One number to balance them all.

What "average" actually means

When someone says average, they mean a typical, representative value — a number that sits somewhere in the thick of the data and sums it up. If your friend asks "how tall are kids in your year?" you don't read out forty numbers. You give one: the average. That's its whole job — to replace a crowd with a single fair stand-in.

Why bother shrinking a crowd down to one number at all? Because our brains are hopeless at holding forty figures at once but brilliant with one. An average lets you compare things instantly: is your class taller than the one next door, did the team play better this season than last, are phones getting cheaper? Each of those is really a question about two whole crowds — and the quickest way to compare crowds is to let one number speak for each.

But "typical" can be measured in three different ways, and each tells a slightly different story:

Mean — add everything up, then share it out equally.
Median — line the numbers up in order and grab the one in the middle.
Mode — the value that shows up the most often.

There's also a fourth number we'll meet, the range. It isn't an average — it doesn't tell you the middle. Instead it measures the spread: how stretched-out the data is from smallest to largest. An average and a range together are a tiny, powerful summary of almost any pile of numbers.

The mean: add it all, share it out

The mean is the one most people picture when they hear "average." The recipe couldn't be simpler:

Say four friends have £2, £4, £6 and £8. Add them: that's £20. There are 4 friends, so 20 ÷ 4 = £5. The mean is like everyone pooling their money in the middle of the table and splitting it back equally — the fair-share number. Some people lose a little, some gain a little, but the total never changes.

Let's do one more with messier numbers, so the recipe really sticks. Suppose five friends scored 6, 7, 7, 9 and 11 on a quiz. Add them: 6 + 7 + 7 + 9 + 11 = 40. Count them: 5 scores. Divide: 40 ÷ 5 = 8. So the mean score is 8 — even though nobody actually scored an 8. Tuck that surprise away; we'll come back to it, because it trips up a lot of people.

That "fair share" idea hides a beautiful second picture. The mean is also the balance point of the data — the exact spot where, if every number were a weight sitting on a ruler, the ruler would tip level and hold still. That's not a coincidence; it's the same idea wearing a different costume. Time to play with it.

The mean is the balance point

Drag the gold dots along the number line, then slide the orange pivot until the plank tips level. The one spot where it balances is the mean — every time.

Why does "add it all and share" turn out to be the same as "find where it balances"? Think about a single dot. The further it sits from the pivot, the harder it pulls that side down — exactly like sitting right at the end of a see-saw lets a small kid lift a much bigger one near the middle. The mean is the one spot where all those left-pulls and right-pulls cancel out perfectly. Nudge the pivot a touch and one side instantly wins. Two recipes, one number.

Now notice what happens when you drag one dot way out to the right. The plank lurches, and to balance it again the pivot has to shuffle toward that runaway dot. That's the mean reacting: because it uses every value in its sum, even one extreme number can drag it off the middle. Hold onto that — it's the mean's biggest strength and its biggest weakness at the same time.

The median: just stand in the middle

The median doesn't care about adding or sharing. It only cares about order. Line your numbers up smallest to largest, walk to the exact middle, and whoever's standing there is the median. Simple — but stubborn in the best way: a giant value out at the edge can't budge it, because the middle of the line is still the middle.

One catch. If you have an odd count there's a single dot in the dead centre — easy. If you have an even count, two dots share the middle, so you take the mean of those two as the median. Press the buttons below to watch it sort itself out, and toggle the count between odd and even to see that rule in action.

Here's the odd case in slow motion. Take the scores 7, 3, 9, 1, 5. Jumbled up they tell you nothing, so line them up first: 1, 3, 5, 7, 9. Walk to the centre — there are two values on each side of the 5 — so the median is 5. Now the even case: add an 11 to get 1, 3, 5, 7, 9, 11. There's no single middle anymore; the 5 and the 7 are tied for the centre, so you average them: (5 + 7) ÷ 2 = 6. Notice the median came out as 6 — a number that isn't even in the list. That's completely fine.

The mode: whatever happens most

The mode is the value that turns up most often — the crowd favourite. It's the only one of the three with a superpower: it works on things you can't add up. You can't take the mean of "strawberry, chocolate, vanilla," but you can absolutely find the most popular flavour. So the mode is the go-to average for categories — colours, names, sizes, choices.

