Normal Distribution & Z-Scores

The Bell Curve Is Everywhere

If you measured the heights of every adult in a large city and plotted them on a chart, you would see a familiar shape: a smooth, symmetrical hill that peaks in the middle and tapers off on both sides. This shape is called the normal distribution, and it is arguably the most important concept in all of statistics.

The normal distribution shows up in a remarkable number of places. Test scores, blood pressure readings, the time it takes to commute to work, manufacturing tolerances on a factory floor, even the errors in scientific measurements -- all of these tend to follow a bell-shaped pattern. The reason is mathematical: whenever a measurement is influenced by many small, independent factors, the result tends to be normally distributed. This principle is closely related to the Central Limit Theorem.

In the chart above, the peak represents the most common value (the mean), and the curve falls off symmetrically on either side. Most values cluster near the center, with fewer and fewer observations appearing as you move toward the extremes.

Mean, Standard Deviation, and Shape

A normal distribution is completely defined by just two numbers: the mean (the center of the curve) and the standard deviation (how spread out the data is). The mean tells you where the peak sits on the number line. The standard deviation tells you how wide or narrow the bell is.

Consider IQ scores, which are designed to follow a normal distribution with a mean of 100 and a standard deviation of 15. Most people score between 85 and 115. A few score below 70 or above 130. Extremely few score below 55 or above 145. Change the standard deviation to 5, and the bell becomes much narrower -- almost everyone clusters between 90 and 110. Change it to 25, and the bell flattens out, with scores spread much more widely.

This is the beauty of the normal distribution: once you know the mean and standard deviation, you know the entire shape and can calculate the probability of any value occurring.

The 68-95-99.7 Rule

One of the most practical features of the normal distribution is the empirical rule, also called the 68-95-99.7 rule. It states that for any normally distributed data:

• About 68% of values fall within 1 standard deviation of the mean.
• About 95% of values fall within 2 standard deviations.
• About 99.7% of values fall within 3 standard deviations.

This rule gives you a quick way to gauge how unusual a value is. If your data is normally distributed and someone reports a value more than 3 standard deviations from the mean, that is extremely rare -- it happens less than 0.3% of the time. Quality control engineers use this idea every day: a factory part that falls outside three standard deviations from the target dimension is flagged as defective.

Z-Scores: A Universal Ruler

Different normal distributions use different units and scales. How do you compare a test score of 82 on an exam where the average is 75 (standard deviation 5) with a score of 720 on the SAT where the average is 500 (standard deviation 100)? You use a z-score.

A z-score tells you how many standard deviations a value is above or below the mean. The formula is straightforward: subtract the mean from the value, then divide by the standard deviation. For the exam score: (82 - 75) / 5 = 1.4. For the SAT: (720 - 500) / 100 = 2.2. The SAT score is more impressive relative to its distribution because it is farther above the mean in standard-deviation units.

A z-score of 0 means the value is exactly average. A positive z-score means it is above average. A negative z-score means it is below average. The magnitude tells you how far from average it is. A z-score of 2.0 means the value is higher than about 97.7% of all values in the distribution.

Z-scores are powerful because they convert any normal distribution into the standard normal distribution -- a bell curve with mean 0 and standard deviation 1. This lets you use a single reference table (or calculator) to find probabilities for any normally distributed variable, regardless of its original scale.

Real-World Applications

The normal distribution and z-scores are not just textbook ideas. Grading on a curve means fitting student scores to a normal distribution. Medical lab results are often flagged as abnormal when they fall beyond 2 standard deviations from the population average. Financial analysts model stock returns using normal distributions (though the tails are often fatter in reality, which is a critical limitation). Insurance companies use normal models to estimate claims.

It is also important to know when the normal distribution does not apply. Income distributions are heavily right-skewed -- a few very high earners pull the mean far above the median. Wait times and survival data are often skewed as well. Count data (like the number of accidents per day) follows other distributions entirely. Always check whether the bell-curve assumption is reasonable before applying these tools.