Skip to the page
Home Campus
Start free

Standard Deviation

Deviations from the mean, and the standard deviation of a list of data: by both forms of the formula and by calculator (σ, dividing by n); and what a large or a small standard deviation says about the data.

What You Should Already Know

  • From lesson 3.3: the range and the interquartile range measure the spread of data, each from only two values. One very large or very small value changes the range a lot but hardly changes the interquartile range.
  • From Secondary 2: the mean xˉ\bar{x} is the sum of the values divided by the number of values, and Σ\Sigma means "the sum of". For nn values, xˉ=Σxn\bar{x} = \dfrac{\Sigma x}{n}.
  • For example, the values 2, 3, 5, 9, 11 have mean xˉ=305=6\bar{x} = \dfrac{30}{5} = 6 and range 11−2=911 - 2 = 9.

Deviations from the Mean

The range and the interquartile range each use only two values of the data. A measure of spread that uses every value asks how far each value is from the mean.

Deviation from the mean

The deviation of a value xx from the mean xˉ\bar{x} is x−xˉx - \bar{x}.

A value above the mean has a positive deviation, a value below the mean a negative one, and a value equal to the mean a deviation of 0.

The deviations of any set of data add up to 0: the values above the mean balance those below it.

Example 1: Deviations

Find the deviation of each of the values 2, 3, 5, 9, 11 from their mean, and add the deviations.

Solution

The mean is xˉ=2+3+5+9+115=305=6.\bar{x} = \dfrac{2 + 3 + 5 + 9 + 11}{5} = \dfrac{30}{5} = 6\text{.}

The deviations are 2−6=−42 - 6 = -4, 3−6=−33 - 6 = -3, 5−6=−15 - 6 = -1, 9−6=39 - 6 = 3 and 11−6=511 - 6 = 5.

The values 2, 3, 5, 9 and 11 each on its own row, with an arrow from the mean 6 to the value: deviations −4, −3, −1, +3 and +5xx = 62−43−35−19+311+50123456789101112

Each arrow starts at the mean and ends at the value: an arrow to the left is a negative deviation, one to the right a positive one.

Sum: −4−3−1+3+5=0.-4 - 3 - 1 + 3 + 5 = 0\text{.}

The deviations are −4-4, −3-3, −1-1, 3 and 5; they add up to 0.

Because the deviations always add up to 0, their mean is always 0, however spread out the data are: it tells us nothing. So each deviation is squared first. Squaring makes every deviation positive or zero, and a value far from the mean, above or below it, gives a large square.

The Standard Deviation

The standard deviation

The standard deviation σ\sigma (the Greek small letter sigma) of nn values with mean xˉ\bar{x} is

σ=Σ(x−xˉ)2n\sigma = \sqrt{\dfrac{\Sigma(x - \bar{x})^2}{n}}

1. Find the mean xˉ\bar{x}. 2. Find each deviation x−xˉx - \bar{x}. 3. Square each deviation. 4. Add the squares: Σ(x−xˉ)2\Sigma(x - \bar{x})^2. 5. Divide by nn, the number of values. 6. Take the square root.

The square root brings the answer back to the units of the data: if the data are in minutes, σ\sigma is in minutes.

Example 2: The standard deviation of a list

Find the standard deviation of 2, 3, 5, 9, 11.

Solution

From Example 1, xˉ=6\bar{x} = 6. A table keeps the working in order: one row for each value, then the deviation and its square.

xxx−xˉx - \bar{x}(x−xˉ)2(x - \bar{x})^2
22−4-41616
33−3-399
55−1-111
993399
1111552525
Total: 3030006060

Σ(x−xˉ)2=16+9+1+9+25=60,\Sigma(x - \bar{x})^2 = 16 + 9 + 1 + 9 + 25 = 60\text{,} and n=5n = 5.

σ=605=12=3.464…\sigma = \sqrt{\dfrac{60}{5}} = \sqrt{12} = 3.464\ldots

The standard deviation is σ=12≈3.46\sigma = \sqrt{12} \approx 3.46 (to 3 significant figures).

