Confidence interval

A confidence interval, usually reported as a 95% CI, is the range around a study's estimate that the data are compatible with. The estimate is the single best guess; the interval says how precise that guess is.

Two readings matter most in practice. When the interval for a difference between two treatments crosses zero, the study did not find a difference it can stand behind, which is not the same as proving there is none. When the interval is wide, usually because the study was small, it cannot rule out a real difference in either direction. Ryan's research articles report the interval alongside the headline number for this reason: a result can sound decisive and still be compatible with no effect at all.

Width is not always a weakness. In the study behind the crown-versus-filling article, root-canal-treated teeth left uncrowned were lost at 6.0 times the rate of crowned teeth, with a 95% CI of 3.2 to 11.3: wide, but even its low end is more than triple the loss rate, so the finding holds.