HN Explorer

Ask HN: Why is the metric Median less popular than Average?

1 points by butterNaN ↗ HN
In almost all real-world instances - From individuals making personal decisions, to inter-governmental groups - whenever one is to try inferring meaning from data, Average seems to be the 'natural' metric of choice. To me, a lot of them could be using Median for better decision-making, but that seems to be no one's first choice.

From a sociological point of view, what is that?

A few points on why I feel Average is the dominant statistic (and counter-points):

* Median requires one to keep an indexed, sorted list. Average can be computed without needing order. _- This seems to be only a positive in computational sense, not real-world-decision-making sense. Governments can wait extra 10 minutes to calculate Median income instead of Average income._

* Average is simple for the Human Brain to make sense of

6 comments

[ 1.9 ms ] story [ 19.5 ms ] thread
mac3n
Average can be derived from totals Sometimes that's all the information that's available.
dsq
In an economic sense, using average is a way of obfuscating inequality. Average salary, for example, includes every salaried employee, such as the multi-million dollar a year senior executive, thus giving the mistaken impression that 'regular' salary is higher than it actually is. Median avoids these illusions.
ggeorgovassilis
You can compute things based on the average, eg. samples*average = totals. There is no interesting statistical property I'm aware of that is computed with medians.

From a philosophical perspective, the median hides outliers which, however, contribute to the total value of a population.

simonblack
On the other hand, you might have 10 million people starving because they have no money for food, but if you include a billionaire in the group you will find that everybody has plenty of money for food.
sn9
You're overthinking it.

It's because most people are mathematically illiterate and most people don't know that there's even a difference between median and average or a reason to use one over the other.

Computationally, computing the average vs the median is virtually instant for 99+% of data sets on modern hardware. I can't imagine what a data set would have to look like for it to take an extra 10 minutes to figure out the median.