📊 Data
🌊Range & Spread
The average can fool you — to know how restless a dataset is, look at how far it scatters.
Two bus routes both claim "5 minutes late on average". Sounds like a tie? In truth one runs four or five minutes late every day, while the other is usually on time but now and then slips by half an hour at a stretch. The summaries from Mean, Median & Mode can't see this difference at all — we need one more quantity to say how far apart the data spreads.
One average, two personalities
- Group A: — mean 5, every value huddled around 5;
- Group B: — mean also 5, yet swinging wildly, propped up by a single 20.
Same average, totally different temperament. The quantity describing that temperament is called spread, and the simplest ruler for it is the range.
Range: maximum minus minimum
Range = maximum − minimum.
Group A: ; Group B: . Group B's range is ten times Group A's, and the restlessness shows at once. The range captures the two farthest endpoints in one sweep — quick, but it uses only two numbers, so whether the middle data huddles or scatters, it has no idea.
The range is only half the story
The range is decided by the two far ends alone: it can say how bad the worst gap gets, but nothing about where the main crowd of data lives. Read it alongside the mean or median, and no single extreme value can lead you astray.
A finer ruler: quartiles
Line the data up from smallest to largest and cut it into four equal parts. Take : the middle value is the median; the middle of the lower half, , is called Q1, and the middle of the upper half, , is Q3. The middle half of the data lives between and , and that distance
is the interquartile range. It only watches the middle 50% of the data, so no matter how the endpoints thrash about, it barely blinks.
Outliers: champions at stretching the range
Slip a into : the range leaps from 3 to , yet the rest of the data still huddles between 2 and 4. Such a towering stranger is an outlier — the moment it arrives, the range stops telling the truth. The IQR is more trustworthy here: 98 sits outside the middle 50%, so the IQR hardly moves.
This is also the referee for the "bus vs train" debate: when the average delay is the same, compare ranges or IQRs and you see instantly who is steady and who is erratic — and which monthly pass to buy.
Spread Lab
Count
7
Mean
6.14
Median
6
Mode
8
Edit the numbers freely and watch the bars scatter and squeeze while the mean line chases along. Two experiments: drag one number very large — the mean gets dragged away while the median stands its ground, the raw power of an outlier; then squeeze all the numbers together — the summaries barely change, but "scattered" versus "tight" is plain to the eye. Compute the range yourself: maximum minus minimum, right there in the numbers you just edited.
Check yourself
Quick quiz
1. What is the range of 3, 7, 2, 9, 4?
2. Two classes both average 70. Class A scores between 65 and 75, class B between 40 and 100. Which is more spread out?
3. Add 100 to 2, 3, 4, 5 (range 3). Which gets dragged harder, the range or the IQR?