1
$\begingroup$

Question: For observations $x_1$, $x_2$, . . . , $x_n$ with sample average $\bar{x}$, we can use an estimator for the population variance: $\hat\sigma^2$ = $\frac1n\cdot $ $\sum\limits_{i=1}^n (x_i - \bar{x})^2$. What is the bias of this estimator?

Supposedly the answer is -$\frac{\sigma^2}n$. The reason this confuses me too is because this question is a one minute question on a multiple choice paper. Is there a shortcut I'm missing or I'm supposed to see? $\text{Bias}\left(\hat\sigma^2\right)$ = $E\left(\hat\sigma^2\right)$ - $\left(\sigma^2\right)$ is the formula I tried to use.

$\endgroup$
3
  • $\begingroup$ The question is not clear. Please post the original exercise. What is the question? $\endgroup$ May 22, 2019 at 14:15
  • $\begingroup$ What have you been doing? Calculation of $\mathbb E\hat\sigma^2$ maybe? Actually I don't know what is meant by Bias($\hat\sigma^2$). $\endgroup$
    – drhab
    May 22, 2019 at 14:16
  • $\begingroup$ I have edited the original question, does this now make sense? @drhab $\endgroup$
    – jpatrickd
    May 22, 2019 at 14:23

2 Answers 2

1
$\begingroup$

You probably know that the expectation of the unbiased estimator is

$$E\left[\frac{1}{n-1}\sum\limits_{i=1}^n (X_i-\overline X )^2\right]=\sigma^2$$

To obtain the expectation of the biased estimator we just have to multiply both sides by $(n-1)$ and divide them by $n$

$$E\left[\frac{1}{n}\sum\limits_{i=1}^n (X_i-\overline X )^2\right]=\sigma^2\cdot \frac{n-1}{n}$$

Now we can calculate the difference: $\sigma^2\left(\frac{n-1}{n}-1\right)=\sigma^2\left(\frac{n-1}{n}-\frac{n}{n}\right)=-\frac{\sigma^2}{n}$

This is the fastest way I can think of.

$\endgroup$
1
$\begingroup$

Observe that $x_{i}-\overline{x}=y_{i}-\overline{y}$ where $y_{i}=x_{i}-\mathbb{E}x_{i}$.

Substituting this and working out $\hat{\sigma}^{2}$ we find: $$\hat{\sigma}^{2}=\frac{1}{n}\sum_{i=1}^{n}y_{i}^{2}-\overline{y}^{2}$$

Here $\mathbb{E}y_{i}^{2}=\sigma^{2}$ so that: $$\mathbb{E}\hat{\sigma}^{2}=\sigma^{2}-\mathbb{E}\overline{y}^{2}=\sigma^{2}-\frac{1}{n^{2}}\mathbb{E}\sum_{i=1}^{n}\sum_{j=1}^{n}y_{i}y_{j}=\sigma^{2}-\frac{1}{n}\sigma^{2}$$

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .