Skip to content

Probability and Probability Distributions

UGC NET Commerce: classical, relative frequency, subjective, axiomatic probability, addition, multiplication rules, Bayes' theorem, binomial, Poisson, normal.

  • 18 min read
  • Updated
On this page · 18 sections
  1. Prerequisites
  2. The Hook: Which Machine Made the Defective Shirt?
  3. 1. Basic Terms
  4. 2. Approaches to Probability
  5. 3. Rules of Probability
  6. 4. Bayes’ Theorem
  7. 5. Random Variables and Expectation
  8. 6. The Binomial Distribution
  9. 7. The Poisson Distribution
  10. 8. The Normal Distribution
  11. 9. The Three Distributions Compared
  12. Memory Tricks
  13. Don’t Confuse
  14. Common Mistakes
  15. Practice
  16. Quick Revision Table
  17. Key Takeaways
  18. What’s Next

Prerequisites

The Hook: Which Machine Made the Defective Shirt?


1. Basic Terms

TermMeaningExample (one throw of a die)
Random experimentOutcome uncertain, possible outcomes knownThrowing a die
Sample space (S)Set of all possible outcomes{1, 2, 3, 4, 5, 6}
EventA subset of the sample space“Even number” = {2, 4, 6}
Mutually exclusive eventsCannot happen together“1” and “6” on one throw
Exhaustive eventsTogether cover the whole sample space“Odd” and “even”
Equally likely eventsNo reason to expect one more than anotherEach face of a fair die
Independent eventsOccurrence of one does not change the probability of the otherHeads on a coin and 6 on a die
Complementary event“Not A”, written \( A' \) or \( \bar{A} \); \( P(A') = 1 - P(A) \)“Not 6”

Probability always lies between 0 (impossible event) and 1 (certain event).


2. Approaches to Probability

ApproachDefinitionAlso calledLimitations
Classical\( P(A) = \dfrac{\text{favourable outcomes}}{\text{total equally likely outcomes}} \)A priori, mathematical; associated with LaplaceNeeds a finite number of equally likely outcomes; the definition is circular (“equally likely” already means “equally probable”)
Relative frequencyLimit of the proportion of times A occurs as trials increase indefinitely: \( P(A) = \lim_{n \to \infty} \dfrac{m}{n} \)Empirical, statistical, a posteriori; developed by Richard von MisesNeeds repeatable trials under identical conditions; the limit is never actually observed
SubjectiveA person’s degree of belief, based on judgement and experiencePersonal, judgementalDifferent people give different values; hard to verify
AxiomaticProbability is any function satisfying a set of axiomsModern; A.N. Kolmogorov (1933)Does not tell you how to assign the values, only the rules they must obey

Kolmogorov’s axioms: (1) \( P(A) \ge 0 \) for every event; (2) \( P(S) = 1 \); (3) for mutually exclusive events, \( P(A \cup B \cup \cdots) = P(A) + P(B) + \cdots \). The classical and relative-frequency definitions both satisfy these axioms, which is why the axiomatic approach includes them.


3. Rules of Probability

Addition theorem

For any two events: \( P(A \cup B) = P(A) + P(B) - P(A \cap B) \).

For mutually exclusive events, \( P(A \cap B) = 0 \), so \( P(A \cup B) = P(A) + P(B) \).

For three events: \( P(A \cup B \cup C) = P(A) + P(B) + P(C) - P(A \cap B) - P(B \cap C) - P(A \cap C) + P(A \cap B \cap C) \).

Conditional probability and multiplication theorem

$$P(A \mid B) = \frac{P(A \cap B)}{P(B)}, \qquad P(A \cap B) = P(B)\,P(A \mid B) = P(A)\,P(B \mid A)$$

For independent events, \( P(A \mid B) = P(A) \), so \( P(A \cap B) = P(A)\,P(B) \).

Total probability

If \( E_1, E_2, \ldots, E_n \) are mutually exclusive and exhaustive, then for any event A:

$$P(A) = \sum_{i} P(E_i)\,P(A \mid E_i)$$

4. Bayes’ Theorem

Named after Thomas Bayes, whose essay was published posthumously in 1763, the theorem reverses a conditional probability: from \( P(A \mid E_i) \) it gives \( P(E_i \mid A) \).

$$P(E_i \mid A) = \frac{P(E_i)\,P(A \mid E_i)}{\sum_j P(E_j)\,P(A \mid E_j)}$$
TermMeaning
\( P(E_i) \)Prior probability: before the evidence
\( P(A \mid E_i) \)Likelihood: chance of the evidence if \( E_i \) is the cause
\( P(E_i \mid A) \)Posterior probability: revised after the evidence

It is the probability of causes, and the base of decision-making that updates beliefs as information arrives (credit scoring, spam filters, medical tests, audit sampling).


