Evidence-appraisal glossary
Shrinkage
A modeling technique that deliberately pulls extreme estimates toward a central value or overall average, trading a little bias for more stable, reliable predictions. It counters the tendency of small or noisy groups to produce exaggerated results.
Also called: shrinkage estimation, partial pooling.
Estimates from small samples or many parallel groups swing widely by chance, and the most extreme ones are usually flukes that will not repeat, the same force behind regression to the mean. Shrinkage methods, including penalized regression and hierarchical models, counter this by discounting extreme values toward a shared center, giving predictions that generalize better to new data. In league tables of hospitals or surgeons, for instance, shrinkage keeps a tiny unit's freak result from topping the chart. The cost is a modest, deliberate bias, accepted because it lowers overall error.
This is a plain-language methodology definition for reading research. It is general education, not medical advice.