UPSC EPFO APFC & EO/AO • RBI Grade B • SSC CGL JSO • SEBI Grade A
⚡ Quick Answer & AI Exam Snapshot (TL;DR)
Elementary Statistics is an explicit, ultra-scoring syllabus component in UPSC EPFO APFC (contributing 6 to 10 direct numerical and formula MCQs). Every aspirant must memorize: (1) Relationship between Arithmetic Mean (AM), Geometric Mean (GM), and Harmonic Mean (HM): AM >= GM >= HM (equality holds iff all observations are identical) and GM^2 = AM x HM for two positive numbers; (2) Karl Pearson’s Empirical Relationship for Moderately Skewed Distributions: Mode = 3 Median – 2 Mean (or Mean – Mode = 3[Mean – Median]); (3) Effect of Change of Origin and Scale on Mean vs Standard Deviation (SD is independent of change of origin [adding/subtracting a constant], but changes with scale [multiplying/dividing]); (4) Coefficient of Variation [CV = (SD / Mean) x 100]; and (5) Skewness (Positive: Mean > Median > Mode; Negative: Mean < Median < Mode) & Kurtosis (Leptokurtic beta_2 > 3, Mesokurtic beta_2 = 3, Platykurtic beta_2 < 3).
1. Measures of Central Tendency: AM, GM, HM, Median & Mode (Formulas & Algebraic Properties)
In UPSC EPFO APFC, 3 to 4 questions test the mathematical properties of Arithmetic Mean (AM), Geometric Mean (GM), Harmonic Mean (HM), Median, and Mode:
1. The 5 Measures of Central Tendency & Their Core Formulas:
• Arithmetic Mean ($\bar{X}$ or AM): $\bar{X} = \frac{\sum_{i=1}^{n} x_i}{n}$ (For two numbers $a$ and $b$: $\text{AM} = \frac{a + b}{2}$).
— Property 1: Algebraic sum of deviations of observations from their Arithmetic Mean is ALWAYS ZERO: $\sum (x_i – \bar{X}) = 0$.
— Property 2: Sum of squared deviations $\sum (x_i – A)^2$ is MINIMUM when taken from the Arithmetic Mean ($A = \bar{X}$)! (Whereas the sum of absolute deviations $\sum |x_i – A|$ is MINIMUM when taken from the MEDIAN!).
• Geometric Mean (GM): $\text{GM} = (x_1 \cdot x_2 \cdots x_n)^{1/n}$ (For two numbers $a$ and $b$: $\text{GM} = \sqrt{ab}$). Best measure for compound growth rates, ratios, percentages, and construction of Index Numbers!
• Harmonic Mean (HM): Reciprocal of the arithmetic mean of reciprocals: $\text{HM} = \frac{n}{\sum (1 / x_i)}$ (For two numbers $a$ and $b$: $\text{HM} = \frac{2ab}{a + b}$). Best measure for average speed over equal distances, time, and rates!
• Arithmetic Mean ($\bar{X}$ or AM): $\bar{X} = \frac{\sum_{i=1}^{n} x_i}{n}$ (For two numbers $a$ and $b$: $\text{AM} = \frac{a + b}{2}$).
— Property 1: Algebraic sum of deviations of observations from their Arithmetic Mean is ALWAYS ZERO: $\sum (x_i – \bar{X}) = 0$.
— Property 2: Sum of squared deviations $\sum (x_i – A)^2$ is MINIMUM when taken from the Arithmetic Mean ($A = \bar{X}$)! (Whereas the sum of absolute deviations $\sum |x_i – A|$ is MINIMUM when taken from the MEDIAN!).
• Geometric Mean (GM): $\text{GM} = (x_1 \cdot x_2 \cdots x_n)^{1/n}$ (For two numbers $a$ and $b$: $\text{GM} = \sqrt{ab}$). Best measure for compound growth rates, ratios, percentages, and construction of Index Numbers!
• Harmonic Mean (HM): Reciprocal of the arithmetic mean of reciprocals: $\text{HM} = \frac{n}{\sum (1 / x_i)}$ (For two numbers $a$ and $b$: $\text{HM} = \frac{2ab}{a + b}$). Best measure for average speed over equal distances, time, and rates!