5. Random Variables and Expectation

A random variable assigns a number to each outcome. It is discrete (countable values: number of defectives) or continuous (any value in a range: weight, time).

  • Expected value: \( E(X) = \sum x\,p(x) \), the long-run average.
  • Variance: \( V(X) = E(X^2) - [E(X)]^2 \).
  • Rules: \( E(aX + b) = aE(X) + b \); \( V(aX + b) = a^2 V(X) \).

6. The Binomial Distribution

Associated with Jacob Bernoulli (Ars Conjectandi, published posthumously in 1713). It counts the number of successes in n independent trials.

Conditions (Bernoulli trials): a fixed number n of trials; each trial has only two outcomes (success, failure); the probability of success p is constant across trials; trials are independent.

$$P(X = r) = \binom{n}{r} p^r q^{\,n-r}, \qquad r = 0, 1, \ldots, n, \quad q = 1 - p$$
PropertyValue
Parametersn and p
Mean\( np \)
Variance\( npq \) (so variance < mean, since q < 1)
SD\( \sqrt{npq} \)
ShapeSymmetric if p = 0.5; positively skewed if p < 0.5; negatively skewed if p > 0.5
Skewness (moment)\( \beta_1 = \dfrac{(q - p)^2}{npq} \)
ModeValue of r around \( (n + 1)p \)
TypeDiscrete

7. The Poisson Distribution

Named after Siméon Denis Poisson (1837). It models the number of occurrences of a rare event in a fixed interval of time, space or quantity.

$$P(X = r) = \frac{e^{-\lambda}\lambda^r}{r!}, \qquad r = 0, 1, 2, \ldots$$
PropertyValue
ParameterOne: \( \lambda \) (= m), the mean number of occurrences
Mean\( \lambda \)
Variance\( \lambda \) (mean = variance)
ShapeAlways positively skewed; approaches symmetry as \( \lambda \) grows
Skewness (moment)\( \beta_1 = 1/\lambda \)
TypeDiscrete, values 0 to infinity
As a limit of binomialWhen \( n \to \infty \), \( p \to 0 \) and \( np = \lambda \) stays finite

Typical uses: defects per metre of cloth, misprints per page, accidents per month at a crossing, calls arriving at a help desk per minute, customers arriving at a counter (queuing theory).


8. The Normal Distribution

The bell-shaped curve. Its first form, as the limit of the binomial, was given by Abraham de Moivre (1733); it was later developed by Laplace and Gauss, which is why it is also called the Gaussian distribution.

$$f(x) = \frac{1}{\sigma\sqrt{2\pi}}\, e^{-\frac{(x - \mu)^2}{2\sigma^2}}, \qquad -\infty < x < \infty$$

Properties

  1. Bell-shaped and symmetric about the mean; mean = median = mode.
  2. Two parameters: \( \mu \) and \( \sigma \). The curve is fixed once both are known.
  3. Asymptotic to the X-axis: the tails approach but never touch it.
  4. Points of inflection at \( \mu \pm \sigma \).
  5. Total area under the curve = 1; half on each side of the mean.
  6. Skewness zero (\( \beta_1 = 0 \)); mesokurtic (\( \beta_2 = 3 \)).
  7. QD \( \approx 0.6745\sigma \) (about \( \frac{2}{3}\sigma \)); MD \( \approx 0.7979\sigma \) (about \( \frac{4}{5}\sigma \)).
  8. A linear function of a normal variable is normal; the sum of independent normal variables is normal.

Area rule

RangeArea (approx.)
\( \mu \pm 1\sigma \)68.27%
\( \mu \pm 2\sigma \)95.45%
\( \mu \pm 3\sigma \)99.73%
\( \mu \pm 1.96\sigma \)95%
\( \mu \pm 2.58\sigma \)99%

Standard normal variable

$$Z = \frac{X - \mu}{\sigma}, \qquad Z \sim N(0, 1)$$

Areas from the mean (0) to z, from the standard table: z = 1 → 0.3413; 1.5 → 0.4332; 1.645 → 0.4500; 1.96 → 0.4750; 2 → 0.4772; 2.58 → 0.4951.

