Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

A random variable is uniformly distributed over $(0,\theta)$. The maximum of a random sample of n, call $y_n$ is sufficient for $\theta$ and it is also the maximum likelihood estimator. Show also that a 100$\gamma$% confidence interval for θ is $(y_n, y_n /(1 − γ )^{1/n})$.

Could anyone tell me how to deal with this problem? Do I have to use the central limit theorem?

share|cite|improve this question
The central limit theorem won't help because it's about the distribution of a mean or a sum, not of a maximum. – Michael Hardy Sep 3 '12 at 12:26
I'm getting $(y_n,y_n/(1-\gamma)^{1/n})$. I'm guessing what you wrote was intended to be that. – Michael Hardy Sep 3 '12 at 12:32
yes, that was a typo. I apologize for the mistake. – adamG Sep 3 '12 at 12:42
In general asymptotic theory doesn't help here because the question requires an exact result. But there is asymptotic theory for the maximum. It is called Gnedenko's theorem and can be applied to the uniform distribution. – Michael Chernick Sep 3 '12 at 16:03
up vote 4 down vote accepted

You need to show that $$ \Pr\left(y_n<\theta<\frac{y_n}{(1-\gamma)^{1/n}}\right) = \gamma. $$ Since $y_n$ is necessarily always less than $\theta$, this probability is the same as $$ \Pr\left(\theta<\frac{y_n}{(1-\gamma)^{1/n}}\right)=\Pr\left( \theta(1-\gamma)^{1/n} < y_n\right) = 1-\Pr\left( y_n < \theta(1-\gamma)^{1/n} \right). $$

Notice that $$ \Pr(y_n < c) = \Pr(\text{All $n$ observations}<c) = \Big( \Pr(\text{A single observation}<c) \Big)^n = \left( \frac c\theta \right)^n. $$ Apply this last sequence of equalities with $\theta(1-\gamma)^{1/n}$ in place of $c$.

share|cite|improve this answer
Thanks Michael, I didn't think about using the given interval and work the reverse way. How would you reason to choose this confidence interval if it was not specified in advance? Could you give an intuitive answer? – adamG Sep 3 '12 at 13:03
This confidence interval has $100\gamma%$ within the interval, and $100(1-\gamma)%$ to the right of the interval. One could also construct a confidence interval with $(100(1-\gamma)/2)%$ in each tail. Or even $$100(1-\gamma)%$ to the left of the interval, with the right endpoint equal to $\infty$. The problem so far doesn't completely specify these things. But one could speak of a unique "confidence distribution", and find that. The second line of displayed $\TeX$ in my answer above is the first step in doing that. – Michael Hardy Sep 3 '12 at 13:18

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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