- Foundation
- About 20 min
- 5 practice questions
Standard deviation of frequency distributions and the short-cut method
Calculate variance and standard deviation for discrete and continuous frequency distributions, and use step deviations yᵢ = (xᵢ − A)/h with σₓ = hσᵧ to simplify the work.
If something looks wrong, please tell us.
For grouped data the formulas for variance carry over with frequencies as weights. When the values are large, the arithmetic can be cut down a great deal by shifting and scaling the data first. This lesson does both.
Discrete frequency distributions
σ² = (1/N) Σfᵢ(xᵢ − x̄)² = Σfᵢxᵢ²/N − x̄², where N = Σfᵢ and x̄ = Σfᵢxᵢ/N.
Worked example 1
Use the made-up book-reading data of lesson 2 (x: 2, 4, 6, 8, 10; f: 4, 5, 5, 4, 2), whose mean is 5.5.
| xᵢ | fᵢ | (xᵢ − 5.5)² | fᵢ(xᵢ − 5.5)² | fᵢxᵢ² |
|---|---|---|---|---|
| 2 | 4 | 12.25 | 49 | 16 |
| 4 | 5 | 2.25 | 11.25 | 80 |
| 6 | 5 | 0.25 | 1.25 | 180 |
| 8 | 4 | 6.25 | 25 | 256 |
| 10 | 2 | 20.25 | 40.5 | 200 |
| Total | 20 | 127 | 732 |
By the definition, σ² = 127/20 = 6.35. By the computing formula, σ² = 732/20 − 5.5² = 36.6 − 30.25 = 6.35. σ = √6.35 ≈ 2.52 books.
Continuous frequency distributions
Represent each class by its mid-point and use the same formula. The answer is an estimate, because we assume every value in a class sits at its mid-point.
The short-cut (step-deviation) method
Choose an assumed mean A (a mid-point near the centre) and a scale h (usually the class width), and put yᵢ = (xᵢ − A)/h, so xᵢ = A + hyᵢ. Then:
- Mean: x̄ = Σfᵢ(A + hyᵢ)/N = A + h × ȳ, where ȳ = Σfᵢyᵢ/N.
- Deviations: xᵢ − x̄ = (A + hyᵢ) − (A + hȳ) = h(yᵢ − ȳ).
- Variance: σₓ² = (1/N)Σfᵢh²(yᵢ − ȳ)² = h² σᵧ².
- So σₓ = hσᵧ, and σₓ² = h²[Σfᵢyᵢ²/N − (Σfᵢyᵢ/N)²].
Short-cut method: x̄ = A + h(Σfᵢyᵢ/N) and σ² = h²[Σfᵢyᵢ²/N − (Σfᵢyᵢ/N)²], with yᵢ = (xᵢ − A)/h.
Note what the derivation shows: the shift A does not change the spread at all, while the scale h multiplies the standard deviation by h.
Short-cut method step by step
- Pick A and h A = a central mid-point, h = class width
- Step deviations yᵢ = (xᵢ − A)/h, small whole numbers
- Column sums Σfᵢyᵢ and Σfᵢyᵢ²
- Mean x̄ = A + h(Σfᵢyᵢ/N)
- Variance σ² = h²[Σfᵢyᵢ²/N − (Σfᵢyᵢ/N)²]
Worked example 2: a continuous distribution
Use the made-up marks of lesson 2 (classes 0-10 to 40-50 with frequencies 5, 8, 15, 8, 4; N = 40). Take A = 25 and h = 10.
| Class | fᵢ | xᵢ | yᵢ | fᵢyᵢ | fᵢyᵢ² |
|---|---|---|---|---|---|
| 0-10 | 5 | 5 | −2 | −10 | 20 |
| 10-20 | 8 | 15 | −1 | −8 | 8 |
| 20-30 | 15 | 25 | 0 | 0 | 0 |
| 30-40 | 8 | 35 | 1 | 8 | 8 |
| 40-50 | 4 | 45 | 2 | 8 | 16 |
| Total | 40 | −2 | 52 |
Mean: x̄ = 25 + 10 × (−2/40) = 25 − 0.5 = 24.5 (as found in lesson 2). Variance: σ² = 100 × [52/40 − (−2/40)²] = 100 × [1.3 − 0.0025] = 129.75. σ = √129.75 ≈ 11.39 marks.
Worked example 3: large values
Made-up heights (cm) of 10 plants: 152, 154, 156, 158, 160, 160, 162, 164, 166, 168. Take A = 160, h = 2: y = −4, −3, −2, −1, 0, 0, 1, 2, 3, 4. Σy = 0 and Σy² = 16 + 9 + 4 + 1 + 0 + 0 + 1 + 4 + 9 + 16 = 60. So ȳ = 0, σᵧ² = 60/10 = 6, x̄ = 160 + 2 × 0 = 160 cm and σₓ = 2√6 ≈ 4.90 cm.
Key takeaways
- Grouped data: σ² = Σfᵢxᵢ²/N − x̄², with mid-points for classes.
- Step deviations yᵢ = (xᵢ − A)/h make the numbers small.
- x̄ = A + hȳ and σₓ = hσᵧ: shifting does not change σ; scaling by h multiplies it by h.
Check your understanding
-
1. For x: 0, 1, 2 with f: 1, 2, 1, what is the variance?
- 1
- 0.25
- 2
- 0.5
Show answer
Answer: D. N = 4, x̄ = (0 + 2 + 2)/4 = 1. Σf(x − 1)² = 1(1) + 2(0) + 1(1) = 2. Variance = 2/4 = 0.5.
-
2. In the short-cut method with A = 45 and h = 10, N = 50 and Σfᵢyᵢ = 15. What is the mean?
- 45.3
- 48
- 60
- 46.5
Show answer
Answer: B. x̄ = A + h(Σfᵢyᵢ/N) = 45 + 10 × (15/50) = 45 + 3 = 48.
-
3. If the step deviations yᵢ = (xᵢ − A)/5 have standard deviation 1.6, the standard deviation of the xᵢ is:
- 1.6
- 0.32
- 8
- 6.6
Show answer
Answer: C. σₓ = hσᵧ = 5 × 1.6 = 8.
-
4. With h = 10, N = 20, Σfᵢyᵢ = 0 and Σfᵢyᵢ² = 45, the standard deviation is:
- 15
- 2.25
- √45
- 22.5
Show answer
Answer: A. σ² = h²[Σfᵢyᵢ²/N − (Σfᵢyᵢ/N)²] = 100 × [45/20 − 0] = 100 × 2.25 = 225, so σ = 15.
-
5. Assertion (A): Choosing a different assumed mean A in the short-cut method does not change the standard deviation obtained. Reason (R): Subtracting the same constant from every value shifts the mean by that constant but leaves every deviation from the mean unchanged.
- Both A and R are true, and R is the correct explanation of A
- Both A and R are true, but R is not the correct explanation of A
- A is true, but R is false
- A is false, but R is true
Show answer
Answer: A. Both are true. Deviations xᵢ − x̄ do not depend on where we start counting, so the spread is unchanged by a shift. R explains A.
Sign in to keep track of the lessons you finish.