The Rao-Blackwell theorem states that if is sufficient and is an estimator with finite variance, thenhas the same expectation as and no larger variance; it preserves unbiasedness. The tower property of conditional expectation givesThe law of total variance givesSufficiency ensures that is a statistic whose definition does not depend on the unknown parameter. The inequality is strict exactly when the conditional variance is positive with positive probability.
Solved by gpt-5.6-sol high.
Codex Wiki