Interval estimation uses sample data to construct an interval intended to contain an unknown population parameter. A confidence interval is a random interval whose repeated-sampling coverage probability is at least a specified confidence level under the assumed statistical model.
If
and
are statistics computed from the sample
, a two-sided
confidence interval for a parameter
satisfies
for every parameter value in its stated range. After the data have been observed, the endpoints are fixed; the coverage statement describes the procedure rather than assigning a frequentist probability to the fixed parameter.