A check on the size of σ

The standard deviation is a kind of typical distance from the mean. It is never less than the smallest distance of a value from the mean, and never more than the largest. In Example 2 the distances are 1 to 5, and σ≈3.46\sigma \approx 3.46 lies between them. An answer of 12, or of 0.35, would be wrong.

A Second Form of the Formula

When the mean is not a whole number, every deviation is a decimal, and squaring them all is slow. The same standard deviation can be worked out from the squares of the values themselves.

The standard deviation: second form

σ=Σx2n−xˉ2\sigma = \sqrt{\dfrac{\Sigma x^2}{n} - \bar{x}^2}

Σx2\Sigma x^2 is the sum of the squares of the values: square each value, then add. It is not (Σx)2(\Sigma x)^2.

The two forms always give the same answer. Use the mean as it is, not rounded: a small error in xˉ\bar{x} becomes a large error once it is squared and subtracted.

Example 3: The same data by the second form

Find the standard deviation of 2, 3, 5, 9, 11 by the second form of the formula.

Solution

Square each value:

xxx2x^2
2244
3399
552525
998181
1111121121
Total: 3030240240

Σx2=4+9+25+81+121=240,\Sigma x^2 = 4 + 9 + 25 + 81 + 121 = 240\text{,} n=5n = 5 and xˉ=6\bar{x} = 6.

σ=2405−62=48−36=12≈3.46\sigma = \sqrt{\dfrac{240}{5} - 6^2} = \sqrt{48 - 36} = \sqrt{12} \approx 3.46

σ≈3.46\sigma \approx 3.46, the same answer as Example 2.

Example 4: A mean that is not a whole number

Find the mean and the standard deviation of the data set A: 10, 11, 10, 11, 10.

Solution

Σx=52\Sigma x = 52 and n=5n = 5, so xˉ=525=10.4\bar{x} = \dfrac{52}{5} = 10.4. The deviations would all be decimals (−0.4-0.4 and 0.60.6), so use the second form.

xxx2x^2
1010100100
1111121121
1010100100
1111121121
1010100100
Total: 5252542542

Σx2=100+121+100+121+100=542.\Sigma x^2 = 100 + 121 + 100 + 121 + 100 = 542\text{.}

σ=5425−10.42=108.4−108.16=0.24=0.4898…\sigma = \sqrt{\dfrac{542}{5} - 10.4^2} = \sqrt{108.4 - 108.16} = \sqrt{0.24} = 0.4898\ldots

Had the mean been rounded to 10, the working would give 108.4−100=8.4≈2.90:\sqrt{108.4 - 100} = \sqrt{8.4} \approx 2.90\text{:} about six times too large. Keep the exact mean.

Mean 10.4; standard deviation σ≈0.490\sigma \approx 0.490 (to 3 significant figures).

Using a Calculator

For more than a few values, the statistics mode of a scientific calculator does the working. The keys differ a little from one model to another, but the steps are these.

The standard deviation on a calculator

1. Put the calculator in statistics mode, for data of one variable, and clear any old data.

2. Key in each value.

3. Open the results. Among other numbers, the calculator lists nn, Σx\Sigma x, Σx2\Sigma x^2, xˉ\bar{x} and two standard deviations. Take only these.

4. Take σx (on some calculators xσn), which divides by nn. This is the standard deviation σ\sigma of this course.

The other one, sx (on some calculators xσn−1), divides by n−1n - 1. It belongs to further statistics, not to this course, and gives a larger answer.

Example 5: Which standard deviation?

Shanti keys 2, 3, 5, 9, 11 into her calculator. It shows xˉ=6\bar{x} = 6, σx =3.464…= 3.464\ldots and sx =3.872…= 3.872\ldots. Which is the standard deviation, and where does the other number come from?

