Confidence Interval

A confidence interval (CI) provides a range around a sample statistic — such as a selection rate or impact ratio — that quantifies the uncertainty inherent in estimating a population value from a sample. A 95% confidence interval means that if the sampling process were repeated many times, approximately 95% of the resulting intervals would contain the true population value. In AI bias auditing, confidence intervals are critical for interpreting impact ratios: an impact ratio of 0.75 with a 95% CI of [0.60, 0.90] tells a different story than 0.75 with a CI of [0.73, 0.77]. The first might be consistent with no adverse impact given sampling variability, while the second almost certainly reflects real bias. Wider confidence intervals result from smaller sample sizes, which is why intersectional subgroup analysis often requires careful statistical interpretation. OnHirely reports confidence intervals alongside impact ratios so that practitioners can distinguish between genuine bias signals and noise from small samples.