On my last exam, this was one of the question I got wrong. However this was the only question I didn't manage to figure out even after the exam. Can anyone help me? With part c) especially?

In order to test the null hypothesis that X has a uniform distribution on the interval $(0,1)$ against the alternative that X has a triangular distribution $[f(x) = 2x$ for $0 < x < 1]$ a random sample of size n is chosen. You want to find the most powerful test at significance level $\alpha = 0.10$

a) If $n = 1$, for what values of X do you reject the null hypothesis?

b) What is the power of the test in a)?

c) If n = 10, for what observations do you reject the null hypothesis?


Since this is a simple vs simple hypothesis test, use NP lemma with $$H_0: f(x)=\,\mathbb{1}_{(0,1)}(x)\quad \text{ against }\quad H_1: f(x)=2x \,\mathbb{1}_{(0,1)}(x).$$ For part c, the likelihood ratio is $$ L = \frac{f_1(x_1,\ldots,x_{10})}{f_0(x_1,\ldots,x_{10})} = 2^{10} x_1\cdots x_{10}. $$ So the rejection region is the set of $x$ for which \begin{align*} L > c &\iff 2^{10} x_1\cdots x_{10} > c \\ &\iff \sum_{i=1}^{10} (-\log x_i) < c'. \end{align*} To find $c'$, notice that $$X_i\sim Unif(0,1) \implies -\log X_i \sim Exp(1)= Gamma(1,1)$$ and $$Y_1,\ldots,Y_n \sim Gamma(1,1) \,\,iid \implies \sum_{i=1}^n Y_i \sim Gamma(n,1).$$


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.