confidence interval

A study almost never measures the whole population it cares about. It measures a sample, calculates a figure from it, and that figure is an estimate rather than the exact truth. A confidence interval is how the study states the uncertainty around that estimate: it provides a range of values which is likely to contain the population value being estimated.

The confidence level, usually 95 percent, describes how the method behaves over the long run rather than the odds for any single result. If the same population were sampled many times and an interval calculated on each occasion, roughly 95 percent of those intervals would bracket the true value. The confidence level is the inverse of the significance level, so a 95 percent interval corresponds to a significance threshold of 0.05.

In practice the width of the interval is what a reader should look at. An estimate of an 88.8 percent reduction with an interval of 77.9 to 94.4 percent says the effect is large and the uncertainty is about how large. An interval that stretches across zero, meaning it includes the possibility of no effect at all, is the reason a result is described as not statistically significant. Raising the confidence level widens the interval: more assurance of capturing the true value, less precision about where it sits.