Mean Squared Error
Mean Squared Error
Mean squared error measures an estimator's average squared distance from the true parameter: MSE(theta-hat) = E[(theta-hat minus theta)2]. It is always nonnegative and combines random spread with systematic offset: MSE = Var(theta-hat) + Bias(theta-hat)2.
Bias is signed before squaring, so a negative and positive bias of the same magnitude make equal contributions to MSE. Comparing estimators requires both components on the same parameter scale. A lower variance does not automatically win if it is achieved by a bias large enough that its square outweighs the variance reduction.
Estimator A has variance 5 and bias 0. Estimator B has variance 1 and bias 3. Which estimator has lower MSE?
A parameter is measured in milligrams. Which statement about its estimator's MSE is correct?