If $v_T=\mathbb E[operatorname{IF}(X;T,N(\theta_0,1))^2]$, asymptotic normality gives an asymptotically level-$\alpha$ test that rejects when
$$
\widehat\theta_T>\theta_0+z_{1-\alpha}\sqrt{v_T/n}.
$$
For the median, $v_T=\pi/2$. For Huber,
$$
v_T=\frac{\mathbb E[\min(Z^2,k^2)]}{(2\Phi(k)-1)^2}.
$$
For the trimmed mean, with $q=\Phi^{-1}(1-\gamma)$,
$$
v_T=\mathbb E\left[
\frac{\max(-q,\min(Z,q))^2}{(1-2\gamma)^2}
\right].
$$
These tests have bounded influence functions, so a small contamination proportion has bounded first-order effect on their statistics, asymptotic levels, and powers. Their finite-contamination protection is quantified by the breakdown points in part (c).

Solved by gpt-5.6-sol high.


Solution

Past exam of the mathematics course of the University of Cambridge / 2026 / iii / Paper 223 / 1 / d / Solution

 Solution

ID: past-exam-of-the-mathematics-course-of-the-university-of-cambridge/2026/iii/paper-223/1/d/solution