2. Two Golden Identities Linking AM, GM, HM & Mean, Median, Mode:
$$\mathbf{\text{Identity 1:} \quad \text{AM} \ge \text{GM} \ge \text{HM} \quad \text{and for 2 positive numbers:} \quad \text{GM}^2 = \text{AM} \times \text{HM}}$$
$$\mathbf{\text{Identity 2 (Karl Pearson’s Empirical Formula):} \quad \text{Mode} = 3 \times \text{Median} – 2 \times \text{Mean}}$$
Equivalent Rearrangements of Identity 2 (Frequently Tested!):
• $\text{Mean} – \text{Mode} = 3 (\text{Mean} – \text{Median})$
• $\text{Median} = \text{Mode} + \frac{2}{3}(\text{Mean} – \text{Mode}) = \frac{1}{3}\text{Mode} + \frac{2}{3}\text{Mean}$
| Statistical Property | Mathematical Average (AM, GM, HM) | Positional Average (Median & Mode) |
|---|---|---|
| Based on All Observations? | YES (Every single value affects AM, GM, and HM). | NO (Median depends only on the middle rank; Mode depends only on the highest frequency). |
| Affected by Extreme Outliers? | Highly Affected (AM is pulled strongly toward extreme values). | NOT Affected! Median is the best measure of central tendency for Open-Ended Class Intervals and skewed income/wealth distributions! |
| Graphical Location Curve | AM, GM, and HM CANNOT be located graphically! | • Median: Located graphically using Ogives (Cumulative Frequency Curves — intersection of Less-Than and More-Than Ogives). • Mode: Located graphically using a Histogram! |
2. Measures of Dispersion: Range, Quartile Deviation, Mean Deviation, Standard Deviation (SD), Variance & CV
1. Absolute vs Relative Measures of Dispersion:
• Range: $L – S$ (Largest minus Smallest) | Coefficient of Range: $\frac{L – S}{L + S}$.
• Inter-Quartile Range (IQR): $Q_3 – Q_1$ | Quartile Deviation (Semi-Interquartile Range): $\text{QD} = \frac{Q_3 – Q_1}{2}$ | Coefficient of QD: $\frac{Q_3 – Q_1}{Q_3 + Q_1}$. (Note: For a symmetrical/normal distribution, $\text{QD} : \text{MD} : \text{SD} \approx 10 : 12 : 15$, i.e., $\text{QD} = \frac{2}{3}\sigma$ and $\text{MD} = \frac{4}{5}\sigma$!).
• Mean Deviation (MD): $\frac{\sum |x_i – A|}{n}$ (Least when calculated around the Median).
• Standard Deviation ($\sigma$ — Coined by Karl Pearson, 1893; also called Root-Mean-Square Deviation):
$$\sigma = \sqrt{\frac{\sum (x_i – \bar{X})^2}{n}} \quad \Big| \quad \text{Variance} = \sigma^2$$
• Standard Deviation of First $n$ Natural Numbers ($1, 2, 3, \dots, n$):
$$\mathbf{\sigma = \sqrt{\frac{n^2 – 1}{12}} \quad \text{and} \quad \text{Variance (}\sigma^2\text{)} = \frac{n^2 – 1}{12}}$$
• Coefficient of Variation (CV — Relative Measure of Dispersion for Comparing Consistency/Stability):
$$\mathbf{\text{CV (\%)} = \frac{\text{Standard Deviation (}\sigma\text{)}}{\text{Arithmetic Mean (}\bar{X}\text{)}} \times 100}$$
Decision Rule: A dataset/stock/batsman with a LOWER Coefficient of Variation (CV) is MORE CONSISTENT, MORE UNIFORM, and MORE STABLE!
• Range: $L – S$ (Largest minus Smallest) | Coefficient of Range: $\frac{L – S}{L + S}$.
• Inter-Quartile Range (IQR): $Q_3 – Q_1$ | Quartile Deviation (Semi-Interquartile Range): $\text{QD} = \frac{Q_3 – Q_1}{2}$ | Coefficient of QD: $\frac{Q_3 – Q_1}{Q_3 + Q_1}$. (Note: For a symmetrical/normal distribution, $\text{QD} : \text{MD} : \text{SD} \approx 10 : 12 : 15$, i.e., $\text{QD} = \frac{2}{3}\sigma$ and $\text{MD} = \frac{4}{5}\sigma$!).
• Mean Deviation (MD): $\frac{\sum |x_i – A|}{n}$ (Least when calculated around the Median).
