Variance of a Point Estimator
Sampling variability is a property of an estimator
An estimator is a random variable because a different sample would produce a different estimate. Its variance measures this repeated-sampling spread around its own expectation.
For independent observations with population variance sigma2, the sample mean has Var(X-bar)=sigma2/n, so its standard error is sigma/square-root-of-n. Increasing n by a factor of four halves this standard error, provided the independence and common-variance assumptions remain suitable.
For a sample proportion from independent Bernoulli trials with success probability p, Var(p-hat)=p(1−p)/n. Low variance means an estimator is precise in repeated sampling; it does not alone prove lack of bias.
For an affine transformation aT+b, Var(aT+b)=a2Var(T): adding a constant shifts all estimates without changing their spread.
Independent observations have population standard deviation 15 units. If n=25 observations are averaged, what is Var(X-bar)?
An independent Bernoulli sample has success probability p=0.30 and size n=200. What is Var(p-hat)?