T1.5 Range, variance, standard deviation, coefficient of variation Start ★ 0
T1.5Playlist 1, episode 5

Same average. Same wait?

T1.4 found one number for the middle of a crowd. Today: a second number, for how far the crowd strays from it. Guess first: you will see your answer again later.

Builds on: T1.4Leads to: T1.6, T3.2, T4.2, T5.2, T6.2
0Last time: the see-saw

Same balance, different crowd

In T1.4, the mean was where the plank balances. Eight waits at each stand (invented): switch stands and watch the triangle.

The mean tells where a crowd balances. It cannot tell how far the dots stray from that point.
Instrument: See-saw, built in T1.4.

1Spread

Same average, different spread

Three everyday pairs. In each pair the average is the same. Answer the question on each card.

0 of 3
Two crowds can share an average and still behave very differently. How far the values stray is called the spread.
In French: dispersion
1Spread

First try: longest minus shortest

Lina’s first five customers waited 1, 3, 4, 5 and 7 minutes (invented). The bracket runs from the shortest wait to the longest. Drag the dots.

Range = largest value − smallest value. Quick, but only the two end dots decide it.
In French: étendue
2The squares

How far from the pivot?

A better idea: use every dot. Each wait gets a bar to the balance point: minus to the left, plus to the right. Drag a dot and add them up.

Distances to the mean always add up to zero: that is why the plank balances. Their plain average can’t measure spread.
In French: écart à la moyenne
2The squares

Each bar grows a square

The trick that stops the cancelling: turn each bar into a square, with the bar as its side. An area is never negative. Drag a dot away and watch its square.

Squaring removes the minus signs, and far dots count extra: twice as far makes a square four times as big.
Instrument: Squares. Back in T3.2, T8.1 and T8.2.
2The squares

One average square

These five waits are everyone Lina served in her first hour: the whole crowd. Three steps pour five squares into one.

Variance = the area of the average square. Standard deviation = its side: back in minutes, like the waits.
In French: variance · écart type

2The squares

One number for the spread

Back to the two stands: eight waits each, taken as the whole crowd. The yellow band reaches one standard deviation to each side of the mean.

The standard deviation says roughly how far a typical value sits from the mean, in the data’s own unit.
Next: this side becomes the tick of the σ-ruler in T1.6.

3Why n − 1

Four free, one forced

Five friends chip in $20 for a gift: $4 each on average. Drag what four of them pay. The friend with the lock pays whatever is left.

Once the mean is known, only n − 1 of the n values are free: the last one is forced. n − 1 is called the degrees of freedom.
In French: degrés de liberté · back in T6.2 and T8.3

3Why n − 1

Draw the bullseye afterwards

Five darts on a wall. The pink diamond is the real bullseye. The green triangle is the middle of your own five darts. The ring shows how far the darts typically land from the spot you measure from.

A small group sits tightest around its own middle. Around the true target it can only look looser.

3Why n − 1

A spoonful does the same

A spoonful of waits can only measure from its own mean. This big jar holds waits of 1 to 7 minutes in equal numbers, so its true mean is 4. Slide the yellow marker: where is the total area smallest?

Measured from its own mean, a sample’s squares are as small as they can be. So a spoonful looks calmer than the jar really is.
Instrument: Jar and spoons, from T1.1.
3Why n − 1

Share among n, or n − 1?

The jar’s true variance is 4. Each spoonful shares its squares two ways: among all n waits, or among n − 1. Scoop many times and compare the long-run averages with 4.

Sharing among n runs too small, by the factor (n − 1)/n. Sharing among n − 1 hits the true variance on average.

!The trap

Variance 4. Four what?

Lina’s five waits: variance 4, standard deviation 2. A customer asks one question. Answer before you look.

The variance is an area, in squared units. Say the standard deviation out loud: it speaks the data’s own unit.

“How far from the 4-minute average is a typical wait?”

Same five waits. Change the unit:

4Compare

Big compared with what?

Two of Lina’s machines (invented). The cup filler pours 500 ml on average, with a standard deviation of 10 ml. The syrup pump gives 30 ml, with a standard deviation of 3 ml.

Coefficient of variation = standard deviation ÷ mean × 100%. It sets a spread next to its own mean, and it has no unit.
In French: coefficient de variation

Which machine is less steady?

5Name it

Name what you did

Each card is something your hand already did. Tap a card, then its name. The book’s symbols appear as you match.

0 of 6
s² and s describe a sample of n values. σ² and σ describe a whole population of N values.
6Real data

Real tables: two spreads

The 244 restaurant tables of T1.2 and T1.4. The green triangle is the mean; the yellow band reaches one standard deviation to each side. The tables are a sample, so the squares are shared among n − 1 = 243.

Read a spread twice: in the data’s unit (standard deviation), and next to its own mean (coefficient of variation).
7Practice

Practice arcade

Six levels, a boss where you build the spread, and a squares lab.

8Remember

Six things to remember, and one trap

Sources and textbook links

The range, variance, standard deviation and coefficient of variation, and the notation s², s, σ², σ, n and N, follow chapter 3 of the course textbook [1]. The 244 restaurant tables are real data [2]. Deviation, variance and standard deviation are defined as in [3]. Two standard results stand behind the n − 1 scenes [4]: the squares are smallest when they are measured from the sample’s own mean, and dividing by n − 1 makes the sample variance right on average. The idea of degrees of freedom follows [5]: each relation the values must satisfy removes one. The two stands, Lina’s waits, the jar of waits, the two machines and the three doors are invented for teaching. “In French” labels give common French statistical vocabulary, not quotations from the book.

  1. Anderson, D. R., Sweeney, D. J., Williams, T. A., Camm, J. D., & Cochran, J. J. (2015). Statistiques pour l’économie et la gestion (5e éd., traduction de la 7e édition américaine d’Essentials of Statistics for Business and Economics par C. Borsenberger). De Boeck Supérieur. Chapitre 3.
  2. Bryant, P. G., & Smith, M. (1995). Practical data analysis: Case studies in business statistics. Irwin. Dataset “tips” (244 tables), as distributed with the R package reshape2 and the seaborn-data repository: https://github.com/mwaskom/seaborn-data
  3. Çetinkaya-Rundel, M., & Hardin, J. (2024). Introduction to Modern Statistics (2nd ed.). OpenIntro. Section 5.4, Variance and standard deviation. https://openintro-ims.netlify.app/explore-numerical
  4. Casella, G., & Berger, R. L. (2002). Statistical inference (2nd ed.). Duxbury. Section 5.2, Theorems 5.2.4 and 5.2.6.
  5. Walker, H. M. (1940). Degrees of freedom. Journal of Educational Psychology, 31(4), 253–269.