Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

If $X_1,\ldots,X_n$ are iid $U(0,\theta)$, and $X(n) = \max(X_1,X_2,\ldots,X_n)$.

H0: $\theta < 1$

HA: $\theta \ge 1$

Reject Ho if $X(n) > c$ Accept H0 if $X(n) \le c$

  1. Find a constant $k$ such that the Type I error is 0.10.

Attempt: I know this is a one sided test, therefore a UMP exists. I know how to do this if the variables came from a normal distribution but am not sure how to do this if the sample comes from a uniform. Do I need to know the distribution of $X(n)$?

2. Determine the power function.

For the normal, I know $P(Z> c + ((\theta_0 - \theta)/(\sigma/\sqrt{n})$ for a simple hypothesis, but I am stuck in this case for a uniform and a one sided hypothesis.

share|cite|improve this question
Who or what gave you the idea that all one-sided hypotheses admit a UMP test? – cardinal Nov 8 '11 at 1:58
up vote 1 down vote accepted

Notice that $\max\{X_1,\ldots,X_n\}\le x$ precisely if $(X_1\le x\text{ and}\ldots\text{and }X_n\le x)$, and the probability of that can be found by using independence, i.e. by multiplying. Remember that $\Pr(X_1\le x)=x/\theta$ That gives you the cumulative distribution function of the maximum.

share|cite|improve this answer
I'm still not sure how to proceed. After you find the cdf which is (x/theta)^n, do you differentiate and find pdf? I am unclear as to what to do next. – lord12 Nov 8 '11 at 17:50
Assuming the null hypothesis is true, you want the probability to be $0.1$ that $\max > c$. So you know that $\Pr(\max > c) = 1 - (c/\theta)^n$ if $c < \theta$ and $=1$ if $c>\theta$. So $1-(c/1)^n = 0.1$ if $c=\text{what}$? (BTW, if the observed sample max is more than 1, that you can be certain that the null hypothesis is false.) – Michael Hardy Nov 9 '11 at 4:09

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.