-1
$\begingroup$

The question reads:

Let $\theta > 0$ and $X \sim \mathcal{U}[0, \theta]$, i.e. $X$ is uniformly distributed on the interval $[0, \theta]$.

Assume that $\theta$ is unknown, but we can observe $X$. For given $\theta_1$, we want to test the hypothesis H0: $\theta \geq \theta_1$ against the alternative H1 : $\theta < \theta_1$. Consider the test which rejects H0, if and only if $X < c$. How should we choose $c$, as a function of $\theta_1$ and $\alpha$, to get a test with confidence level $\alpha$? Carefully justify your answer.

I struggle to understand how to approach this question/ how to carry it out? Any help would be much appreciated as this question is due in today. Thanks!!

$\endgroup$
  • $\begingroup$ What is uniformly distributed? $\endgroup$ – the_candyman Feb 26 at 14:55
  • $\begingroup$ @the_candyman Sorry, completely blanked out the first bit of the question, have edited the post to make it correct now. Should make sense. $\endgroup$ – James Debenham Feb 26 at 15:00
3
$\begingroup$

The significance level $\alpha$ corresponds to the maximum Type I error you are willing to accept for the test; that is to say, the erroneous conclusion to reject the null hypothesis when it is true: $$\Pr[\text{reject } H_0 \mid H_0 \text{ true}] \le \alpha.$$ Since you are already told that the criterion to reject $H_0$ is if $X < c$, and you are also told that the null hypothesis is $H_0 : \theta \ge \theta_1$, this becomes $$\Pr[X < c \mid \theta \ge \theta_1] \le \alpha.$$

Questions you should ask yourself:

  1. For a fixed $\theta$, what is the conditional probability $\Pr[X < c \mid \theta]$?
  2. As $\theta$ increases, does $\Pr[X < c \mid \theta]$ increase, or decrease, for a fixed $c$?
  3. How does this inform the choice of $\theta$ under $H_0$, and $c$ as a function of $\theta_1$ and $\alpha$, to ensure a level $\alpha$ test?
$\endgroup$

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.