Let $Z_{1},\ldots, Z_{n}$ randon variables i.i.d with symmetric contunuous distribution in $\theta$.

We consider hypotesis test $$\begin{array}{rcl} H_{0}: && \theta=\theta_{0} \\ H_{a}: && \theta>\theta_{0} \\ \end{array}$$

Let $T_{\theta}$ be a statistic, and let $$\alpha=\mathsf{P}(T_{\theta_{0}}\geq c | H_{0} \mbox{ true})$$ is significance level, that is, reject the null hypothesis if $T_{\theta_{0}}(Z)\geq c$.

My question: What is the power of $T$?

Remark: We remember that

Definition. The power of a hypothesis test is the probability of making the correct decision if the alternative hypothesis is true. That is, the power of a hypothesis test is the probability of rejecting the null hypothesis $H_{0}$ when the alternative hypothesis $H_{a}$ is the hypothesis that is true.

I know that $$\textrm{power}=1-\mathsf{P}(T_{\theta_{a}}\geq k|H_{a}\mbox{ is true})=\mathsf{P}(T_{\theta_{a}}\leq k|H_{a}\mbox{ is true})$$ where $k$ depend of $c$. I do not understand as from $ c $ I can get $ k $ in general. For example, for statistic $$T_{\theta}=\frac{\sqrt{n}(\overline{Z}-\theta)}{s}$$ It is easy to see the dependency of $ c $ (see this link). But it is not easy for the case of the signed statistic $$B=\sum_{i=1}^{n}\psi(Z_{i}-\theta) \quad \mbox{ where } \quad \psi(x):=\left\{\begin{array}{ll} 1 & x>0\\ 0 & x\leq 0 \end{array} \right.$$ or the Wilcoxon signed kank statistic $$W^{+}=\sum_{i=1}^{n}\psi(Z_{i}-\theta)R^{+}_{i}$$ where $R^{+}=(R^{+}_{1},\ldots,R^{+}_{n})$ is the rank vector of $|Z_{1}|,\ldots,|Z_{n}|$. I really have not been able to see that dependence in these cases.


Your Answer

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

Browse other questions tagged or ask your own question.