Normal approximation to binomial: when n is large and p not too close to 0 or 1 (a common rule of thumb: np and nq both at least 5), use \( \mu = np \), \( \sigma = \sqrt{npq} \), with a continuity correction of 0.5 (for “at least 30”, use 29.5).


9. The Three Distributions Compared

BasisBinomialPoissonNormal
TypeDiscreteDiscreteContinuous
Parametersn, p\( \lambda \)\( \mu \), \( \sigma \)
Meannp\( \lambda \)\( \mu \)
Variancenpq (< mean)\( \lambda \) (= mean)\( \sigma^2 \)
ShapeSymmetric only if p = 0.5Always positively skewedAlways symmetric
UseFixed number of yes/no trialsRare events in an intervalMeasurements: heights, errors, sales, sampling distributions
Associated withJacob BernoulliS.D. Poissonde Moivre, Laplace, Gauss
graph LR
    A["Binomial
n, p"] -->|"n large, p small,
np = λ"| B["Poisson
λ"] A -->|"n large,
p near 0.5"| C["Normal
μ = np, σ² = npq"] B -->|"λ large"| C

Memory Tricks

Don’t Confuse

ThisvsThat
Mutually exclusive: cannot occur togethervsIndependent: one does not affect the other’s probability
Prior probabilityvsPosterior probability (after evidence, by Bayes)
Classical: a priori, equally likelyvsRelative frequency: a posteriori, long-run proportion
Binomial: fixed n, variance < meanvsPoisson: rare events, variance = mean
Discrete (binomial, Poisson)vsContinuous (normal)
±1.96σ = 95%vs±2σ ≈ 95.45%
Points of inflection: \( \mu \pm \sigma \)vsTails: asymptotic, never touch the axis

Common Mistakes


Practice

All questions below are original practice questions written in the UGC NET format. They are not previous-year questions.

Level 1: Foundation

  1. P(A) = 0.5, P(B) = 0.4 and P(A ∩ B) = 0.2. Find P(A ∪ B).
  2. A Poisson variable has variance 3. What is its mean?
  3. Which approach to probability did Kolmogorov formalise?
Answers
  1. \( 0.5 + 0.4 - 0.2 = \) 0.7. (A and B are also independent here, since \( 0.5 \times 0.4 = 0.2 \).)
  2. 3: mean = variance in a Poisson distribution.
  3. The axiomatic approach (1933).

Level 2: Exam-style MCQs

Q1. (Practice) In a binomial distribution the mean is 6 and the variance is 4.2. The number of trials n is:

  • (a) 10
  • (b) 15
  • (c) 20
  • (d) 30
Answer and explanation

(c) 20. \( q = 4.2/6 = 0.7 \), so \( p = 0.3 \) and \( n = 6/0.3 = 20 \).

Q2. (Practice) Daily demand for a product is normally distributed with mean 500 units and SD 50 units. The probability that demand on a day exceeds 600 units is approximately (area from 0 to z = 2 is 0.4772):

  • (a) 0.0228
  • (b) 0.4772
  • (c) 0.9772
  • (d) 0.0456
Answer and explanation

(a) 0.0228. \( z = (600 - 500)/50 = 2 \); area beyond \( = 0.5 - 0.4772 = 0.0228 \). Option (d) is the two-tailed area, a trap.

Q3. (Practice) Two firms bid for a contract. The probability that firm X wins is 0.4 and that firm Y wins is 0.35; only one can win. The probability that neither wins is:

  • (a) 0.14
  • (b) 0.25
  • (c) 0.39
  • (d) 0.75
Answer and explanation

(b) 0.25. The events are mutually exclusive: \( P(X \cup Y) = 0.75 \), so \( P(\text{neither}) = 1 - 0.75 = 0.25 \). Option (c), \( 0.6 \times 0.65 \), wrongly treats them as independent.

Q4. (Practice) Consider the following statements about the Poisson distribution:

  1. Its mean and variance are equal.
  2. It is a limiting form of the binomial when n is large and p is small.
  3. It is always symmetrical.

Which of the statements given above is/are correct?

  • (a) 1 only
  • (b) 1 and 2 only
  • (c) 2 and 3 only
  • (d) 1, 2 and 3
Answer and explanation

(b) 1 and 2 only. The Poisson distribution is positively skewed (\( \beta_1 = 1/\lambda \)).