Solution

σx divides by nn: 605=12=3.464…,\sqrt{\dfrac{60}{5}} = \sqrt{12} = 3.464\ldots\text{,} the answer of Example 2.

sx divides by n−1=4n - 1 = 4: 604=15=3.872….\sqrt{\dfrac{60}{4}} = \sqrt{15} = 3.872\ldots\text{.} It is not the standard deviation of this course.

The calculator also shows n=5n = 5, Σx=30\Sigma x = 30 and Σx2=240\Sigma x^2 = 240: checking nn shows that no value was missed or keyed twice.

The standard deviation is σx: σ≈3.46\sigma \approx 3.46. The other, 3.87, is sx, which divides by n−1n - 1.

What the Standard Deviation Tells Us

Reading a standard deviation

A small standard deviation: the values are close to the mean. The data are consistent.

A large standard deviation: the values are spread out from the mean. The data vary a lot.

σ=0\sigma = 0 only when every value is the same; σ\sigma is never negative. It has the units of the data.

The standard deviation describes the spread only. It says nothing about whether the values are large or small: that is the mean's job.

Example 6: The same mean, different spreads

Data set A is 10, 11, 10, 11, 10 and data set B is 4, 8, 11, 13, 16. (a) Show that the two sets have the same mean. (b) Find the standard deviation of B, and compare it with that of A (Example 4).

Solution

(a) Each set has Σx=52\Sigma x = 52 and n=5n = 5, so each mean is 525=10.4\dfrac{52}{5} = 10.4.

(b) Either form will do. Example 4 used the second form; here the first form is used, to show that it works as well when the deviations are decimals with one place.

xxx−xˉx - \bar{x}(x−xˉ)2(x - \bar{x})^2
44−6.4-6.440.9640.96
88−2.4-2.45.765.76
11110.60.60.360.36
13132.62.66.766.76
16165.65.631.3631.36
Total: 52520085.285.2

σ=85.25=17.04=4.127…≈4.13\sigma = \sqrt{\dfrac{85.2}{5}} = \sqrt{17.04} = 4.127\ldots \approx 4.13

Set A, 10, 11, 10, 11 and 10, each on its own row over a scale from 2 to 18, with the mean 10.4 as a dashed line and an arrow from it to each value: every arrow is shorter than 1Set Axx = 10.4101110111024681012141618Set B, 4, 8, 11, 13 and 16, each on its own row over a scale from 2 to 18, with the mean 10.4 as a dashed line and an arrow from it to each value: arrows from 0.6 to 6.4 longSet Bxx = 10.44811131624681012141618

Set A has σ≈0.490\sigma \approx 0.490 and set B σ≈4.13\sigma \approx 4.13. The diagrams show why: A's values are all within 1 of the mean, while B's reach from 4 to 16.

(a) Both means are 10.4. (b) σ≈4.13\sigma \approx 4.13 for B against 0.490 for A: B is far more spread out, and A far more consistent, although their means are the same.

Example 7: One very large value

The 15 driving times, in minutes, of lesson 3.3 are 13, 14, 18, 18, 19, 21, 22, 22, 24, 25, 27, 28, 30, 33, 43. Their interquartile range is 10 minutes. (a) Find their standard deviation. (b) The time of 43 minutes is changed to 93 minutes. Find the new standard deviation, and compare the change with that in the interquartile range.

Solution

(a) By calculator: n=15n = 15, Σx=357\Sigma x = 357, Σx2=9335\Sigma x^2 = 9335, xˉ=23.8\bar{x} = 23.8, and σx =7.476…= 7.476\ldots

Check by the second form: 933515−23.82=622.33…−566.44=55.89…=7.476…\sqrt{\dfrac{9335}{15} - 23.8^2} = \sqrt{622.33\ldots - 566.44} = \sqrt{55.89\ldots} = 7.476\ldots

(b) Now Σx=407\Sigma x = 407, Σx2=16 135\Sigma x^2 = 16\,135, xˉ=27.13…\bar{x} = 27.13\ldots, and σx =18.42…= 18.42\ldots

Lesson 3.3 found that the interquartile range stays 10 minutes: the quartiles do not move.

