So many times I look at workforce dashboards and they show the mean age, mean length of service, maybe even mean salary. As Good Math, Bad Math and others have often pointed out, using the mean is a bad idea – lazy at best, horribly misguiding at worst. And here’s why, according to Dr Salary:
The median salary is often much closer than the arithmetic mean to what common intuition would give for the typical salary. Strangely, people, political parties, newspapers, even statisticians and (gasp) PayScale continue to calculate the arithmetic mean and present it as a "typical" salary answer, when median salary would be much closer to what people want to know.
Outliers ruin the mean – the mean is really only useful for data which is evenly spread…and very little HR data IS evenly spread. Here’s the example given by Darwin’s Finance:
- You live in a town of 1000 residents who are all earning roughly $80,000 per year.
- To keep it simple, the median and the mean are roughly $80,000 give or take.
- Your town is real nice and reminds Warren Buffet of his childhood. He decides to move in.
- He takes in $1Billion this year.
While the median income for the town remains at $80,000 (because the middle number is still $80,000 as are the other 999, there is only one outlier making a billion), the arithmetic mean income for the town is 1.08 Million Dollars!
Sure it’s an extreme example, but you get the point – one or two outliers are enough to ruin your measure. So, whatever you do, don’t ask the business to make decisions using the mean average – it’s just too important.