- Advanced
- About 18 min
- 5 practice questions
Deviation identities and their shortcuts
Use the identities behind variance and mean deviation to answer hard questions in one line: sums of squared deviations about any point, variances of arithmetic progressions, and why the mean deviation never exceeds the standard deviation.
If something looks wrong, please tell us.
At the advanced level, statistics questions are short to state and quick to solve if you know the identities behind the formulas, and painfully long if you do not. This lesson collects those identities, proves the ones you need, and shows the one-line answers they give. As in the earlier courses, variance divides by n (the NCERT convention): σ² = Σ(x − x̄)²/n.
Sum of squared deviations about any point
Write x − a = (x − x̄) + (x̄ − a), square and add. The cross term is 2(x̄ − a)Σ(x − x̄), which is zero because deviations from the mean add to zero. So Σ(x − a)² = Σ(x − x̄)² + n(x̄ − a)² = n[σ² + (x̄ − a)²]. Two consequences follow at once: the sum of squared deviations is least when a = x̄, and you can compute Σ(x − a)² without the data.
Σ(x − a)² = n[σ² + (x̄ − a)²], least at a = x̄. Σ|x − a| is least at a = median. For the first n natural numbers σ² = (n² − 1)/12; for n terms of an AP with common difference d, σ² = d²(n² − 1)/12. MD about the mean ≤ σ.
Worked example 1
Ten observations have mean 8 and variance 4. Find Σ(x − 5)². By the identity, Σ(x − 5)² = 10[4 + (8 − 5)²] = 10 × 13 = 130. No observation is needed.
Variance of an arithmetic progression
For 1, 2, …, n: Σx = n(n + 1)/2 and Σx² = n(n + 1)(2n + 1)/6. So σ² = (n + 1)(2n + 1)/6 − (n + 1)²/4 = (n + 1)[2(2n + 1) − 3(n + 1)]/12 = (n + 1)(n − 1)/12 = (n² − 1)/12. Any AP a, a + d, …, a + (n − 1)d is a + d × (0, 1, …, n − 1), a shift and a scale of consecutive integers, so its variance is d²(n² − 1)/12; the first term a does not matter.
Worked example 2
The variance of 1, 2, …, 10 is 99/12 = 8.25. The ten numbers 5, 9, 13, …, 41 form an AP with d = 4, so their variance is 16 × 99/12 = 132 and their standard deviation is √132 ≈ 11.49. Conversely, if the variance of 1, 2, …, n is 10, then n² − 1 = 120 and n = 11.
Mean deviation never exceeds standard deviation
Let the distances from the mean be |d₁|, …, |dₙ|. Their mean is MD and the mean of their squares is σ². The variance of these distances is σ² − MD², and a variance cannot be negative, so MD ≤ σ, with equality only when all distances are equal. For 1, 2, 3, 4, 10 the mean is 4, MD = (3 + 2 + 1 + 0 + 6)/5 = 2.4 and σ² = (9 + 4 + 1 + 0 + 36)/5 = 10, so σ ≈ 3.16. The inequality is a quick check on any answer that gives both.
Worked example 3: the median minimises absolute deviations
For 2, 3, 7, 8, 20, the median is 7 and Σ|x − 7| = 5 + 4 + 0 + 1 + 13 = 23; at any other a the sum is larger. Moving a away from the median by a small step brings it closer to fewer than half of the values and further from more than half, so the sum grows.
- Variance is unchanged by adding a constant, so shift data to small numbers first.
- Σ(x − x̄) = 0 always; Σ(x − x̄)² = nσ².
- Variance 0 means every observation equals the mean.
Key takeaways. The identity Σ(x − a)² = n[σ² + (x̄ − a)²] replaces long calculations. The variance of an AP depends only on n and d. The median minimises the sum of absolute deviations, and MD about the mean never exceeds σ.
Check your understanding
-
1. What is the variance of the first 20 natural numbers?
- 35
- 33
- 33.25
- 36.75
Show answer
Answer: C. σ² = (n² − 1)/12 = 399/12 = 33.25.
-
2. What is the standard deviation of 3, 7, 11, 15, 19, 23, 27?
- 64
- 8
- 4
- 2√7
Show answer
Answer: B. An AP with d = 4 and n = 7: σ² = 16 × 48/12 = 64, so σ = 8.
-
3. Twenty observations have mean 15 and standard deviation 3. What is Σ(x − 10)²?
- 680
- 500
- 180
- 580
Show answer
Answer: A. Σ(x − 10)² = 20[9 + (15 − 10)²] = 20 × 34 = 680.
-
4. For any data set, which statement is always true (MD taken about the mean, variance dividing by n)?
- MD ≥ σ
- MD = σ²
- MD ≤ σ²
- MD ≤ σ
Show answer
Answer: D. σ² − MD² is the variance of the absolute deviations, which is never negative, so MD ≤ σ. The others fail, e.g. for 0, 1 (MD = σ = 0.5 > σ² = 0.25) or for 1, 2, 3, 4, 10 (MD 2.4 < σ ≈ 3.16).
-
5. For the data 2, 3, 7, 8, 20, what is the least possible value of Σ|x − a|?
- 24
- 20
- 23
- 25
Show answer
Answer: C. The least value occurs at the median, 7: 5 + 4 + 0 + 1 + 13 = 23.
Sign in to keep track of the lessons you finish.