The standard deviation uses every value, and the value 93 is about 66 minutes from the new mean. Its squared deviation, about 4340, is most of the total, so σ\sigma more than doubles.

(a) σ≈7.48\sigma \approx 7.48 minutes (b) σ≈18.4\sigma \approx 18.4 minutes. One very large value more than doubled the standard deviation, but did not change the interquartile range.

So each measure of spread has its strength. The range and the interquartile range each use only two values. The standard deviation uses every value; for that reason one very large or very small value affects it, while the interquartile range is hardly affected. Lesson 3.7 uses this to choose between them.

Common Mistakes

  • Taking the mean of the deviations as the spread. The deviations always add up to 0, so their mean is always 0: square them first.
  • Forgetting the square root. In Example 2, 605=12\dfrac{60}{5} = 12 is not the standard deviation; 12≈3.46\sqrt{12} \approx 3.46 is.
  • Taking sx on the calculator, which divides by n−1n - 1: 3.87 in Example 5. Take σx (xσn), which divides by nn.
  • Working out (Σx)2(\Sigma x)^2 in place of Σx2\Sigma x^2. For 2, 3, 5, 9, 11, (Σx)2=302=900(\Sigma x)^2 = 30^2 = 900, but Σx2=240\Sigma x^2 = 240.
  • Rounding the mean before using the second form. In Example 4 a mean of 10 in place of 10.4 gives 2.90 in place of 0.490.
  • Taking the square root of Σx2n\dfrac{\Sigma x^2}{n} alone and then subtracting xˉ2\bar{x}^2. The square root covers the whole difference.
  • Reading a large standard deviation as large values. The standard deviation says how spread out the values are, not how large they are.

Practice Questions

Try each question first, then open the solution to check your working. Give each standard deviation to 3 significant figures, and keep the full value until the last line.

Question 1

For the values 4, 7, 7, 10, 12: (a) find the mean and the deviation of each value from the mean, and check that the deviations add up to 0; (b) find the standard deviation.

ShowHide solution

(a) xˉ=405=8\bar{x} = \dfrac{40}{5} = 8. Deviations: −4-4, −1-1, −1-1, 2, 4. Sum: −4−1−1+2+4=0.-4 - 1 - 1 + 2 + 4 = 0\text{.}

The values 4, 7, 7, 10 and 12 each on its own row, with an arrow from the mean 8 to the value: deviations −4, −1, −1, +2 and +4xx = 84−47−17−110+212+402468101214

(b) Squares: 16, 1, 1, 4, 16; Σ(x−xˉ)2=38\Sigma(x - \bar{x})^2 = 38.

σ=385=7.6=2.756…\sigma = \sqrt{\dfrac{38}{5}} = \sqrt{7.6} = 2.756\ldots

(a) Mean 8; deviations −4-4, −1-1, −1-1, 2, 4 (sum 0) (b) σ≈2.76\sigma \approx 2.76

Question 2

Mei Ling cycled to school on six days. Her times, in minutes, were 18, 21, 19, 24, 20, 18. Find the mean and the standard deviation of her times.

ShowHide solution

Σx=120\Sigma x = 120, n=6n = 6, so xˉ=20\bar{x} = 20 minutes. The mean is a whole number, so the first form is quick.

xxx−xˉx - \bar{x}(x−xˉ)2(x - \bar{x})^2
1818−2-244
21211111
1919−1-111
2424441616
20200000
1818−2-244
Total: 120120002626

σ=266=4.333…=2.081…\sigma = \sqrt{\dfrac{26}{6}} = \sqrt{4.333\ldots} = 2.081\ldots

Mean 20 minutes; standard deviation about 2.08 minutes.

Question 3

Find the standard deviation of 3, 4, 4, 5, 7, 9 by the second form of the formula.