• Standard Deviation ($\sigma$ — Coined by Karl Pearson, 1893; also called Root-Mean-Square Deviation):
$$\sigma = \sqrt{\frac{\sum (x_i – \bar{X})^2}{n}} \quad \Big| \quad \text{Variance} = \sigma^2$$
• Standard Deviation of First $n$ Natural Numbers ($1, 2, 3, \dots, n$):
$$\mathbf{\sigma = \sqrt{\frac{n^2 – 1}{12}} \quad \text{and} \quad \text{Variance (}\sigma^2\text{)} = \frac{n^2 – 1}{12}}$$
• Coefficient of Variation (CV — Relative Measure of Dispersion for Comparing Consistency/Stability):
$$\mathbf{\text{CV (\%)} = \frac{\text{Standard Deviation (}\sigma\text{)}}{\text{Arithmetic Mean (}\bar{X}\text{)}} \times 100}$$
Decision Rule: A dataset/stock/batsman with a LOWER Coefficient of Variation (CV) is MORE CONSISTENT, MORE UNIFORM, and MORE STABLE!
2. Effect of Change of Origin (Adding/Subtracting $a$) and Scale (Multiplying by $b$) — Tested Every Year!
Suppose every observation $X$ is transformed into $Y = a + bX$:
• New Mean ($\bar{Y}$): Affected by BOTH Origin and Scale → $\mathbf{\bar{Y} = a + b\bar{X}}$.
• New Standard Deviation ($\sigma_Y$): INDEPENDENT of Change of Origin ($a$), but affected by Scale ($|b|$) → $\mathbf{\sigma_Y = |b| \cdot \sigma_X}$!
• New Variance ($\sigma_Y^2$): Independent of Origin ($a$), multiplied by $b^2$ → $\mathbf{\sigma_Y^2 = b^2 \cdot \sigma_X^2}$!
Example: If the SD of $x_1, x_2, \dots, x_n$ is $5$, and we add $10$ to every number and multiply by $-3$ (i.e., $Y = 10 – 3X$), then the New SD is $|-3| \times 5 = 15$, and the New Variance is $(-3)^2 \times 25 = 9 \times 25 = 225$!
Suppose every observation $X$ is transformed into $Y = a + bX$:
• New Mean ($\bar{Y}$): Affected by BOTH Origin and Scale → $\mathbf{\bar{Y} = a + b\bar{X}}$.
• New Standard Deviation ($\sigma_Y$): INDEPENDENT of Change of Origin ($a$), but affected by Scale ($|b|$) → $\mathbf{\sigma_Y = |b| \cdot \sigma_X}$!
• New Variance ($\sigma_Y^2$): Independent of Origin ($a$), multiplied by $b^2$ → $\mathbf{\sigma_Y^2 = b^2 \cdot \sigma_X^2}$!
Example: If the SD of $x_1, x_2, \dots, x_n$ is $5$, and we add $10$ to every number and multiply by $-3$ (i.e., $Y = 10 – 3X$), then the New SD is $|-3| \times 5 = 15$, and the New Variance is $(-3)^2 \times 25 = 9 \times 25 = 225$!
