Sample Size (n)

The number of observations a result was computed on: the foundation of its reliability.

Sample size (n) is the number of cases or measurements used to compute a statistic, e.g. a correlation or a beta. The larger the n, the less random and more reliable the result. With a small n, even a seemingly strong relationship can be down to chance. It's a key quality filter for any number based on historical data: a result from n = 500 observations is far more trustworthy than the same result from n = 12. In practice, always read a statistic together with its sample. A high correlation on a small sample is a weak hint, not strong evidence.

In plain words

Tells you how much data is behind a number. Little data = the result could be a fluke. Lots of data = you can trust it more.

Example

A correlation of 0.8 at n = 15 is unreliable, easily a coincidence. The same correlation at n = 500 is a solid, repeatable relationship.

Also known as: próbka, n, liczba obserwacji, sample size

Back to the glossary