Weak Law of Large Numbers, biased expectation?

231 Views Asked by At

I want to show that:

$$\hat{\sigma^2}=(1/n)\sum^{n}_{i=1} ( X_i-\bar{X} )^2$$

is a consistent estimator of $\sigma^2$.

I was using the Weak Law of Large Numbers in the sense that: $$E(X_i-\bar{X })^2=\sigma^2[1-1/n]$$, and I also showed that $$\operatorname{Var}(X_i-\bar{X })^2< \infty$$

Given those two, I said that the average of $(X_i-\bar{X })^2$ was $\hat{\sigma^2}$, and that the estimator converges in probability to the sigma square.

The problem is that the expectation shows that my estimator is biased (the conditions state that $EX_i$ should be unbiased and equal to $\mu$).

I know the traditional answer to show this is consistent is to plug plus and minus $\mu$ inside my estimator and show that $\bar{X}\to\mathrm{Prob}( \mu_x)$, and also show that it converges in probability to sigma square, then everything is fine.

I just want to know if the first solution is valid or not, or if the expectation of what I consider"Xi" has to be unbiased for $\mu$. Thanks in advance.