# Medians

• Oct 5th 2009, 12:43 PM
Let $\displaystyle \phi (x)$ be a strictly increasing function of a real variable. Show that the result of applying the sign test to a sample $\displaystyle X_1, \dots , X_n$ with hypothetical median $\displaystyle M_0$ is identical to that of applying the test to the transformed sample $\displaystyle \phi (X_1), \dots , \phi(X_n)$ with hypothetical median $\displaystyle \phi (M_0)$.
Consider a random sample $\displaystyle X_1, \dots ,X_n$ from an unknown distribution $\displaystyle F(x - \theta)$ where $\displaystyle \theta$ is unknown and $\displaystyle F(0) = \frac{1}{2}$, i.e. $\displaystyle \theta$ is the population median. Show that the Hodges-Lehmann point estimate of $\displaystyle \theta$ arising from the Sign Test is the sample median of the data.
Now in the notes it says that when $\displaystyle \theta = M$, $\displaystyle \theta$ is symmetric around some $\displaystyle a_0$ which occurs when $\displaystyle a_0 = \frac{n}{2}$.
So its symmetrical around $\displaystyle X_{\frac{n}{2}}$ which if you performed the sign test on would show its the median...! Hellary I cant really explain my thoughts I have no idea how I'm gonna write this down...