The Standard Normal Distribution

What Makes a Normal Distribution Standard?

A normal model can describe measurements whose values cluster around a center, with fewer observations farther from that center. But normal models may have different centers and spreads, so a probability statement is not fully described by saying only that a variable is normal. The standard normal distribution is the particular normal model with center zero and standard deviation one. It gives a common reference curve for understanding normal probabilities.

The standard normal variable is usually denoted by \(Z\). The notation \(Z\sim N(0,1)\) means that \(Z\) has a normal distribution with mean \(0\) and variance \(1\). Since standard deviation is the square root of variance, its standard deviation is also \(1\). The values of \(Z\) are dimensionless: they are measured relative to this model’s own scale, not in units such as seconds or centimeters.

The density curve is bell-shaped and symmetric about \(z=0\). Its highest point is at zero, and it tapers toward the horizontal axis in both directions. The entire area under the curve is \(1\), because it represents the total probability of all possible values of \(Z\). As with other continuous models, probabilities correspond to areas under the curve, not to the height of the curve at a single point.

Reading the Density Curve

The function \(\phi\) gives the height of the density curve at each horizontal coordinate. For example, \(\phi(0)\) is the height directly above zero. That height is not the probability that \(Z=0\). For a continuous random variable, the probability of any one exact value is zero; probabilities are areas over intervals of values.

The standard normal density is positive at every finite \(z\), so the curve stays above the horizontal axis. It is symmetric because the formula depends on \(z^2\): replacing \(z\) by \(-z\) leaves the density unchanged. Thus, points equally far from zero have equal curve heights, such as \(\phi(-1)=\phi(1)\). Equal heights alone are not probabilities; area over a range of values is needed to obtain probability.

Cumulative Probabilities and \(\Phi\)

A useful way to record probabilities from the standard normal model is to accumulate area from the far left of the curve up to a chosen coordinate. The resulting function is called the standard normal cumulative distribution function, written with a capital Greek letter phi, \(\Phi\). Lowercase \(\phi\) refers to the density; uppercase \(\Phi\) refers to cumulative probability.

The variable \(t\) inside the integral is a placeholder for values along the horizontal axis; \(z\) is the upper limit that identifies the cumulative cutoff. The function \(\Phi\) always takes values between \(0\) and \(1\). As \(z\) moves from far left to far right, the accumulated area increases from values near \(0\) toward \(1\).

The probability notation \(P(Z\le z)\) and the area notation \(\int_{-\infty}^{z}\phi(t)\,dt\) describe the same quantity in two ways. The probability notation emphasizes the event about the random variable; the integral emphasizes the area under the density curve. Because the model is continuous, including the endpoint does not change the probability: \(P(Z\le z)=P(Z<z)\).

How the Density and Cumulative Function Fit Together

The density curve and the cumulative distribution function show different views of the same model. The density \(\phi(z)\) gives the curve’s height at a location. The cumulative function \(\Phi(z)\) gives the area accumulated up to that location. Where the density is high, adding a small amount to the cutoff adds area relatively quickly; where the density is low, the cumulative area changes more slowly.

This connection can be stated precisely. If \(a<b\), the increase in cumulative probability between \(a\) and \(b\) is the area under the density curve between those coordinates. Since \(\phi(t)\) is positive, this increase is positive. This fact explains why the cumulative function never decreases.

The result follows directly by subtracting the two cumulative integrals. The area accumulated up to \(b\) includes all the area accumulated up to \(a\), plus the area from \(a\) to \(b\). Because the density is positive throughout that interval, the extra area is positive. By the Fundamental Theorem of Calculus, differentiating the integral with upper limit \(z\) gives \(\Phi'(z)=\phi(z)\).

Symmetry also provides a useful check on cumulative values. The standard normal curve places equal areas on either side of zero. In particular, the cumulative area to the left of \(-1\) equals the area to the right of \(1\). Since \(\Phi(1)\approx0.8413\), the remaining area on the right is about \(1-0.8413=0.1587\), and therefore \(\Phi(-1)\approx0.1587\). This is a check based on the symmetry of the model, not a different definition of \(\Phi\).

Why This Representation Matters

The standard normal density and its cumulative function provide a shared language for describing normal-model probabilities. The curve makes the shape and symmetry visible, while \(\Phi\) assigns a numerical cumulative probability to any cutoff. A cumulative value is useful precisely because it connects a location on the horizontal axis to the amount of probability accumulated to its left.

In applications, a model may be stated in the original units of a measurement, while the standard normal model uses its own scale. The standard normal curve is therefore a foundational reference, not an assumption that every measurement is automatically standard normal. Whether a variable follows a normal model must be justified by the situation or by an appropriate analysis.

Common Mistakes

• Calling a density height a probability. A value such as \(\phi(1)\approx0.2420\) is the curve’s height at \(1\), not the probability that \(Z=1\). Probability is represented by area.
• Mixing up \(\phi\) and \(\Phi\). Lowercase \(\phi(z)\) is the density height. Uppercase \(\Phi(z)\) is the cumulative probability up to \(z\).
• Forgetting what the cutoff means. \(\Phi(z)\) describes area to the left of \(z\), not area to the right and not the area at \(z\).
• Assuming every normal variable is standard normal. Standard normal means specifically mean zero and standard deviation one. A different normal model has different parameters.