70 Statistics for Management
70.1 What is Statistics?
Statistics = the science of collecting, organising, analysing, interpreting and presenting data — Croxton & Cowden. Modern statistics has two branches: Descriptive Statistics (summarising data) and Inferential Statistics (drawing conclusions about populations from samples). Foundational figures: Karl Pearson (correlation), R.A. Fisher (ANOVA, MLE), W.S. Gosset (t-test), Jerzy Neyman (hypothesis testing), C.R. Rao and P.C. Mahalanobis (Indian Statistical Institute, 1931).
70.2 Types of Data
- Qualitative / Categorical — Nominal (gender, brand) · Ordinal (ranking).
- Quantitative — Discrete (count) · Continuous (height, weight).
- Time-series vs Cross-sectional vs Panel data.
- Primary vs Secondary data.
70.3 Levels of Measurement — Stevens (1946)
- Nominal — names, categories (Yes/No).
- Ordinal — ordered ranks (Likert).
- Interval — equal intervals, arbitrary zero (temperature °C).
- Ratio — true zero, ratios meaningful (income, height).
70.4 Measures of Central Tendency
- Mean — arithmetic average; sum/n.
- Median — middle value (50th percentile).
- Mode — most frequent value.
- Geometric Mean — nth root of product; for ratios.
- Harmonic Mean — reciprocal of avg of reciprocals; for rates.
- Quartiles · Deciles · Percentiles — positional measures.
70.5 Measures of Dispersion
- Range = Max − Min.
- Interquartile Range (IQR) = Q3 − Q1.
- Mean Deviation.
- Variance (σ²) = Σ(X − X̄)² / n.
- Standard Deviation (σ) = √Variance.
- Coefficient of Variation (CV) = σ / X̄ × 100.
- Quartile Deviation = (Q3 − Q1)/2.
70.6 Measures of Shape
-
Skewness — asymmetry.
- Positive (right-skewed) — tail on right; Mean > Median > Mode.
- Negative (left-skewed) — tail on left.
- Karl Pearson’s coefficient = (Mean − Mode) / σ; or 3(Mean − Median)/σ.
-
Kurtosis — peakedness.
- Mesokurtic = normal (k = 3).
- Leptokurtic = peaked (k > 3).
- Platykurtic = flat (k < 3).
70.7 Probability
- Classical = favourable/total outcomes.
- Empirical = relative frequency.
- Subjective = personal belief.
- Axiomatic — Kolmogorov (1933).
- Conditional probability = P(A|B) = P(A∩B)/P(B).
- Bayes’ Theorem — Reverend Bayes (1763) — P(A|B) = P(B|A)·P(A)/P(B).
- Independent events — P(A∩B) = P(A)·P(B).
70.8 Probability Distributions
| Distribution | Use case |
|---|---|
| Binomial | Bernoulli trials; success/failure |
| Poisson | Rare events per time period |
| Hypergeometric | Without-replacement sampling |
| Normal (Gauss) | Bell curve; many natural phenomena |
| Standard Normal (Z) | μ=0, σ=1 |
| Student’s t | Small-sample inference; Gosset 1908 |
| Chi-square (χ²) | Goodness of fit, independence |
| F-distribution | ANOVA, variance ratios; Fisher-Snedecor |
| Exponential | Time between events |
| Uniform | Equal probability over range |
70.9 Central Limit Theorem
CLT (Laplace 1810; refined by Lyapunov, Lindeberg-Lévy) — distribution of sample means approaches Normal as sample size n increases, regardless of population distribution shape. n ≥ 30 is rule of thumb.
70.10 Index Numbers
- Laspeyres’ Index — base-year weights.
- Paasche’s Index — current-year weights.
- Fisher’s Ideal Index — √(Laspeyres × Paasche).
- Marshall-Edgeworth — average of base and current weights.
- CPI (Consumer Price Index) — India: CPI-IW, CPI-AL, CPI-RL.
- WPI (Wholesale Price Index).
- IIP (Index of Industrial Production).
- Tests: Time Reversal · Factor Reversal · Circular.
70.11 Time-Series Analysis
- Trend (T) — long-term direction.
