Mathematics
68-95-99.7 Rule
Quick fact
In a normal distribution, roughly 1 in 370 observations fall beyond three standard deviations from the mean, making such extreme values rare indeed.
Why this is interesting
Ever wonder how test scores, heights, or IQ scores are distributed? Why is it that most people score near the average, but a few are far above or below?
Read the full explanation
Understanding 68-95-99.7 Rule
Imagine a bell-shaped curve—the normal distribution. Its peak is the average (mean). The spread is measured by standard deviation. The 68-95-99.7 rule says: about 68% of all data points lie within one standard deviation of the mean (both sides). For example, if adult male heights average 175 cm with a standard deviation of 7 cm, then 68% of men are between 168 cm and 182 cm. Going out two standard deviations covers about 95% of people (161–189 cm). Three standard deviations include 99.7% (154–196 cm). Almost everyone is within that range.
A deeper explanation
This rule works because the normal distribution is a probability density function with known mathematical properties. The area under the curve between two points corresponds to the proportion of data. Standard deviation measures natural variability: the wider the curve, the more scattered the data. Z-scores tell how many standard deviations a value is from the mean. The rule stems from integrating the Gaussian function: for a standard normal, the integral from -1 to 1 is ~0.6827, -2 to 2 is ~0.9545, -3 to 3 is ~0.9973. Knowing this, you can quickly assess probabilities and detect outliers without complex calculations. In practice, the rule is a shortcut for understanding data spread, quality control (e.g., Six Sigma), and grading curves.