Statistics • Topic 3 of 3

Coefficient of Variation and Comparing Variability

Standard deviation tells you the spread of one data set in its own units. But suppose you want to compare the consistency of two sets that are measured differently, or that simply have very different averages — say monthly rainfall against daily temperature, or the run-scoring of a batsman who averages $80$ against one who averages $25$. A raw standard deviation cannot settle that comparison fairly, because a larger mean naturally tends to come with a larger spread.

The fix is to express the spread as a percentage of the mean. This is the coefficient of variation (C.V.), a unit-free, relative measure of dispersion:

$$\text{C.V.}=\dfrac{\sigma}{\bar{x}}\times 100\quad(\bar{x}\neq 0)$$

Because the units cancel in the ratio $\dfrac{\sigma}{\bar{x}}$, the C.V. is a pure number (a percentage) and can compare any two distributions directly. The rule is simple:

$$\text{smaller C.V.}\ \Rightarrow\ \text{more consistent (less variable)}$$

The data set with the greater C.V. is the more variable (less consistent); the one with the smaller C.V. is the more uniform or reliable. The comparison below shows two players with the same average but different consistency:

PlayerMean $\bar{x}$S.D. $\sigma$C.V. $=\dfrac{\sigma}{\bar{x}}\times100$Verdict
A$50$$5$$10\%$more consistent
B$50$$15$$30\%$more variable

When two sets happen to have the same mean, the comparison reduces to comparing standard deviations directly — the means cancel and a smaller $\sigma$ already means greater consistency.

Deeper Insight — why "relative" spread beats "absolute" spread: The coefficient of variation exists because dispersion only means something in proportion to the scale you are working at. A standard deviation of two kilograms is enormous for the weights of medicines but trivial for the weights of adults; the same number tells opposite stories depending on the average it is set against. By dividing by the mean we strip out that scale, leaving a clean statement of "how much does the data wobble relative to its own typical size?" This is exactly why C.V. is the right tool for comparing consistency across players, factories, crops or share prices that operate at different levels. Two cautions worth carrying forward: the C.V. is undefined when $\bar{x}=0$ and becomes unstable when the mean is close to zero, and it is only meaningful for data on a ratio scale with a true zero — comparing the C.V. of temperatures in Celsius would be nonsense, since the choice of zero is arbitrary. Used within those limits, the C.V. turns the vague word "consistent" into a number you can defend.

Comparing two data sets by coefficient of variation Same mean — lower C.V. is more consistent Set A: C.V. = 10%mean Set B: C.V. = 30%mean
1
Worked Example
A data set has mean $\bar{x}=40$ and standard deviation $\sigma=8$. Find its coefficient of variation.
Solution
  1. Apply $\text{C.V.}=\dfrac{\sigma}{\bar{x}}\times 100$.
  2. $\text{C.V.}=\dfrac{8}{40}\times 100=20\%$.

Answer: C.V. $= 20\%$.

2
Worked Example
Two firms A and B pay mean wages of $\bar{x}_A=5250$ and $\bar{x}_B=5200$, with standard deviations $\sigma_A=900$ and $\sigma_B=816$. Which firm has more consistent wages?
Solution
  1. Firm A: $\text{C.V.}_A=\dfrac{900}{5250}\times 100\approx 17.14\%$.
  2. Firm B: $\text{C.V.}_B=\dfrac{816}{5200}\times 100\approx 15.69\%$.
  3. Firm B has the smaller coefficient of variation.

Answer: Firm B has more consistent wages (lower C.V. $\approx 15.69\%$).

3
Worked Example
Batsman X averages $50$ runs with $\sigma=5$; batsman Y averages $40$ runs with $\sigma=5$. Both have the same standard deviation — who is the more consistent scorer?
Solution
  1. Equal $\sigma$ does not mean equal consistency, because the means differ — compare C.V.
  2. X: $\text{C.V.}=\dfrac{5}{50}\times 100=10\%$.
  3. Y: $\text{C.V.}=\dfrac{5}{40}\times 100=12.5\%$.
  4. X has the smaller C.V.

Answer: Batsman X is more consistent (C.V. $=10\%$ vs $12.5\%$).

4
Worked Example
For the data $5, 7, 9, 11, 13$, find the mean, standard deviation, and coefficient of variation.
Solution
  1. Mean $\bar{x}=\dfrac{5+7+9+11+13}{5}=\dfrac{45}{5}=9$.
  2. Deviations $(x_i-9)$: $-4, -2, 0, 2, 4$; squares $16, 4, 0, 4, 16$ summing to $40$.
  3. $\sigma=\sqrt{\dfrac{40}{5}}=\sqrt{8}=2\sqrt{2}\approx 2.83$.
  4. $\text{C.V.}=\dfrac{2.83}{9}\times 100\approx 31.4\%$.

Answer: $\bar{x}=9$, $\sigma\approx 2.83$, C.V. $\approx 31.4\%$.

5
Worked Example
The C.V. of a distribution is $25\%$ and its standard deviation is $12.5$. Find the mean.
Solution
  1. Rearrange $\text{C.V.}=\dfrac{\sigma}{\bar{x}}\times 100$ to give $\bar{x}=\dfrac{\sigma}{\text{C.V.}}\times 100$.
  2. $\bar{x}=\dfrac{12.5}{25}\times 100=50$.

Answer: Mean $\bar{x}=50$.

6
Worked Example
Two crops give yields with $\bar{x}_1=60$, $\sigma_1=9$ and $\bar{x}_2=75$, $\sigma_2=12$. Which crop's yield is more stable?
Solution
  1. Crop 1: $\text{C.V.}_1=\dfrac{9}{60}\times 100=15\%$.
  2. Crop 2: $\text{C.V.}_2=\dfrac{12}{75}\times 100=16\%$.
  3. Crop 1 has the smaller C.V., so its yield varies less relative to its mean.

Answer: Crop 1 is more stable (C.V. $=15\%$ vs $16\%$).

Key Points

  • Coefficient of variation $\text{C.V.}=\dfrac{\sigma}{\bar{x}}\times 100$ is a unit-free, relative measure of spread.
  • It lets you compare variability across data sets with different units or different means.
  • Smaller C.V. means more consistent/uniform; larger C.V. means more variable.
  • When two sets share the same mean, comparing standard deviations alone is enough.
  • C.V. is undefined for $\bar{x}=0$ and only meaningful for ratio-scale data with a true zero.
Tap an option to check your answer0 / 4
Q1.The coefficient of variation (CV) is:
Explanation: CV $=\dfrac{\sigma}{\bar x}\times100$.
Q2.A higher CV indicates the data is:
Explanation: Larger CV $\Rightarrow$ more variability.
Q3.The CV is measured in:
Explanation: CV is a percentage, unit-free.
Q4.To compare the consistency of two data sets, we use:
Explanation: CV compares relative variability.