3. Skewness (Asymmetry), Kurtosis (Peakedness) & Probability Distributions (Binomial, Poisson, Normal)
| Statistical Concept | Types / Formulas | Exact Exam Conditions & Benchmarks |
|---|---|---|
| 1. Skewness (Lack of Symmetry) Karl Pearson’s Coefficient ($S_k$): $S_k = \frac{\text{Mean} – \text{Mode}}{\sigma} = \frac{3(\text{Mean} – \text{Median})}{\sigma}$ Bowley’s Quartile Skewness: $S_B = \frac{Q_3 + Q_1 – 2\text{Median}}{Q_3 – Q_1}$ |
• Symmetrical Distribution ($S_k = 0$) • Positively Skewed (Right-Tailed, $S_k > 0$) • Negatively Skewed (Left-Tailed, $S_k < 0$) |
• Symmetrical (Bell-Shaped): $\mathbf{\text{Mean} = \text{Median} = \text{Mode}}$ and $Q_3 – \text{Median} = \text{Median} – Q_1$. • Positively Skewed (Long tail to the RIGHT): $\mathbf{\text{Mean} > \text{Median} > \text{Mode}}$. • Negatively Skewed (Long tail to the LEFT): $\mathbf{\text{Mean} < \text{Median} < \text{Mode}}$ (i.e., $\text{Mode} > \text{Median} > \text{Mean}$). |
| 2. Kurtosis (Peakedness / Flatness of Curve) Measured by Pearson’s 4th Moment Ratio: $\beta_2 = \frac{\mu_4}{\mu_2^2}$ and $\gamma_2 = \beta_2 – 3$ |
• Mesokurtic (Normal Curve) • Leptokurtic (Tall/Slender Peak, Fat Tails) • Platykurtic (Flat-Topped Curve) |
• Mesokurtic (Normal): $\mathbf{\beta_2 = 3}$ (Excess Kurtosis $\gamma_2 = 0$). • Leptokurtic (More Peaked): $\mathbf{\beta_2 > 3}$ ($\gamma_2 > 0$). • Platykurtic (Flatter than Normal): $\mathbf{\beta_2 < 3}$ ($\gamma_2 < 0$). |
| 3. Binomial, Poisson & Normal Distributions | Mean vs Variance Relationships | • Binomial Distribution ($n, p, q$ where $p + q = 1$): $\text{Mean} = np$ | $\text{Variance} = npq$ → In a Binomial Distribution, Mean is ALWAYS GREATER than Variance ($np > npq$ since $0 < q < 1$)! • Poisson Distribution (Rare Events, parameter $\lambda$): $\mathbf{\text{Mean} = \text{Variance} = \lambda}$! • Normal (Gaussian) Distribution: Symmetrical bell curve where $\mu \pm 1\sigma$ covers 68.27% of area, $\mu \pm 2\sigma$ covers 95.45%, and $\mu \pm 3\sigma$ covers 99.73%! |
🎯 High-Yield Exam One-Liners (Last-Minute Revision Box)
- For any set of positive numbers, Arithmetic Mean (AM) >= Geometric Mean (GM) >= Harmonic Mean (HM), and for two numbers, GM^2 = AM x HM.
- The algebraic sum of deviations of a set of observations from their Arithmetic Mean is ALWAYS zero: Sum(X – Mean) = 0.
- The sum of squared deviations is MINIMUM when measured from the Arithmetic Mean, whereas the sum of absolute deviations (Mean Deviation) is MINIMUM when measured from the Median.
- Geometric Mean (GM) is the best average for measuring growth rates, ratios, and constructing Index Numbers, whereas Harmonic Mean (HM) is best for average speeds over equal distances.
- Median is a positional average that is unaffected by extreme outliers and is the most suitable measure of central tendency for open-end class intervals.
- Median is located graphically using Ogives (Cumulative Frequency Curves), whereas Mode is located graphically using a Histogram.
- Karl Pearson’s empirical formula for a moderately skewed distribution is: Mode = 3 Median – 2 Mean (or Mean – Mode = 3[Mean – Median]).
- Standard Deviation (SD) and Variance are completely INDEPENDENT of a change of origin (adding or subtracting a constant to every value does not change SD or Variance).
- If every observation in a dataset with Standard Deviation ‘s’ is multiplied by a constant ‘k’, the new Standard Deviation becomes |k| x s, and the new Variance becomes k^2 x s^2.
- The Variance of the first ‘n’ natural numbers (1, 2, 3, …, n) is (n^2 – 1) / 12, and the Standard Deviation is sqrt((n^2 – 1) / 12).
- Coefficient of Variation (CV) = (Standard Deviation / Mean) x 100; a lower CV indicates higher consistency, stability, and homogeneity.
- For a symmetrical (Normal) distribution, Mean = Median = Mode; for a Positively Skewed distribution, Mean > Median > Mode; and for a Negatively Skewed distribution, Mean < Median < Mode.
- In a Normal Distribution, Quartile Deviation (QD) = (2/3) x SD, and Mean Deviation (MD) = (4/5) x SD, giving the ratio QD : MD : SD = 10 : 12 : 15.
- Kurtosis is measured by beta_2 = mu_4 / (mu_2)^2: Mesokurtic (Normal) has beta_2 = 3, Leptokurtic (peaked) has beta_2 > 3, and Platykurtic (flat) has beta_2 < 3.
- In a Binomial Distribution (parameters n and p), Mean = np and Variance = npq, so Mean is ALWAYS greater than Variance (np > npq).
