Skip to content

Statistics (Paper I — MCQ, part of 20 marks with Mathematics)

MCQ-level: definitions, formulas, standard properties, small mental computations. Also reused in Paper II (noise, random signals) and Financial Management (forecasting).


1. Introduction to Statistics

Statistics is the science of collecting, organizing, presenting, analyzing and interpreting numerical data for decision-making.

Branch Deals With
Descriptive Summarizing data (tables, graphs, averages, dispersion)
Inferential Drawing conclusions about a population from a sample (estimation, hypothesis testing)
Term Meaning
Population Entire group under study
Sample Subset actually observed
Parameter Population measure (μ, σ)
Statistic Sample measure (\(\bar{x}\), s)
Variable Characteristic that varies (quantitative: discrete/continuous · qualitative)

Scales of measurement: Nominal → Ordinal → Interval → Ratio (increasing information).


2. Measures of Central Tendency

Mean (Arithmetic)

\[ \boxed{\bar{x} = \frac{\sum x_i}{n}} \qquad \bar{x} = \frac{\sum f_ix_i}{\sum f_i} \;\text{(frequency data)} \]
  • Uses all observations; affected by extreme values (outliers).
  • Combined mean: \(\bar{x}_{12} = \frac{n_1\bar{x}_1 + n_2\bar{x}_2}{n_1 + n_2}\)

Median

Middle value of ordered data — positional average.

\[ \text{Median} = \left(\frac{n+1}{2}\right)^{th} \text{value (odd } n\text{)}; \quad \text{mean of two middle values (even } n\text{)} \]

Grouped: \(Md = L + \frac{\frac{N}{2} - cf}{f} \times h\)

  • Not affected by extreme values → best for skewed data (income!).

Mode

Most frequent value. Grouped: \(Mo = L + \frac{f_1 - f_0}{2f_1 - f_0 - f_2} \times h\)

Empirical Relation (favorite MCQ)

\[ \boxed{\text{Mode} = 3\,\text{Median} - 2\,\text{Mean}} \]

Skewness and the Three Averages

Distribution Relation
Symmetrical Mean = Median = Mode
Positively skewed (right tail) Mean > Median > Mode
Negatively skewed (left tail) Mean < Median < Mode

Other means: GM \(= \sqrt[n]{x_1x_2\cdots x_n}\) (growth rates), HM \(= \frac{n}{\sum 1/x_i}\) (rates/speeds); \(AM \ge GM \ge HM\).


3. Measures of Dispersion

Dispersion measures the spread/scatter of data around the center.

Measure Formula Notes
Range \(X_{max} - X_{min}\) Crude, quick
Quartile deviation \(\frac{Q_3 - Q_1}{2}\) Middle 50% spread
Mean deviation \(\frac{\sum\|x_i - \bar{x}\|}{n}\) Absolute deviations
Variance \(\sigma^2 = \frac{\sum(x_i - \bar{x})^2}{n} = \frac{\sum x_i^2}{n} - \bar{x}^2\) Squared units
Standard deviation \(\sigma = \sqrt{\sigma^2}\) Same units as data; most used
Coefficient of variation \(CV = \frac{\sigma}{\bar{x}} \times 100\%\) Unit-free — compares consistency of two series; smaller CV = more consistent

Properties of σ: unaffected by change of origin (\(x \pm a\)), scaled by change of scale (\(\sigma_{kx} = |k|\sigma_x\)). Sample variance uses \(n - 1\) (unbiased).


4. Correlation

Correlation measures the degree of linear relationship between two variables.

Karl Pearson's Coefficient

\[ \boxed{r = \frac{\sum(x-\bar{x})(y-\bar{y})}{\sqrt{\sum(x-\bar{x})^2\sum(y-\bar{y})^2}} = \frac{Cov(x,y)}{\sigma_x\sigma_y}} \]

Properties (MCQ favorites):

  • \(-1 \le r \le +1\)
  • \(r = +1\) perfect positive, \(r = -1\) perfect negative, \(r = 0\) no linear relation
  • Unit-free; unaffected by change of origin and scale
  • Correlation ≠ causation

Spearman's Rank Correlation

\[ \boxed{\rho = 1 - \frac{6\sum d_i^2}{n(n^2 - 1)}} \]

(\(d_i\) = rank differences) — for ordinal/qualitative data.


5. Regression

Regression estimates the average value of one variable for a given value of the other (prediction line).

Regression line of y on x: \(y - \bar{y} = b_{yx}(x - \bar{x})\), where

\[ \boxed{b_{yx} = r\frac{\sigma_y}{\sigma_x}} \qquad \boxed{b_{xy} = r\frac{\sigma_x}{\sigma_y}} \]

Properties:

  • \(r = \pm\sqrt{b_{yx} \cdot b_{xy}}\) (sign = sign of the b's; both b's always same sign)
  • Both lines intersect at \((\bar{x}, \bar{y})\)
  • If \(r = \pm 1\): the two lines coincide; if \(r = 0\): lines are perpendicular
Correlation Regression
Degree of relationship Nature/equation of relationship
Symmetric (\(r_{xy} = r_{yx}\)) Not symmetric (\(b_{yx} \ne b_{xy}\))
No prediction Used for prediction/forecasting

