You are on page 1of 1

Normal approximation of the Poisson distribution

If X is a random variable which has a Poisson distribution with mean λ, and λ is large (λ > 15),
then the distribution of X is approximately normal with mean λ and variance λ.
approx
X ∼ Po(λ) where λ > 15 =⇒ X ∼ N(λ, λ)

This normal approximation can be used to approximate probabilities which are too tedious to
calculate using the Poisson distribution.

Remark: The Poisson distribution is a discrete distribution taking only non-negative integer
values, while the normal distribution is a continuous distribution taking all values. When using
the normal approximation to calculate probabilities, a continuity correction is used:

Poisson Normal approximation


P(X ≤ x) P(X ≤ x + 0.5)
P(X < x) P(X < x − 0.5)
P(X ≥ x) P(X ≥ x − 0.5)
P(X > x) P(X > x + 0.5)

For example, let X ∼ Po(20).

To calculate P(X ≤ 25), instead of using the the Poisson distribution which is tedious:

P(X ≤ 25) = P(X = 0) + P(X = 1) + P(X = 2) + · · · + P(X = 25)


202 2025
= e−20 (1 + 20 + 2!
+ ··· + 25!
)
= 0.888,

it can be approximated using the normal distribution N(20, 20).

For the continuity correction, 0.5 is added to 25 since any number ≤ 25.5, when rounded to
the nearest integer is ≤ 25.
approx
P(X ≤ 25) = P(X ≤ 25.5)
25.5−20
= P(Z ≤ √
20
)
= P(Z ≤ 1.230)
= Φ(1.230)
= 0.891

Although the probability obtained is not the same as the actual value (0.888 correct to 3
significant figures), it is very close.

You might also like