A confidence interval gives a range around a point estimate within which the true value plausibly falls, conventionally at ninety-five percent confidence. A narrow interval indicates precision while a wide interval indicates uncertainty.
An interval including one indicates no statistically significant association.
Alternative Names:
CI, 95% Confidence Interval
Why it Matters?
Whether the interval crosses one is the practical test in causation litigation, since an interval spanning one includes the possibility of no association at all. An expert citing a relative risk of two point one with a confidence interval from zero point eight to five point four is reporting an estimate consistent with no effect. Reading the interval rather than the point estimate is the necessary discipline in evaluating epidemiological testimony.
Frequently Confused with
Related terms
Frequently asked questions
What does an interval including one indicate?
Why read the interval rather than the estimate?
