Standard deviation uses the original units because it is the square root of variance. Squaring each difference from the mean gives variance squared units, such as cm². Taking the square root brings the result back to centimeters, so you can compare the spread directly with the measurements.
If a mean length is 4 cm, a standard deviation of about 1.633 cm describes spread on that same length scale. A variance of about 2.667 cm² describes the same data through squared differences. The numbers differ because the two measures express spread in different ways.
This lesson follows the units through one small, invented dataset. It also shows what happens when you move every value, convert units, or reverse signs. For related measures of center and spread, start with the statistics tools hub.
Why Does Variance Have Squared Units?
Variance has squared units because its calculation squares the differences between observations and their mean. Subtracting two lengths leaves a length: 2 cm minus 4 cm is −2 cm. Multiplying that difference by itself gives 4 cm².
A difference from the mean is called a deviation. A negative deviation places a value below the mean, while a positive deviation places it above. Both signs matter for location, but they would cancel if we simply added the deviations.
For lengths of 2, 4, and 6 cm, the mean is 4 cm. The deviations are −2, 0, and 2 cm, whose sum is zero. A zero sum doesn’t mean there is no spread; it means the signed differences balance around the mean.
Squaring gives 4, 0, and 4 cm², which add to 8 cm². All observations now contribute nonnegative amounts to the total. A difference twice as large contributes four times as much after squaring, so larger departures receive greater weight.
Dividing by a count doesn’t remove the squared units. A count of observations is a number, not a length. Thus, 8 cm² divided by three observations still has units of cm², just as dividing a length by a plain number still leaves a length.
The unit cm² doesn’t mean variance measures a physical area. Here it records the multiplication of a length difference by itself. A variance calculated from waiting times would use minutes², even though nobody waited for a “square minute.”
How Does The Square Root Restore The Units?
The square root reverses the squaring of the units. For the complete population 2, 4, and 6 cm, variance is 8/3 cm². Its square root is √(8/3) cm, or about 1.633 cm, so the standard deviation returns to the unit used for each length.
Follow the calculation below from left to right. The first subtraction keeps centimeters, the squaring changes them to cm², and the final square root returns them to centimeters. The number and its unit both pass through these steps.
| Length | Deviation from 4 cm | Squared deviation |
|---|---|---|
| 2 cm | −2 cm | 4 cm² |
| 4 cm | 0 cm | 0 cm² |
| 6 cm | 2 cm | 4 cm² |
| Total | 0 cm | 8 cm² |
The exact variance is 8/3 cm², not 2.667 cm². The decimal is a rounded display, and the exact standard deviation is √(8/3) cm. Keep the fraction until the last step to avoid carrying a rounding error into the square root.
Why not take each square root before averaging? That would undo each squared deviation separately and produce absolute distances from the mean. Standard deviation instead takes one square root after the squared differences have been combined and divided.
OpenStax’s explanation of spread defines standard deviation as the square root of variance and distinguishes population and sample formulas. The examples here use invented lengths to make those steps and their units easy to check.
Do Sample And Population Formulas Change The Units?
Sample and population formulas change the divisor, but they don’t change the units. Both formulas combine squared deviations and then take a square root. The population formula divides by the number of values; the usual sample formula divides by one less than that number.
If 2, 4, and 6 cm represent a sample, their mean and squared deviations stay the same. The sample variance is 8/(3 − 1) = 4 cm². The sample standard deviation is √4 = 2 cm.
Compare that with the population standard deviation of about 1.633 cm. The larger sample result comes from the smaller divisor, not from a change of units. Choose the formula based on what the dataset represents, rather than which answer looks more convenient.
The symbols also help identify the choice. Population standard deviation is often written σ, and sample standard deviation is written s. Report which one you used when checking results with the standard deviation calculator.
A constant dataset such as 4, 4, and 4 cm has zero spread. Every deviation is zero, so both formulas give zero standard deviation. A population containing one value also has zero spread, but the usual sample formula is undefined for one value because its divisor is zero.
What Happens When You Convert Centimeters To Millimeters?
