result with its uncertainty so readers know the range of reasonable values.

Uncertainty from a Single Measurement and Your Instrument

The simplest case is when you take one measurement with one tool. Your uncertainty comes from the precision of that tool — how finely it can divide what you are measuring.

For an analog instrument (one with a needle or a scale you read by eye), absolute uncertainty is usually half the smallest division you can see. If a thermometer has marks every 1 degree, you can reasonably estimate to the nearest 0.5 degrees, so your uncertainty is ± 0.5 degrees. If a ruler has millimeter marks, your uncertainty is ± 0.5 mm. You are splitting the smallest gap in half because you can estimate between the marks, but not more precisely than that.

For a digital instrument (one with a number on a display), check the manufacturer's manual or the back of the device. Most digital scales state something like "± 0.1 g" or "± 1°C". That number is your uncertainty. If no manual is available, a safe assumption is that the uncertainty equals the smallest digit shown — a scale that displays to 0.1 g has an uncertainty of about ± 0.1 g.

Once you have your absolute uncertainty, you can calculate relative uncertainty (also called percent uncertainty) by dividing the absolute uncertainty by your measurement and multiplying by 100. If you measured 50 grams with an uncertainty of ± 0.5 grams, your relative uncertainty is (0.5 / 50) × 100 = 1%. Relative uncertainty tells you whether your measurement is good or poor: 1% is quite good, 10% is rough, 50% is barely useful.

Combining Uncertainties When You Add or Subtract Measurements

Often you do not measure one thing directly — you measure two or more things and combine them. If you measure the length of two boards and add them to find the total length, you have two separate uncertainties to combine.

When you add or subtract measurements, add the absolute uncertainties together. Do not add the relative uncertainties. If board A is 50 ± 0.5 cm and board B is 30 ± 0.3 cm, the total length is (50 + 30) ± (0.5 + 0.3) = 80 ± 0.8 cm. The combined uncertainty is 0.8 cm, not 0.5 + 0.3 = 0.8 relative to the total (which would be wrong). This rule holds whether you are adding or subtracting — the uncertainties always add.

The reason is that uncertainties can work for or against you. Board A might be 0.5 cm longer than you think, or 0.5 cm shorter. Board B might be 0.3 cm longer or shorter. In the worst case, both errors go the same direction, so you add them. In the best case, they cancel out, but you cannot count on that, so you plan for the worst.

Combining Uncertainties When You Multiply or Divide Measurements

When you multiply or divide measurements, the rule is different: add the relative uncertainties, not the absolute ones. If you measure the length and width of a rectangle and multiply them to find area, you combine relative uncertainties.

Suppose length is 10 ± 0.5 cm (relative uncertainty = 0.5/10 = 5%) and width is 6 ± 0.3 cm (relative uncertainty = 0.3/6 = 5%). The area is 10 × 6 = 60 cm². The combined relative uncertainty is 5% + 5% = 10%. So the area is 60 ± 10% = 60 ± 6 cm², or 54 to 66 cm². You add the relative uncertainties and then convert back to absolute uncertainty if you need it.

This rule applies to multiplication, division, and any combination of the two. If you divide one measurement by another, still add the relative uncertainties. The logic is that when you multiply or divide, errors compound differently than when you add or subtract — a small relative error in one measurement becomes a larger relative error in the result.

Computing Uncertainty from Repeated Measurements

If you measure the same thing several times and get slightly different results each time, you can calculate uncertainty using standard deviation. This method is more powerful than the instrument-precision method because it captures real variation in your process, not just the limits of your tool.

First, measure the same thing at least three times (more is better). Write down each result. Then calculate the average (mean) by adding all results and dividing by how many you took. Next, subtract the average from each individual result to find the difference for each measurement. Square each difference, add all the squared differences together, and divide by the number of measurements minus one. Finally, take the square root of that number. That is your standard deviation.