- Cyclical (C) — business-cycle fluctuations.
- Seasonal (S) — repeating pattern.
- Irregular (I) — random.
Multiplicative model: Y = T × C × S × I. Additive model: Y = T + C + S + I.
Methods: Moving averages · Exponential smoothing · ARIMA (Box-Jenkins 1970) · Holt-Winters.
70.12 Indian Statistical System
- ISI — Indian Statistical Institute (P.C. Mahalanobis, 1931, Kolkata).
- CSO / NSO — Central Statistical Office, now National Statistical Office (under MoSPI).
- NSSO / NSO Field Operations — surveys.
- MoSPI — Ministry of Statistics and Programme Implementation.
- RBI — financial statistics.
- Census of India — Registrar General.
- Mahalanobis — pioneer of Indian sample surveys; Mahalanobis Distance (1936).
70.13 Modern Trends
- Big Data analytics.
- Bayesian methods revival.
- Machine learning — supervised/unsupervised.
- Causal inference — Pearl, Imbens (Nobel 2021).
- Real-time streaming analytics.
- Reproducible research — R, Python, Jupyter.
- Statistical AI — automated ML.
- Data visualisation — Tableau, Power BI.
70.14 Practice Questions
Indian Statistical Institute was founded in 1931 by:
View solution
Stevens (1946) identified how many measurement scales?
View solution
Central Limit Theorem rule-of-thumb sample size:
View solution
Fisher's Ideal Index is the geometric mean of:
View solution
Bayes' Theorem (1763) computes:
View solution
When Mean > Median > Mode, distribution is:
View solution
CV (Coefficient of Variation) is:
View solution
Student's t-distribution (1908) is by:
View solution
"Mahalanobis Distance" was developed in:
View solution
Time-series components do NOT include:
View solution
Axiomatic probability (1933) is by:
View solution
ARIMA models (1970) are by:
View solution
Median is the:
View solution
India's headline retail inflation is measured by:
View solution
Match:
| (i) | t-distribution | (a) | Mahalanobis |
| (ii) | ANOVA | (b) | Gosset |
| (iii) | Distance | (c) | Kolmogorov |
| (iv) | Axiomatic probability | (d) | R.A. Fisher |
View solution
70.14.1 Advanced Format Questions
A: Median is preferred over mean for skewed data.
R: Mean is affected by extreme values.
View solution
Dispersion measures: (i) Range. (ii) Variance. (iii) SD. (iv) CV.
View solution
Data: 2, 4, 4, 4, 5, 5, 7, 9. Mean is:
View solution
Mean = 50, SD = 5. CV is:
View solution
70.15 Quick Recall
- Statistics — descriptive vs inferential. Pioneers: Pearson · Fisher · Gosset · Neyman · Mahalanobis · C.R. Rao.
- Stevens (1946) 4 scales: Nominal · Ordinal · Interval · Ratio.
- Central tendency: Mean · Median · Mode · GM · HM.
- Dispersion: Range · IQR · MD · Variance · SD · CV = σ/X̄ × 100.
- Skew: positive (Mean>Median>Mode) · negative · Pearson coeff.
- Kurtosis: Meso (3) · Lepto (>3) · Platy (<3).
- Probability: Classical · Empirical · Subjective · Axiomatic (Kolmogorov 1933) · Conditional · Bayes (1763) · Independence.
- Distributions: Binomial · Poisson · Hypergeometric · Normal · t (Gosset 1908) · χ² · F · Exponential · Uniform.
- CLT (Laplace 1810) — n ≥ 30.
- Index: Laspeyres (base) · Paasche (current) · Fisher Ideal (√LP) · Marshall-Edgeworth · CPI · WPI · IIP. Tests: Time/Factor/Circular reversal.
- Time-series: T · C · S · I; ARIMA (Box-Jenkins 1970) · Holt-Winters · Exponential smoothing.
- India: ISI (Mahalanobis 1931) · CSO/NSO/MoSPI · Mahalanobis Distance (1936).
- Modern: Big data · Bayesian revival · ML · Causal inference (Pearl, Imbens Nobel 2021) · streaming · R/Python · Tableau.