Why the Average Lies

A Billionaire Walks
Into a Bar

Twenty strangers at a bar have a perfectly ordinary average net worth. Then one very specific billionaire walks in. Nobody in the room got richer. The average did.

Try it below

Act one

Meet the room

Twenty people, twenty ordinary net worths. Watch what one very rich stranger does to the number everyone quotes, and the number that barely notices him.

Each dot is one person, placed by net worth. Nobody has moved yet.

Average (mean) $0
Median $0

Net worth figures are stylized round numbers for illustration, not exact published figures.

Act two

How big a room would fix this?

Maybe twenty people is just too small a room. Pick a bigger one and see how much space it takes before one billionaire stops dominating everyone else's average.

Average net worth in the room $6.19B

With 20 ordinary people and one specific billionaire, the average is already absurd.

It would take roughly 2.6 million ordinary people in the room before his presence stops more than doubling the average. The median, meanwhile, would barely have moved at all.

Why statisticians reach for the median

One number can hijack an average. It can't hijack a median.

The mean adds up every value and divides by how many there are, so one number large enough can drag the whole result with it. The median just asks which value sits in the middle once everyone is lined up in order. Move the richest person in the room to the moon and the median doesn't even notice, because it never looked at how large that value was, only at its position in the line.

This is exactly why government statistics agencies usually report median household income, not mean household income, when describing what a "typical" household earns. A country's income distribution has a long tail of very high earners, and the mean gets pulled toward that tail every time. The median stays anchored to where most people actually are.

Statisticians call the median a robust statistic for exactly this reason: it resists distortion from extreme values. The mean is not robust. It reports whatever is in the room, extreme guests included.

  • A neighborhood of modest homes can report a shockingly high "average home price" the moment a single mansion sells nearby, even though every other house is worth the same as last year.
  • A class's "average test score" can look great because of two students who scored perfectly, while most of the class is still struggling. The median score tells that story more honestly.