Probability Distributions: Binomial, Poisson,
Normal
1. First Principles: What is a “Distribution”?
Before I jump into the specifics, I must not assume you know what a
“distribution” is.
A distribution in probability/statistics is a description of how likely
different outcomes are.
Imagine rolling a die. The distribution says: “Each of the numbers 1, 2,
3, 4, 5, 6 has probability 1/6.”
That’s all it is: a map from “possible outcomes” → “probabilities.”
Sometimes outcomes are discrete (countable: 0, 1, 2, 3 people).
Sometimes they are continuous (uncountable: height can be 170.2
cm, 170.25 cm, etc.).
Okay. With that bedrock set, now let’s look at each distribution in turn.
2. The Binomial Distribution
a) The Context
Discrete outcomes only. I cannot have 2.35 people succeeding; I
can only have 2 or 3.
We are counting successes in a fixed number of independent trials.
b) Defining Conditions
1. There are exactly n trials. (e.g., I flip a coin 10 times: n = 10).
2. Each trial can only have two outcomes: “success” or “failure.” (Heads
or Tails, Defective or Non-defective, Male or Female).
3. The probability of success is p.
o Then automatically, the probability of failure is q = 1 − p.
o Notice: p + q = 1. This is a postulate of probability theory:
the sum of probabilities of all mutually exclusive exhaustive
outcomes must equal 1.
4. All trials are independent: the outcome of one trial does not affect
another.
c) The Formula
The probability of exactly k successes in n trials is:
k()
P ( X=k )= n p q
k n− k
where:
( nk)= k ! ( n− k ) ! is the number of ways to choose k successes out of n
n!
trials.
p = probability of getting success k times.
k
q
n −k
= probability of getting failure the remaining n−k times.
d) Mean and Variance
Mean: μ=n p .
Variance: σ 2=n p q.
That means on average, if I flip 100 coins (n=100) with p=0.5, I expect 50
heads, and the variability (spread) around 50 is about √25 = 5.
e) Normal Approximation
If n is large and p is not extreme (not too close to 0 or 1), then the binomial
shape looks “bell-shaped.” In fact:
X ∼ Bin ( n , p ) can be approximated by N ( μ=n p , σ 2 =n p q ) .
3. The Poisson Distribution
a) The Context
The Poisson is like a limiting case of the Binomial.
It applies when events are rare but possible over a large number of
opportunities.
Think: number of phone calls arriving at a call center in one minute,
number of printing errors in a book, meteorites hitting Earth in a year.
b) Defining Conditions
1. The number of trials n is very large (say n > 50).
2. The probability of success p in a single trial is very small.
3. The expected number of successes n p= λ stays finite (and usually < 5
in applications).
c) The Formula
The probability of observing k events is:
−λ k
e λ
P ( X=k )=
k!
where:
λ=n p is the expected (mean) number of events.
e ≈2.71828 is the base of natural logarithms.
d) Mean and Variance
Mean = λ .
Variance = λ . That is a unique property: variance equals mean.
e) Relation to Binomial
You can think of Poisson as:
−λ k
lim
n → ∞ , p → 0 ,n p= λ ()
n p k ( 1 − p )n − k = e λ .
k k!
This is why people say Poisson is Binomial’s child for rare events.
4. 🟢 The Normal Distribution
a) The Context
The Normal is the most famous distribution of all.
It applies to variables with a continuous range of values.
Examples: human height, exam scores, measurement errors, weights,
times.
It’s called “normal” because historically many natural processes were
observed to follow it.
b) Defining Formula
The probability density function (pdf) is:
2
( x− μ )
−
1 2σ
2
f ( x )= e
σ √2 π
where:
μ = mean (the center).
σ = standard deviation (the spread).
π ≈ 3.14159.
c) Properties
The curve is perfectly symmetric around the mean.
About 68% of values lie within 1σ, 95% within 2σ, 99.7% within 3σ (the
empirical rule).
It is continuous: the probability of an exact value is 0; we only talk
about ranges.
d) Relation to Binomial and Poisson
If n is large and p not extreme, Binomial → Normal.
If λ is large, Poisson → Normal.
So the Normal is the “big umbrella” many other distributions approximate.
5. Linking Them All Together
Binomial: Fixed number of independent yes/no trials. Discrete.
Poisson: Special case of Binomial when n large, p small, np finite.
Rare events. Discrete.
Normal: Continuous bell-shaped curve. Approximates Binomial (when
n large) and Poisson (when λ large).
So I can see them as part of a family tree:
Binomial ──> Poisson (rare events)
│
└──> Normal (large n, balanced p)
Poisson ──> Normal (large λ)
6. Everyday Analogies
Binomial: Like flipping a fixed number of coins and counting heads.
Poisson: Like counting the number of people who knock on your door
in one hour — could be 0,1,2,… and arrivals are rare but possible.
Normal: Like measuring heights in a classroom; most people cluster
around the mean, few are very tall or very short.