Probability Distributions - Study Notes
Chapter Summary
Frequency distributions are of two types. Observed frequency distributions are built from actual data or experiment; theoretical (probability) distributions are built from expectations based on past experience. This unit studies two discrete theoretical distributions, the binomial and the Poisson, and one continuous distribution, the normal distribution.
Binomial Distribution
- Discovered by James Bernoulli (1654-1705) in 1700 and published in 1713, eight years after his death.
- Bernoulli trial: a random experiment with only two outcomes, success S and failure F, with probabilities p and q. Examples: tossing a coin (head or tail), throwing a die (even or odd).
- Definition: X follows a binomial distribution with parameters n and p if \( P(X = x) = {}^nC_x\, p^x q^{n-x},\; x = 0, 1, 2, \ldots, n \), where \( q = 1 - p \). We write \( X \sim B(n, p) \) and call X a binomial variate.
- The probabilities are the successive terms of the binomial expansion \( (q + p)^n \).
Conditions
- The number of trials n is finite.
- The trials are independent of each other.
- The probability of success p is constant for each trial.
- Every trial has only two possible outcomes, success or failure.
Mean, variance and properties
- Mean \( E(X) = np \); Variance \( = npq \); Standard deviation \( = \sqrt{npq} \).
- Symmetrical if \( p = q = 0.5 \); skew symmetric if \( p \ne q \).
- Positively skewed if \( p < 0.5 \); negatively skewed if \( p > 0.5 \).
- Variance is less than mean: \( npq = (np)q < np \).
- Check: if given data lead to \( q > 1 \), the binomial is impossible. Example: mean 12 and standard deviation 4 give \( q = 16/12 = 4/3 \), so such a distribution cannot exist.
Poisson Distribution
- Derived in 1837 by the French mathematician Simeon D. Poisson.
- It approximates the binomial when n is large and p is close to zero, and describes events that occur rarely.
- Examples: number of bacteria in one cubic centimetre; printing mistakes per page of a text book; alpha particles emitted by a radioactive substance in a fraction of a second; road accidents in a particular interval of time per day; lightnings per second.
Conditions (limiting case of binomial)
- n, the number of trials, is indefinitely large (\( n \to \infty \)).
- p, the constant probability of success in each trial, is very small (\( p \to 0 \)).
- \( np = \lambda \) is finite, so \( p = \lambda/n \) and \( q = 1 - \lambda/n \).
Definition and properties
- \( P(X = x) = \dfrac{e^{-\lambda}\lambda^x}{x!},\; x = 0, 1, 2, \ldots;\; \lambda > 0 \).
- Mean \( = \lambda \) and variance \( = \lambda \). The Poisson is the only distribution in which mean and variance are equal.
- In examinations the needed value of \( e^{-\lambda} \) must be given.
- Successive terms: \( P(X = x + 1) = \dfrac{\lambda}{x + 1} P(X = x) \). Example: \( P(0) = 0.2725 \) gives \( \lambda = 1.3 \) and \( P(1) = 0.2725 \times 1.3 = 0.3543 \).
- Expected frequency of x successes in N sets \( = N \times P(X = x) \).
Normal Distribution
- A continuous distribution for quantities such as weight, height and salary. First introduced by De Moivre in 1733; Laplace and Gauss were also associated with its development.
- It is a limiting case of the binomial when n is infinitely large and neither p nor q is very small.
- Density: \( f(x) = \dfrac{1}{\sigma\sqrt{2\pi}} e^{-\frac{(x-\mu)^2}{2\sigma^2}},\; -\infty < x < \infty \). We write \( X \sim N(\mu, \sigma^2) \).
Properties of the normal curve
| Property | Statement |
|---|---|
| Shape | Bell shaped and symmetrical about \( x = \mu \) |
| Averages | Mean, median and mode coincide |
| Asymptote | The x-axis; the tails never touch it |
| Position | No part of the curve lies below the x-axis |
| Points of inflexion | \( x = \mu \pm \sigma \) |
| Peak | Single peak (unimodal); maximum ordinate \( \dfrac{1}{\sigma\sqrt{2\pi}} \) at \( x = \mu \) |
| Total area | Equal to 1 |
Area property
- \( P(\mu - \sigma < X < \mu + \sigma) = 0.6826 \) (about 68.27%)
- \( P(\mu - 2\sigma < X < \mu + 2\sigma) = 0.9544 \) (about 95.5%)
- \( P(\mu - 3\sigma < X < \mu + 3\sigma) = 0.9973 \) (about 99.7%)
Standard Normal Distribution
- \( Z = \dfrac{X - \mu}{\sigma} \) is the standard normal variate with mean 0 and standard deviation 1: \( Z \sim N(0, 1) \).
- Density: \( \varphi(z) = \dfrac{1}{\sqrt{2\pi}} e^{-z^2/2},\; -\infty < z < \infty \).
- The table gives the area between 0 and z. The area to the right (or left) of 0 is 0.5.
Reading the table
- \( P(Z > a) = 0.5 - P(0 < Z < a) \) for \( a > 0 \). Example: \( P(Z > 1.09) = 0.5 - 0.3621 = 0.1379 \).
- \( P(Z < -a) = 0.5 - P(0 < Z < a) \). Example: \( P(Z < -1.65) = 0.5 - 0.4505 = 0.0495 \).
- Limits on opposite sides: add. \( P(-1 < Z < 1.96) = 0.3413 + 0.4750 = 0.8163 \).
- Limits on the same side: subtract. \( P(1.25 < Z < 2.75) = 0.4970 - 0.3944 = 0.1026 \).
- If a given right-tail area exceeds 0.5, the z value is negative.
Worked example
X is normal with mean 30 and SD 5. For \( P(26 \le X \le 40) \): \( z = -0.8 \) and \( z = 2 \), so the probability is \( 0.2881 + 0.4772 = 0.7653 \). For \( P(X \ge 45) \): \( z = 3 \), so \( 0.5 - 0.49865 = 0.00135 \).
More for this chapter
Book Back Questions27 textbook MCQs · solved
Additional MCQs7 extra MCQs · solved
Practice TestInteractive · instant score
Book Back TestTest yourself on the textbook set
Additional MCQ TestTest yourself on the extra set
Formula SheetAll key formulas