The 'average salary' that nobody actually earns.
Nine employees earn $30k; the CEO earns $3M. What's this company's 'average salary'?
Three spokespeople summarizing the same data differently.
Experiment
Hands-on experiment
Predict first — nine employees earn $30k; the CEO earns $3M. The company's average salary is…?
Move the CEO's salary (the 9 employees stay at $30k)
Drag the slider. Only the CEO's dot and the mean's fulcrum get pulled right — the median line stays nailed in place.
Mean salary
$3.27M
Median salary
$3,000k
Read more — why it exists · insights · common mistakes · formulasExpand ▾
Why
Why does this exist?
'The average salary', 'the average home price', 'average life expectancy' — the world summarizes groups into single numbers. And that number often feels wrong compared to your reality.
Because there's more than one way to summarize. Add-and-divide (mean), line-up-and-pick-the-middle (median), most-common-value (mode) — the three can produce wildly different numbers from the same data.
Especially in top-skewed distributions like income and wealth, the mean gets dragged by a handful of extremes. The habit of checking which spokesperson is talking — that's the first step to not being fooled by news and statistics.
Insight
Insights from the video
“The mean isn't democratic — one big number dominates the vote.”
In a mean, every value participates in proportion to its size. The CEO's $3M speaks louder than a hundred employees combined. The median is one-person-one-vote.
“The question chooses the spokesperson.”
'Total payroll budget?' — the mean. 'A typical employee's life?' — the median. 'The most common case?' — the mode. Choosing a representative isn't taste; it's a function of the question.
Misconception
Common misconceptions
Half the people earn above average and half below.
That's the median's property. The mean gets dragged by extremes, so in skewed distributions like salaries, most people can be 'below average'.
One representative number is enough — the mean IS the group.
Every summary throws information away. Depending on what got discarded (skew, outliers, spread), some spokesperson will sound like a liar. Picking the representative that fits the question is where statistical literacy begins.
Formula
Writing it as math
Meet the three spokespeople, in formulas.
The mean
Add everything, divide by the count. Every value participates by its size, so it's sensitive to extremes — the same principle as a center of mass.
The median
Line up and take the middle. It sees only rank, not size, so the CEO earning $3M or $300M moves it not at all.
The mode
The majority's representative. Like 'the most common shoe size', it tells you the case you'll actually meet most often.
In Real Life
Where you meet it in real life
Salary and housing statistics
Why governments publish both mean and median — the gap between them is itself information: the size of the skew.
A shoe store's inventory
Order for the average size 9.4 and you go broke. Inventory follows the mode — the best-selling size.
The illusion in game revenue
Average spend per user is dragged up by a few 'whales'. To know the typical player, look at the median (usually $0).
The life-expectancy trap
The historical 'average lifespan of 40' was dragged down by infant mortality. Adults back then often lived to 60–70.
Try Yourself
Test yourself
Q1For the data [1, 2, 2, 3, 100], what are the mean and the median?Show answer ▾
Mean 21.6, median 2. One 100 dragged the mean up over tenfold while the median didn't budge — the difference in outlier resistance.
Q2The class average is 70. Can fewer than half the students score 70 or above?Show answer ▾
Yes — a few very high scores can pull the mean up. The 'half above, half below' property belongs to the median, not the mean.
Q3A headline reads 'Half the nation earns below the average income'. Is that strange?Show answer ▾
Not at all — with a top-skewed income distribution it's natural. In fact 'half earn below the MEDIAN' is always true, hence not news. Distinguish mean from median and the headline's identity shows itself.
Try answering yourself before revealing it — getting it wrong is where learning starts.
Connection
Concepts connect
Previous concept
Expected Value
EV is the probability world's mean — know the mean's properties and EV's follow.
← Expected Value labLeads to next
Variance & Standard Deviation
If a representative summarizes the 'center', variance summarizes the 'scatter' around it. The second piece of information the mean hides awaits.
Go to the Variance & Standard Deviation lab →Related
Labs worth exploring together
Related lab
Take-home Pay
While we're on salaries — see the structure between salary and bank account.
Go to the Take-home Pay lab →