Unbiased estimator
Also known as: Unbiasedness
An unbiased estimator satisfies $E_\theta[W]=\theta$ for every $\theta$ in the parameter space. Then MSE equals variance.
Examples: $\bar X$ for a mean; $S^2=(n-1)^{-1}\sum(X_i-\bar X)^2$ for Normal variance. Unbiasedness is a finite-sample property — consistency is the large-sample analogue. Among unbiased linear estimators of a regression slope, Gauss–Markov picks the minimum-variance one (BLUE).