Let $X \sim U(0,\theta)$. Using moments for finding an estimator for the $\theta$ parameter, I obtained $\theta=2 \overline X $ which is unbiased.

I want to prove this estimator is consistent. I know that I shall verify:

$$\lim _{n\to \infty} E(\hat\theta_n) = \theta $$

$$\lim _{n\to \infty} V(\hat\theta_n) = 0$$

How can I prove this? Can be properly done without using Chebyshev inequality? I know that $V(\hat \theta)=0$ because it's unbiased, so I assume the limit involving the variance is zero. But how about the limit of the mean of $\ n$ estimators?


  • 1
    $\begingroup$ This doesn't bear directly on the question asked, but it may be worth noting that $2\overline X$ is not a really good estimator of $\theta.$ The best unbaised estimator is $\dfrac{n+1}n \max\{X_1,\ldots,X_n\}. \qquad$ $\endgroup$ – Michael Hardy Jul 3 '18 at 16:12

You've got the situation reversed. The fact that $\hat \theta_n$ is unbiased means that $\operatorname{E}[\hat \theta_n] = \theta$, but it does not say anything about the variance of the estimator.

The simplest thing to do is actually calculate the variance of $\bar X$, using the fact that the variance of the sum of independent random variables equals the sum of the variances of each random variable. That is to say, $$\operatorname{Var}[X_1 + \cdots + X_n] \overset{\text{ind}}{=} \operatorname{Var}[X_1] + \cdots + \operatorname{Var}[X_n].$$ Since each observation is also identically distributed, the RHS is simply $n \operatorname{Var}[X_1]$, or $n$ times the variance of a single observation. Compute this variance, show it is finite, and then show $$\operatorname{Var}[\hat \theta_n] = \frac{4}{n} \operatorname{Var}[X_1]$$ and then the limit as $n \to \infty$ is zero.

You also have some other errors in your notation. For example, you should write $$\hat \theta_n = 2 \bar X,$$ not $\theta = 2\bar X$, which makes no sense since $\theta$ is a parameter, and $\hat \theta_n$ is an estimator of that parameter.

  • $\begingroup$ Excellent, thank you very much. $\endgroup$ – Hernán Jul 3 '18 at 3:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.