Numbers can have modes too, of course. In the shoe-size list 6, 7, 7, 7, 8, 9 the size 7 turns up three times — more than anything else — so the mode is 7. And here's a neat detail: the mode is the only average guaranteed to be a value you could actually point to in your data, because by definition it has to be one of your numbers.

Mode is also honest about ties. A dataset can have no mode at all (everything appears the same number of times) or several modes (two flavours tie for first — that's "bimodal"). Cast some votes below and watch the tallest bar light up as the mode.

Range: how spread out is it?

An average tells you the middle, but it stays quiet about something important: are the numbers all bunched up together, or flung far apart? That's what the range answers, and its recipe is a one-liner:

Imagine two classes that both have a mean test score of 70. In the first class everyone scored between 66 and 74 — a range of 8, nicely consistent. In the second, scores ran from 30 to 100 — a range of 70, all over the place. Same average, wildly different story. That's why the range matters: it catches the spread the average hides. (Look back at the balance demo — its "Range" readout has been quietly tracking your largest dot minus your smallest the whole time.)

One warning, though. Because the range only ever looks at the two end values, a single outlier can blow it wide open. If forty-nine people are aged 10 to 12 and one grandparent is 70, the range leaps to 60 — even though almost everyone is bunched tightly together. So treat the range as a quick, rough measure of spread: handy, but just as easily fooled by one extreme value as the mean is.

All four, on one tiny dataset

Let's run every tool over the same little pile of numbers, so the whole toolkit clicks at once. Imagine the number of pets owned by seven friends:

Mean — add them (0+1+1+2+2+2+8 = 16) and divide by 7 → about 2.3 pets each.
Median — they're already in order, so the middle one (the 4th of 7) is 2.
Mode — the value that shows up most is 2 (three friends have two pets).
Range — largest minus smallest, 8 − 0 = 8.

See the story those four numbers tell together? The mean (2.3) got quietly nudged upward by that one friend with eight pets, while the median and the mode both sit calmly at 2 — the genuinely typical number. Meanwhile the range of 8 is waving a little flag: "careful, this data is more stretched-out than any single average lets on." Four quick calculations, and you already understand the whole group.

The outlier: mean lurches, median holds

An outlier is a value that sits far away from the rest — a billionaire wandering into a room of regular folks, a single 100 in a sea of 50s. Outliers are the moment the three averages split up and stop agreeing.

Picture a tiny company where five workers earn around £30,000–£40,000. Now crank up the boss's pay with the slider. Watch the mean sprint to the right, chasing the big number, while the median sits calmly in the middle, barely moving. This is exactly why "the average salary" can be misleading — it usually means the mean, and one huge salary can quietly inflate it.

Outliers aren't always villains, and they aren't always real. Sometimes an outlier is the most interesting thing in the whole dataset — the record-breaking athlete, the once-in-a-century storm. Other times it's just a slip: someone typed 500 when they meant 50. Either way, spotting one is your cue to slow down and ask two questions — should this number really be here? and which average can I trust while it is?

Rule of thumb: when there's a wild outlier and you want a number that reflects the typical person, reach for the median. When every value genuinely matters and there are no monsters, the mean is your friend.

Which average should you use?

There's no single "best" average — the right one depends on the data and the question. Pick a scenario and see which one earns the job, and why.

Boil it down to three quick questions: Are the values words or categories? Use the mode. Is there a wild outlier? Lean on the median. Is the data fairly even, and does every value count? The mean is perfect.

And here's a grown-up secret: you don't always have to pick just one. Scientists and journalists often report a couple of numbers side by side — say, the median and the range — precisely because no single value can capture everything. The average tells you the middle; the range tells you the spread. Together they're honest in a way either one alone simply can't be.

Two traps people fall into

Averages are everywhere

Once you spot them, you can't stop. These three little numbers run quietly under huge parts of daily life:

So the next time you read "the average anything," you've got a kind of x-ray vision: which average is this really, could an outlier be skewing it, and would a different one tell a fairer story? That single question turns you from someone numbers happen to into someone who reads them on purpose.