Overview
The median absolute deviation, MAD, measures how spread out data are in a way that one extreme value can't distort. It's the median distance of each value from the median.
Enter your values in any order.
Worked examples
| Inputs | Result |
|---|---|
| Your numbers12, 16, 14, 21, 19 | Median absolute deviation: 3Median: 16Q1/Q3: 14 / 19 |
Show 2 more examples
| Your numbers12, 14, 16, 19, 100 | Median absolute deviation: 3Median: 16Q1/Q3: 14 / 19 |
| Your numbers2, 4, 4, 5, 9 | Median absolute deviation: 1Median: 4Q1/Q3: 4 / 5 |
How it works
Find the median, then each value's absolute distance from it, then the median of those distances. Because it uses medians throughout, an outlier barely changes it, unlike the standard deviation. Multiplying MAD by 1.4826 gives an estimate of the standard deviation for normal data.
The formula
Practical tips
- For 12, 14, 16, 19 and 21, the median is 16, the distances are 4, 2, 0, 3 and 5, and the MAD is 3.
- Changing 21 to 100 leaves the MAD at 3, while the standard deviation jumps from 3.65 to about 38.
- Don't confuse it with the mean absolute deviation, which averages the distances from the mean.
Frequently asked questions
Why use MAD instead of the standard deviation?
It resists outliers, so it's better for messy data or for spotting the outliers themselves.
How do I flag outliers with MAD?
A common rule flags values more than about 3 MADs (scaled by 1.4826) from the median.
Is MAD the same as mean absolute deviation?
No, though both are abbreviated MAD. Check which one a source means.