Q5. (Practice) In the three-machine example (A 50%, B 30%, C 20% of output; defect rates 1%, 2%, 5%), the probability that a defective item came from machine A is:

  • (a) 0.50
  • (b) 0.238
  • (c) 0.005
  • (d) 0.021
Answer and explanation

(b) 0.238. \( P(A \mid D) = 0.005/0.021 = 5/21 \approx 0.238 \). Option (c) is the joint probability, (d) the total probability of a defect, and (a) the prior.

Level 3: Assertion–reason, match and sequence

For assertion–reason questions, use these options:

  • (a) Both A and R are true, and R is the correct explanation of A
  • (b) Both A and R are true, but R is not the correct explanation of A
  • (c) A is true, but R is false
  • (d) A is false, but R is true

Q1. (Practice) Assertion (A): In a binomial distribution the variance is always less than the mean. Reason (R): The variance npq equals the mean np multiplied by q, which is less than 1.

Answer and explanation

(a). For 0 < p < 1, q < 1, so npq < np. R explains A directly.

Q2. (Practice) Assertion (A): Two mutually exclusive events with positive probabilities are independent. Reason (R): For mutually exclusive events, \( P(A \cap B) = 0 \).

Answer and explanation

(d). A is false: independence needs \( P(A \cap B) = P(A)P(B) > 0 \), but for mutually exclusive events it is 0, so they are dependent. R is true.

Q3. (Practice) Match List I with List II.

List I (Approach)List II (Feature)
A. ClassicalI. Degree of personal belief
B. Relative frequencyII. Set of axioms; Kolmogorov
C. SubjectiveIII. Equally likely outcomes; a priori
D. AxiomaticIV. Long-run proportion of repeated trials
  • (1) A-III, B-IV, C-I, D-II
  • (2) A-IV, B-III, C-I, D-II
  • (3) A-III, B-I, C-IV, D-II
  • (4) A-II, B-IV, C-I, D-III
Answer and explanation

(1) A-III, B-IV, C-I, D-II.

Q4. (Practice) Match the distribution with its property.

List IList II
A. BinomialI. Mean = variance
B. PoissonII. Continuous, mesokurtic
C. NormalIII. Variance less than mean
  • (1) A-III, B-I, C-II
  • (2) A-I, B-III, C-II
  • (3) A-II, B-I, C-III
  • (4) A-III, B-II, C-I
Answer and explanation

(1) A-III, B-I, C-II.

Q5. (Practice) Arrange the following in increasing order of the area of a normal curve they cover:

  • A. \( \mu \pm 1.96\sigma \)

  • B. \( \mu \pm 1\sigma \)

  • C. \( \mu \pm 3\sigma \)

  • D. \( \mu \pm 2.58\sigma \)

  • (1) B, A, D, C

  • (2) B, D, A, C

  • (3) A, B, D, C

  • (4) B, A, C, D

Answer and explanation

(1) B, A, D, C. About 68.27%, 95%, 99% and 99.73%.

Q6. (Practice) Arrange the steps in applying Bayes’ theorem:

  • A. Divide each joint probability by the total

  • B. List the prior probabilities of the causes

  • C. Multiply each prior by the likelihood of the observed event

  • D. Add the joint probabilities to get the total probability of the event

  • (1) B, C, D, A

  • (2) C, B, D, A

  • (3) B, D, C, A

  • (4) B, C, A, D

Answer and explanation

(1) B, C, D, A. Prior → joint → total → posterior.


Quick Revision Table

ItemFormula or fact
ClassicalFavourable / total equally likely; Laplace; a priori
Relative frequencyLong-run proportion; von Mises; a posteriori
SubjectivePersonal belief
AxiomaticKolmogorov, 1933: non-negativity, P(S) = 1, additivity
Addition\( P(A \cup B) = P(A) + P(B) - P(A \cap B) \)
Multiplication\( P(A \cap B) = P(A)P(B \mid A) \); independent: \( P(A)P(B) \)
Bayes (1763)Posterior = prior × likelihood / total
Expectation\( E(X) = \sum xp(x) \); \( V(aX + b) = a^2V(X) \)
Binomial\( \binom{n}{r}p^rq^{n-r} \); mean np; variance npq
Poisson\( e^{-\lambda}\lambda^r/r! \); mean = variance = λ; rare events
NormalSymmetric, mean = median = mode, \( \beta_2 = 3 \), inflection at \( \mu \pm \sigma \)
Areas±1σ 68.27%; ±2σ 95.45%; ±3σ 99.73%; ±1.96σ 95%; ±2.58σ 99%
Z\( (X - \mu)/\sigma \)

Key Takeaways

What’s Next