What Is a Normal Distribution?
What does a normal distribution actually look like?
Picture a bell. The middle is tall and crowded, the tails on each side flatten out. That is the whole idea. Heights, measurement errors, and a lot of test scores pile up this way: average is common, extremes are rare.
Two numbers describe the entire curve. The mean sets where the peak sits. The standard deviation sets how wide the bell spreads. Small standard deviation, tight and tall. Large one, wide and flat.
What is the normal distribution formula?
The formula looks scary, but it is doing one job: giving the height of the curve at any value x.
f(x) = (1 / (sigma times the square root of 2 pi)) times e to the power of negative (x minus mu) squared divided by 2 sigma squared.
In plain words, mu is the mean and sigma is the standard deviation. The e term makes the curve fall off smoothly as you move away from the mean. The fraction out front just scales it so the total area under the curve equals 1, which is what makes it a probability distribution. You rarely compute it by hand. The site's free calculator does the plugging and gives you the area, which is the probability you usually want.
What is the 68 95 99.7 rule?
This is the part worth memorizing.
| Range | Share of data |
|---|---|
| Within 1 standard deviation | about 68% |
| Within 2 standard deviations | about 95% |
| Within 3 standard deviations | about 99.7% |
So a value three standard deviations out is genuinely unusual. That is where the phrase outlier earns its keep.
Worked example: a z score
Say a test has a mean of 100 and a standard deviation of 15. You score 130. The z score is (130 minus 100) divided by 15, which is 2. You are two standard deviations above average, higher than roughly 97.7% of test takers. That single number, the z score, is how the normal distribution turns raw scores into comparable positions.
One surprising detail
The bell curve shows up in the strangest places, including the distribution of pinky finger lengths and the spread of errors in astronomy. It also is not universal. Income and city sizes follow skewed curves, so anyone claiming everything is normally distributed is overselling it.
Do the math: open the normal distribution