Definition
Plain language
A way of writing a chance so that being extra sure keeps counting, instead of getting stuck near 100%.
As stated in the literature
The logarithm of p/(1-p); an unbounded scale on which additive shifts correspond to multiplicative changes in odds, so effects remain measurable in saturated regimes where probability differences vanish.
Also called: log odds, log-odds scale
Why it matters: It keeps effects visible when a system is already near-certain, which is where ordinary percentage differences stop telling you anything.
For example, moving from 99 percent sure to 99.9 percent sure barely registers as a change in probability, but on the log-odds scale it is the same size of step as moving from 50 to 90 percent.