ShowHide solution

Σx=32\Sigma x = 32, n=6n = 6, so xˉ=326=163=5.333…,\bar{x} = \dfrac{32}{6} = \dfrac{16}{3} = 5.333\ldots\text{,} a recurring decimal. Keep it as 163\dfrac{16}{3} (or in the calculator's memory).

xxx2x^2
3399
441616
441616
552525
774949
998181
Total: 3232196196

σ=1966−(163)2=32.666…−28.444…=4.222…=2.054…\sigma = \sqrt{\dfrac{196}{6} - \left(\dfrac{16}{3}\right)^2} = \sqrt{32.666\ldots - 28.444\ldots} = \sqrt{4.222\ldots} = 2.054\ldots

σ≈2.05\sigma \approx 2.05

Question 4

Darren keyed the five values 4, 7, 7, 10, 12 into his calculator and wrote "standard deviation =3.08= 3.08". Explain his mistake, and give the right answer.

ShowHide solution

Σ(x−xˉ)2=38\Sigma(x - \bar{x})^2 = 38 (Question 1). Dividing by n−1=4n - 1 = 4 gives 9.5=3.082…\sqrt{9.5} = 3.082\ldots: Darren read sx.

The standard deviation divides by n=5n = 5: 7.6=2.756…\sqrt{7.6} = 2.756\ldots, shown as σx (or xσn).

He took sx, which divides by n−1n - 1. The standard deviation is σx ≈2.76\approx 2.76.

Question 5

Ten pupils recorded the number of books each read in a term. For their numbers xx, Σx=85\Sigma x = 85 and Σx2=777\Sigma x^2 = 777. Find the mean and the standard deviation.

ShowHide solution

xˉ=8510=8.5\bar{x} = \dfrac{85}{10} = 8.5 books.

σ=77710−8.52=77.7−72.25=5.45=2.334…\sigma = \sqrt{\dfrac{777}{10} - 8.5^2} = \sqrt{77.7 - 72.25} = \sqrt{5.45} = 2.334\ldots

Mean 8.5 books; standard deviation about 2.33 books.

Question 6

The five values 2, 4, 6, 8 and kk have mean 6. Find kk, and the standard deviation of the five values.

ShowHide solution

The sum is 5×6=305 \times 6 = 30, so 2+4+6+8+k=30,2 + 4 + 6 + 8 + k = 30\text{,} 20+k=3020 + k = 30 and k=10k = 10.

Deviations: −4-4, −2-2, 0, 2, 4. Squares: 16, 4, 0, 4, 16; sum 40.

σ=405=8=2.828…\sigma = \sqrt{\dfrac{40}{5}} = \sqrt{8} = 2.828\ldots

k=10k = 10; σ≈2.83\sigma \approx 2.83

Question 7

Without working anything out, put these three sets in order of standard deviation, smallest first, and give a reason. Then check by working out the standard deviations of Q and R. P: 5, 5, 5, 5. Q: 4, 5, 6. R: 1, 5, 9.

ShowHide solution

All three sets have mean 5. In P every value is 5, so every deviation is 0 and σ=0\sigma = 0. In Q the values are 1 away from the mean (or 0); in R, 4 away. So P, then Q, then R.

Q: squares 1, 0, 1, sum 2: σ=23=0.816…\sigma = \sqrt{\dfrac{2}{3}} = 0.816\ldots

R: squares 16, 0, 16, sum 32: σ=323=3.265…\sigma = \sqrt{\dfrac{32}{3}} = 3.265\ldots

P (σ=0\sigma = 0), then Q (σ≈0.816\sigma \approx 0.816), then R (σ≈3.27\sigma \approx 3.27): the further the values from the mean, the larger σ\sigma.

Question 8

The ages, in years, of five cousins are 3, 5, 8, 10, 14. (a) Find the mean and the standard deviation of their ages. (b) Without working them out again, write down the mean and the standard deviation of their ages 10 years from now, and explain.