Converting centimeters to millimeters multiplies standard deviation by 10 and variance by 100. Each length and each difference from the mean becomes ten times its previous number. Squaring those differences introduces a factor of 100; taking the square root returns the factor to 10.
The same three lengths become 20, 40, and 60 mm, with a mean of 40 mm. Their deviations are −20, 0, and 20 mm. The squared deviations add to 800 mm², giving population variance 800/3 mm².
The population standard deviation is √(800/3) mm, or about 16.330 mm. That is the same physical spread as about 1.633 cm. The larger number reflects a smaller measuring unit, not a change in the lengths themselves.
Convert the squared units carefully too. Since 1 cm equals 10 mm, 1 cm² equals 100 mm². Multiplying the variance number by only 10 would mix a length conversion with a squared-length quantity and produce the wrong result.
Don’t compare spread numbers until their units match. A standard deviation of 16 mm is smaller than 2 cm, even though 16 is larger than 2. Convert either result first, then compare values expressed in the same unit.
Converting back follows the same rule in reverse. Divide the millimeter standard deviation by 10 to obtain centimeters, and divide the millimeter variance by 100 to obtain square centimeters.
How Do Shifts And Scale Changes Affect Spread?
Adding the same number to every value leaves standard deviation unchanged because the mean moves by that number too. Add 10 cm to our lengths and the list becomes 12, 14, and 16 cm. The mean becomes 14 cm, leaving the deviations at −2, 0, and 2 cm.
Variance therefore remains 8/3 cm², and population standard deviation remains about 1.633 cm. The whole group moved to a new location without becoming wider. This separates a measure of center from a measure of spread.
Multiplication changes spread instead of merely moving it. Multiply each numerical value by 10, and every deviation becomes ten times as large. Variance becomes 100 times as large, while standard deviation becomes ten times as large.
A negative multiplier reverses direction but cannot create negative spread. Treat the numbers as signed positions in centimeters, rather than physical lengths. Multiplying 2, 4, and 6 cm by −3 gives −6, −12, and −18 cm, with mean −12 cm.
The deviations become 6, 0, and −6 cm. Population variance becomes 24 cm², and population standard deviation becomes √24 cm, about 4.899 cm. The sign reversal changes positions, not the requirement that spread be nonnegative.
The general rule uses the absolute value of the multiplier: multiplying all values by a scales standard deviation by |a|. Variance scales by a². For a = −3, standard deviation triples and variance becomes nine times as large.
These rules assume the same change applies to every observation. Changing just one value can alter both the mean and the spread in less simple ways. Multiplying every value by zero makes them all equal, producing zero standard deviation for any valid dataset.
What Does Standard Deviation Tell You?
Standard deviation describes spread around the mean in a unit you can read alongside the data. It is not the ordinary average of absolute distances from the mean. For our population example, those absolute distances are 2, 0, and 2 cm, whose average is 4/3 cm.
That average absolute distance is about 1.333 cm, while standard deviation is about 1.633 cm. Both answers use centimeters, but they come from different operations. Matching units doesn’t make two measures interchangeable.
Standard deviation also doesn’t guarantee a fixed share of observations within one standard deviation of the mean. In our three-value population, only 4 cm lies within 4 ± 1.633 cm. That is one of three observations, not about 68%.
The familiar 68% guideline applies approximately to bell-shaped normal distributions. It isn’t a rule for every list or distribution shape. Check the data’s shape before using standard deviation to describe how many observations lie within a particular interval.
How Should You Report Standard Deviation Units?
Report the standard deviation with the original measurement unit and identify the formula you used. For this example, write “population standard deviation ≈ 1.633 cm.” If the values are a sample, write “sample standard deviation = 2 cm” instead.
Keep variance clearly labeled as a separate quantity with squared units. Use the variance calculator when you want to inspect that step. A label such as “variance = 8/3 cm²” helps prevent a squared result from being mistaken for a length.
The same unit check works for seconds, dollars, kilograms, or other numerical measurements. Differences keep the original unit, squared differences create squared units, and the final square root restores the original unit. Read the unit beside the number, because both are part of the answer.