The mean is the most-used measure of center, but there are times when it is recommended to use the median for data display and analysis. When might it be appropriate to use the median instead of the mean?
1 Answer
Jan 2, 2015
When there are a few extreme values in your data set.
Example:
You have a dataset of 1000 cases with values not too far apart. Their mean is 100, as is their median. Now you replace just one case with a case that has value 100000 (just to be extreme). The mean will rise dramatically (to almost 200), while the median will be unaffected.
Calculation:
1000 cases, mean=100, sum of values=100000
Lose one 100, add 100000, sum of values=199900, mean=199.9
Median (=case 500+501)/2 stays the same.
