Evidence-appraisal glossary
Optimism
Optimism is the gap between how well a model appears to perform on the same data used to build it and how well it truly performs on new data. Models almost always look better on their home turf than they really are.
Also called: optimism-corrected performance.
When you develop and test a model on one dataset, it fits that data's quirks, so measures like the c-statistic come out inflated. Optimism is that inflation, and resampling methods such as bootstrapping estimate it so performance can be corrected downward. Ignoring optimism is a common reason published models disappoint when other groups try to use them.
This is a plain-language methodology definition for reading research. It is general education, not medical advice.