Plain-language meaning
A confidence interval gives a range around an estimate. It combines the estimate, sample size, variation and chosen confidence level into a lower and upper bound.
Example
If a sample mean is 100 and the margin of error is 5, a simple interval is 95 to 105. For proportions, methods such as Wilson intervals are often preferred over the simplest normal approximation when sample sizes are modest.
How to interpret it
A 95% confidence interval does not mean there is a 95% probability that this exact interval contains the true value. It means that, over repeated samples using the same method, about 95% of the intervals would contain the true value.
Limitations
A confidence interval does not fix biased sampling, poor measurement, missing data or a weak study design. It also depends on the method chosen for a mean, proportion or other statistic.
FAQ
What makes a confidence interval narrower?
Larger sample size, lower variability and a lower confidence level usually make the interval narrower.
Is a confidence interval the same as a prediction interval?
No. A confidence interval estimates uncertainty around a parameter. A prediction interval estimates where a future observation may fall.