Standard error
How much an estimate would bounce around across repeated samples.
What it means
The standard error is the standard deviation of a sample statistic's sampling distribution, quantifying how much an estimate such as a mean would vary from sample to sample. It shrinks as sample size grows, typically in proportion to one over the square root of n, which is why precision improves slowly with added data. The standard error is the engine behind confidence intervals and test statistics, converting variability into statements about uncertainty. It must not be confused with the standard deviation of the data itself: the latter describes spread among individuals, the former describes the precision of a summary computed from them.
Examples
A sample mean of 100 with a standard error of 2 implies the estimate would typically vary by about 2 points across repeated samples.
A poll's reported 'margin of error' of three points is the standard error dressed up. Going from 500 respondents to 2,000 only halves it — precision gets expensive fast.
An app team measures average session length from 100 users, then from 100 different users, and gets 4.1 and 4.6 minutes. That gap is sampling noise, not changed behaviour.
First described in Foundational inferential statistics.