Normal Distribution: Two Parameters, One Bell Curve

Why it exists

The problem is describing how probability spreads across every possible value of a real-valued random variable. The normal distribution solves this by offering a single continuous curve where only two parameters, the mean and the variance, are needed to define the relative likelihood of any outcome. This collapses an infinite set of possibilities into a compact formula governed by a center and a spread.

The mental model

Picture a symmetric bell. The peak sits exactly at the mean, which is the expectation of the distribution. The width of the bell is controlled by the standard deviation, the positive square root of the variance. A narrow bell means values stay close to the center; a wide bell means the variable can land far from the mean while still being probable. Once you set those two knobs, the entire shape is locked in.

How it works

The distribution is defined by its probability density function. You supply two fixed parameters: the mean and the variance. The mean shifts the curve left or right along the real number line. The variance scales the curve; larger variance flattens and widens it, while smaller variance tightens it around the mean. The standard deviation is the positive square root of the variance, giving a direct measure of spread in the same units as the random variable. Any specific value receives a density from this curve, with the highest density occurring exactly at the mean.

When to use it

Use it when you are modeling a real-valued random variable that is continuous and whose outcomes concentrate symmetrically around a known average. It fits when you can quantify the spread as variance or standard deviation and you need a density that peaks at the expectation.

When not to use it

Avoid it if the variable is discrete rather than continuous, or if it is bounded, because the Gaussian density extends across the entire real line and can assign density to impossible values. It also fails when data is skewed or has multiple peaks, since the model assumes one symmetric center. Forcing normality onto such data misrepresents both the center and the tail probabilities.

One canonical example

Consider a normal deviate with mean zero and variance one. Its standard deviation is one, and the probability density function peaks at zero. If you shift the mean to one hundred and raise the variance to twenty-five, the standard deviation becomes five, the peak moves to one hundred, and the curve tightens so that most density sits between ninety-five and one hundred and five. This shows how the two parameters alone completely redraw the distribution.