Scribd
Upload
4 views7 pages

Lecture 6

This lecture discusses the approximations of binomial distributions using normal and Poisson distributions. It highlights the need for these approximations due to the complexity of calculating cumulative binomial probabilities, especially for large n. The normal approximation is suitable when σ = npq ≥ 3, while the Poisson approximation is applicable when n is large and either p or q is small, with npq < 3.

Uploaded by

cakesmasher117
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views7 pages

Lecture 6

This lecture discusses the approximations of binomial distributions using normal and Poisson distributions. It highlights the need for these approximations due to the complexity of calculating cumulative binomial probabilities, especially for large n. The normal approximation is suitable when σ = npq ≥ 3, while the Poisson approximation is applicable when n is large and either p or q is small, with npq < 3.

Uploaded by

cakesmasher117
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Lecture 6: Binomial Approximated by Normal or

Poisson

STOR 435, SS1, 2025

5/19/2025

435-SS1-2025 approximating binomial


normal approximation to binomial

Histogram vs Density Curve

435-SS1-2025 approximating binomial


normal approximation to binomial continued

Why need it? The binomial probability formula need not be


convenient for calculating cumulative binomial probabilities.
For example, let X ∼ Bin(100, 1/3) and calculate
48  
X 100
P(X ≤ 48) = (1/3)k (2/3)100−k = · · ·
k
k=0

A certain numerical approximation scheme is needed.


We are to introduce two kinds of approximations: normal
approximations (Pitman: Sections 2.2, 2.3) and Poisson
approximations (Pitman: Section 2.4).

435-SS1-2025 approximating binomial


normal approximation to binomial continued

Let X ∼ Bin(n, p), Z ∼ N(0, 1). For a ≤ b and large n,


 
a − µ−0.5 b − µ+0.5
P(a ≤ X ≤ b) ≈ P ≤Z≤
σ σ
   
b − µ+0.5 a − µ−0.5
= Φ −Φ ,
σ σ
√
where µ = np, σ = npq, and the terms ±0.5 are referred to as
the correction for continuity which makes the approximation
more accurate.
A rule of thumb: n and p play a collective role. The normal
√
approximation scheme works well if σ = npq ≥ 3. The larger
the σ, the better.

435-SS1-2025 approximating binomial


Example 1
Let X ∼ N(µ, σ 2 ) represent the lifetime of light bulbs with
µ = 1000 and σ = 200 (in hours). Sample 100 light bulbs with
replacement. Let Y be the total number of light bulbs in the
sample with lifetimes exceeding 1050 hours. Find P(Y ≥ 45).
Note: Y ∼ Bin(n, p) with n = 100 and
 
1050 − 1000
p = P(X > 1050) = P Z > = P(Z > 0.25) = 0.4013.
200

Hence
   
100 − np + 0.5 45 − np − 0.5
P(Y ≥ 45) ≈ Φ √ −Φ √
npq npq
= 1 − Φ(0.89) = 0.1867.

Note: Normal distributions appear twice in this problem for


different reasons.

435-SS1-2025 approximating binomial


normal approximation need not work

435-SS1-2025 approximating binomial


Poisson approximation

Conditions: Let X ∼ Bin(n, p) where n is large, and either p


√
or q is small, such that npq < 3.
Poisson approximation to binomial:

µk
P(X = k) ≈ e−µ , k = 0, 1, 2, ...
k!
k
{e−µ µk! , k = 0, 1, 2, ...} form a Poisson probability
distribution with parameter µ > 0.
Note: Poisson approximations apply to single-point
binomial probabilities, whereas normal approximations
apply to cumulative binomial probabilities.

435-SS1-2025 approximating binomial

You might also like