6. Probability

Definitions

\[ \boxed{P(A) = \frac{\text{favorable outcomes}}{\text{total outcomes}}} \qquad 0 \le P(A) \le 1, \quad P(\bar{A}) = 1 - P(A) \]

Laws

Addition:

\[ P(A \cup B) = P(A) + P(B) - P(A \cap B) \]

(mutually exclusive: \(P(A \cap B) = 0\))

Multiplication:

\[ P(A \cap B) = P(A)\,P(B|A) \]

(independent: \(P(A \cap B) = P(A)P(B)\))

Conditional: \(P(A|B) = \frac{P(A \cap B)}{P(B)}\)

Bayes' theorem:

\[ \boxed{P(A_i|B) = \frac{P(A_i)P(B|A_i)}{\sum_j P(A_j)P(B|A_j)}} \]

Quick values: two dice sum 7 → \(\frac{6}{36} = \frac{1}{6}\); at least one head in 2 tosses → \(\frac{3}{4}\); 52-card deck: P(ace) \(= \frac{1}{13}\).


7. Random Variables and Distributions

Random Variable

A random variable assigns a numerical value to each outcome of a random experiment. Discrete (countable values, PMF) or continuous (interval values, PDF).

\[ E(X) = \sum x_ip_i \qquad Var(X) = E(X^2) - [E(X)]^2 \]

Properties: \(E(aX + b) = aE(X) + b\); \(Var(aX + b) = a^2Var(X)\).

Standard Distributions

Distribution Type PMF/PDF Mean Variance Models
Binomial \(B(n,p)\) Discrete \(\binom{n}{x}p^xq^{n-x}\) \(np\) \(npq\) n independent success/fail trials (bit errors in a frame)
Poisson \(P(\lambda)\) Discrete \(\frac{e^{-\lambda}\lambda^x}{x!}\) \(\lambda\) \(\lambda\) Rare events per interval (call arrivals — links to Erlang)
Normal \(N(\mu,\sigma^2)\) Continuous bell curve \(\mu\) \(\sigma^2\) Noise, measurement errors
Uniform Continuous \(\frac{1}{b-a}\) \(\frac{a+b}{2}\) \(\frac{(b-a)^2}{12}\) Quantization error
Exponential Continuous \(\lambda e^{-\lambda x}\) \(\frac{1}{\lambda}\) \(\frac{1}{\lambda^2}\) Holding/service times

Facts: mean = variance ⇒ Poisson; binomial → Poisson as \(n\to\infty, p\to 0\), \(np = \lambda\); binomial → normal for large \(n\).

Normal Distribution Properties (MCQ favorites)

  • Symmetric, bell-shaped; mean = median = mode
  • Standard normal: \(Z = \frac{X - \mu}{\sigma}\), \(N(0, 1)\)
  • 68–95–99.7 rule: \(\mu \pm 1\sigma\) → 68.3%, \(\mu \pm 2\sigma\) → 95.4%, \(\mu \pm 3\sigma\) → 99.7%
  • Total area = 1; points of inflection at \(\mu \pm \sigma\)

8. Samples and Sampling

Concepts

Sampling is selecting a representative subset to estimate population characteristics — cheaper, faster, sometimes the only option (destructive testing).

Random (probability) Methods Non-random Methods
Simple random (lottery/random numbers) Convenience
Systematic (every \(k^{th}\), \(k = N/n\)) Judgment/purposive
Stratified (divide into homogeneous strata, sample each) Quota
Cluster (sample whole groups) Snowball

Errors

  • Sampling error — chance difference between sample and population (decreases as \(n\) increases)
  • Non-sampling error — bias, bad questionnaire, non-response (can occur even in census)

Key Results

  • Standard error of mean: \(SE = \frac{\sigma}{\sqrt{n}}\) — halving SE needs sample
  • Central Limit Theorem: for large \(n\) (≥ 30), the sampling distribution of \(\bar{x}\) is approximately normal regardless of population shape
  • Census vs sample survey: complete coverage/costly vs partial/economical

9. Speed Facts for MCQs

Fact Value
Mode (empirical) \(3Md - 2\bar{x}\)
\(AM \ge GM \ge HM\) Equality iff all values equal
CV \(\frac{\sigma}{\bar{x}} \times 100\), unit-free
\(r\) range \([-1, +1]\); \(r = \pm\sqrt{b_{yx}b_{xy}}\)
Regression lines meet at \((\bar{x}, \bar{y})\)
Binomial mean/var \(np\), \(npq\) (var < mean)
Poisson mean = var \(\lambda\)
Normal areas 68 – 95 – 99.7
SE of mean \(\sigma/\sqrt{n}\)
Two dice sum 7 1/6

Key Exam Points — Statistics

  • Median resists outliers; mean uses all data; empirical relation Mode = 3Median − 2Mean.
  • CV (not σ) compares consistency between two data sets.
  • Correlation: strength of linear relation; regression: prediction equation; \(r = \pm\sqrt{b_{yx}b_{xy}}\).
  • Binomial (np, npq) vs Poisson (λ, λ) vs Normal (68-95-99.7) — the distribution trio covers most MCQs.
  • CLT and \(SE = \sigma/\sqrt{n}\) are the two sampling facts asked repeatedly.