Theorem
Standard deviation is at least the mean deviation
Statistics • Measures of Dispersion
Statement
For any observations, with and , we have . Equality holds exactly when every is the same number.
How to use it
- 1Call each distance from the mean . Then MD is the mean of the , and is the mean of the .
- 2For any numbers, the mean of the squares minus the square of the mean is a variance, and a variance is never negative. So , which gives .
- 3The two are equal when all the distances are the same, for example the data
: both are . So the safe statement is , not .
Worked example
For the data , check that .
Step-by-step
- 1Mean ; distances
Why: Find each distance from the mean.
- 2
- 3, so
Why: Average the squared distances, then take the root.
- 4
Answer
Watch out
Values that are not all equal can still give (for example ), so a strict is not always true.
Builds on
Also known as
SD greater than or equal to MD, mean deviation less than standard deviation, relation between SD and mean deviation