Measure or calculation

Standard deviation

Where you meet this term: IB Diploma

A measure of how widely the values in a data set are spread around their mean: the larger the standard deviation, the more the values vary.

In plain words

A measure of how far the values typically lie from their mean: the bigger it is, the more spread out the data are.

In more depth

Standard deviation = √(Σ(x − mean)² ÷ n), in the same units as the data. It uses every value, so outliers enlarge it. In business it measures variability or risk, such as how much weekly sales or delivery times vary; software estimating it from a sample divides by n − 1, which gives a slightly larger figure.

Formula

Standard deviation = √(Σ(x − mean)² ÷ n), where x is each value and n is the number of values

See it worked in the toolkit guide to descriptive statistics

In context

It is calculated from how far each value lies from the mean, so it uses all the data, and extreme values increase it. Two products can have the same mean weekly sales but very different standard deviations, and the one with the larger figure is harder to plan production and stock for. It is expressed in the same units as the data.

Example

Two cafés both average 200 customers a day; one varies between 190 and 210 and the other between 100 and 300, so the second has a much larger standard deviation and needs more flexible staffing.

Common misconceptions

  • Misconception

    A large standard deviation means the values themselves are large.

    In fact

    It measures spread around the mean, not the size of the mean: two data sets with the same mean can have very different standard deviations.

Frameworks, tools and diagrams

Suggested next

  • Median Linked through Mean and Quartiles and Descriptive statistics
  • Mode Linked through Mean and Descriptive statistics
  • Quality standard A related name