ShowHide solution

(a) xˉ=405=8\bar{x} = \dfrac{40}{5} = 8. Deviations −5-5, −3-3, 0, 2, 6; squares 25, 9, 0, 4, 36; sum 74.

σ=745=14.8=3.847…\sigma = \sqrt{\dfrac{74}{5}} = \sqrt{14.8} = 3.847\ldots

(b) Every age goes up by 10, so the mean goes up by 10, to 18 years. Every value and the mean move together, so every deviation stays the same, and so does σ\sigma.

(a) Mean 8 years; σ≈3.85\sigma \approx 3.85 years (b) mean 18 years; σ≈3.85\sigma \approx 3.85 years still, because the deviations do not change.

Question 9

The masses, in grams, of 8 eggs in a box are 52, 55, 58, 54, 60, 57, 53, 59. (a) Find the mean and the standard deviation. (b) Two more eggs, each of mass 56 g, are put in the box. Without working it out, say whether the standard deviation of the 10 masses is larger or smaller, and why. Then check.

ShowHide solution

(a) Σx=448\Sigma x = 448, so xˉ=4488=56\bar{x} = \dfrac{448}{8} = 56 g. Deviations: −4-4, −1-1, 2, −2-2, 4, 1, −3-3, 3; squares 16, 1, 4, 4, 16, 1, 9, 9; sum 60.

σ=608=7.5=2.738…\sigma = \sqrt{\dfrac{60}{8}} = \sqrt{7.5} = 2.738\ldots g.

(b) The new eggs are at the mean, 56 g, so the mean stays 56 g and their deviations are 0. The sum of the squares stays 60 but is now divided by 10: the standard deviation is smaller.

Check: σ=6010=6=2.449…\sigma = \sqrt{\dfrac{60}{10}} = \sqrt{6} = 2.449\ldots g.

(a) Mean 56 g; σ≈2.74\sigma \approx 2.74 g (b) Smaller: σ≈2.45\sigma \approx 2.45 g, because two values at the mean add nothing to Σ(x−xˉ)2\Sigma(x - \bar{x})^2.

Question 10

The heights of the pupils in class 4X have standard deviation 4.2 cm, and those in class 4Y 7.9 cm. Priya says, "So the pupils in 4Y are taller." Is she right? What do the two standard deviations tell us?

ShowHide solution

No. A standard deviation measures spread, not size. To compare which class is taller on average, we need the means.

The heights in 4Y are more spread out about their mean than those in 4X; the heights in 4X are more alike.

No. The standard deviations say only that 4Y's heights vary more and 4X's are more consistent; the means are needed to say which class is taller.

Key Points

  • The deviation of a value from the mean is x−xˉx - \bar{x}; the deviations always add up to 0.
  • Standard deviation: σ=Σ(x−xˉ)2n,\sigma = \sqrt{\dfrac{\Sigma(x - \bar{x})^2}{n}}\text{,} or σ=Σx2n−xˉ2;\sigma = \sqrt{\dfrac{\Sigma x^2}{n} - \bar{x}^2}\text{;} both give the same answer.
  • Use the exact mean, round only at the end, and give σ\sigma to 3 significant figures in the units of the data.
  • On a calculator in statistics mode, take σx (xσn), which divides by nn, not sx (xσn−1).
  • A small standard deviation: values close to the mean, consistent data. A large one: values spread out.
  • The standard deviation uses every value, so one very large or very small value affects it; the interquartile range is hardly affected.
  • The standard deviation measures spread only, never the size of the values.

Previous lesson3.4 Box-and-Whisker Plots Next lesson3.6 Standard Deviation of Grouped Data

All Secondary 4 (Grade 10) lessons

Home Campus has lessons for Primary 3 to Secondary 4, and practice questions with worked solutions and worksheets for Primary 3 to Secondary 2. Register free to practise and keep track of progress.

This lesson follows the Singapore Secondary 4 syllabus and was checked question by question before it was published. Spotted a mistake? Tell us.