- In a Poisson Distribution (used for rare events), Mean is ALWAYS exactly equal to Variance (Mean = Variance = lambda).
- In a Normal Distribution, Mean +/- 1 SD covers 68.27% of observations, Mean +/- 2 SD covers 95.45%, and Mean +/- 3 SD covers 99.73%.
📝 UPSC & Regulatory Level Practice MCQs (With Detailed Solutions)
Q1. In a moderately skewed frequency distribution, the Arithmetic Mean is 45 and the Median is 42. Using Karl Pearson’s empirical relationship, what is the Mode of the distribution, and is the distribution positively or negatively skewed?
Correct Answer: (A) — Using Karl Pearson’s empirical formula: Mode = 3(Median) – 2(Mean) = 3(42) – 2(45) = 126 – 90 = 36. Since Mean (45) > Median (42) > Mode (36), the distribution is Positively Skewed (right-tailed).
Q2. The Standard Deviation of 15 observations is found to be 6. If every observation in the dataset is first increased by 25 and then multiplied by -4 (i.e., Y = -4(X + 25) = -4X – 100), what will be the new Standard Deviation and the new Variance of the transformed dataset?
Correct Answer: (A) — Standard Deviation is completely independent of a change of origin (adding 25 or subtracting 100 has zero effect), and is multiplied by the absolute value of the scale factor |-4| = 4. Since SD can never be negative, New SD = |-4| x 6 = 24, and New Variance = (24)^2 = 576.
Q3. For two positive numbers, if their Arithmetic Mean (AM) is 25 and their Harmonic Mean (HM) is 16, what is their Geometric Mean (GM)?
Correct Answer: (B) — For any two positive numbers, GM^2 = AM x HM. Therefore, GM = sqrt(AM x HM) = sqrt(25 x 16) = sqrt(400) = 20.
Q4. What is the Variance of the first 11 natural numbers (1, 2, 3, …, 11)?
Correct Answer: (A) — The Variance of the first ‘n’ natural numbers is given by the formula (n^2 – 1) / 12. Substituting n = 11: Variance = (11^2 – 1) / 12 = (121 – 1) / 12 = 120 / 12 = 10. (The Standard Deviation is sqrt(10)).
Q5. For a Binomial Distribution with parameters n (number of trials) and p (probability of success), if the Mean of the distribution is 12 and the Standard Deviation is 2 (so Variance = 4), what are the values of ‘p’ (probability of success) and ‘n’ (number of trials)?
Correct Answer: (A) — In a Binomial Distribution, Mean = np = 12 and Variance = npq = 4. Dividing Variance by Mean gives: q = npq / np = 4 / 12 = 1/3. Since p + q = 1, p = 1 – 1/3 = 2/3. Substituting p = 2/3 into np = 12 gives n x (2/3) = 12 -> n = 18.
❓ Frequently Asked Questions (Exam FAQs)
Q: Can the Variance of a Binomial Distribution ever be greater than its Mean?
Never! In a Binomial Distribution, Mean = np and Variance = npq. Since q (probability of failure) is always a fraction strictly between 0 and 1 (0 < q < 1), Variance (npq) is ALWAYS strictly less than Mean (np). If an exam question gives a Binomial Distribution with Variance > Mean, that distribution is mathematically impossible!
Q: Which measure of dispersion is used to compare the variability of two series measured in different units (e.g., height in cm vs weight in kg)?
The Coefficient of Variation (CV = [SD / Mean] x 100), introduced by Karl Pearson, is a unit-free (dimensionless) pure percentage and is the standard relative measure used to compare variability or consistency across datasets with different units or widely different means.
🚀 Continue Your Preparation on CrackTarget
- UPSC EPFO APFC 2026 Hub: Full syllabus notes, 30/60/90-day study planner, PYQs (2012–2025) & free mocks.
- UPSC DRDO SAO Grade-II Hub: 90-day roadmap, Office Procedure, HRM, Marketing & Accounting notes.
- PFRDA Grade A & NABARD Grade A Hub: Paper-2 specialist notes, PYQ vault & real-UI mock tests.
- Complete Free Study Materials Library & CrackTarget Press Book Store: Access 280+ free notes and specialist books.
- Join Official Communities: Telegram Channel | YouTube | Instagram | Facebook | X (Twitter)