For example, if you timed an event five times and got 10.2, 10.5, 10.1, 10.3, and 10.4 seconds: the average is (10.2 + 10.5 + 10.1 + 10.3 + 10.4) / 5 = 10.3 seconds. The differences from average are −0.1, +0.2, −0.2, 0, and +0.1. Squared, they are 0.01, 0.04, 0.04, 0, and 0.01. The sum is 0.10. Divide by (5 − 1) = 4 to get 0.025. The square root is 0.158, which you round to ± 0.16 seconds. Your result is 10.3 ± 0.16 seconds.

Standard deviation captures scatter in your measurements. If your five times had been 10.0, 10.1, 10.3, 10.5, and 10.6 seconds, the standard deviation would be larger because the results scattered more. This method is especially useful when you suspect your instrument is precise but your technique (your hand, your reaction time, your setup) introduces variation.

Reporting Your Result With Uncertainty

Once you have calculated uncertainty, write your final result in the form: measurement ± uncertainty, with units. Examples: 25.3 ± 0.2 grams, 98.6 ± 0.5 °C, 5.67 ± 0.08 meters per second.

Round your uncertainty to one or two significant figures (usually one). If your uncertainty is 0.0347, round it to 0.03 or 0.035, not 0.0347. Then round your measurement to the same decimal place as your uncertainty. If your uncertainty is ± 0.2 cm, your measurement should be stated to the nearest 0.1 cm, not to 0.01 cm. Reporting 25.34 ± 0.2 looks silly because the uncertainty is larger than the last digit you reported.

Include units with both the measurement and the uncertainty, or state the units once after both numbers. "50 ± 2 grams" and "50 grams ± 2 grams" are both clear. "50 ± 2" is not, because a reader will not know whether the uncertainty is 2 grams or 2 percent.

Common Mistakes When Computing Uncertainty

One frequent error is forgetting to add uncertainties when combining measurements. If you measure two things and add them, you must add the uncertainties too. Ignoring the second measurement's uncertainty makes your final result look more precise than it really is.

Another mistake is using the wrong rule for the type of combination. Adding relative uncertainties when you should add absolute ones (or vice versa) gives a wrong answer. Write down whether you are adding/subtracting or multiplying/dividing, then explore the correct rule.

A third error is reporting too many significant figures in your uncertainty or in your measurement. If your uncertainty is ± 0.2, reporting the measurement as 25.347 is misleading — the last two digits are meaningless noise. Round to match your uncertainty.

Finally, some people forget to report uncertainty at all. A measurement without uncertainty is incomplete. Always state the range, whether as absolute uncertainty (± 0.5) or relative uncertainty (± 5%), so readers know how much to trust your number.

Frequently Asked Questions

What is the difference between absolute and relative uncertainty?

Absolute uncertainty is the actual range, stated in the same units as your measurement (like ± 0.5 cm). Relative uncertainty is that range as a percentage of your measurement (like ± 5%). Relative uncertainty is useful for comparing how good different measurements are — a 1% relative uncertainty is better than a 10% relative uncertainty, regardless of what you measured.

Do I always have to measure something multiple times to calculate uncertainty?

No. If you measure once with a tool, your uncertainty comes from the tool's precision (half the smallest division for analog, or the stated precision for digital). You only need to measure multiple times if you want to capture variation in your technique or process. Both methods are valid; use whichever fits your situation.

What if I multiply three measurements together instead of two?

Add the relative uncertainties of all three. If you multiply A, B, and C, add their relative uncertainties together to get the relative uncertainty of the result. The rule scales to any number of measurements.

Can uncertainty be negative?

No. Uncertainty is always stated as a positive number (± means plus or minus). It represents a range, not a direction. An uncertainty of ± 0.5 means the true value could be 0.5 higher or 0.5 lower, not that it is definitely lower.

What should I do if my calculated uncertainty is very large?

A large uncertainty means your measurement is not very precise. You can try to reduce it by using a more precise tool, measuring more carefully, or taking more repeated measurements to average out random errors. If you cannot reduce it, report it honestly — a large uncertainty is better than a false claim of precision.