Testing order statistics and finding if it is MP test using Neyman Pearson lemma.

85 Views Asked by At

Problem: Suppose X1,X2,...Xn follows exponential (mean=1). Only the largest values of Xi are recorded and denoted by T. To test H0:n=5 vs H1:n=10, show that the most powerful test of size 0.05 rejects H0 for T>c, for some c. Also find c.

Here my question is that I can use NP lemma for joint pdf but can I use the same for order statistics. Alternatively I can just redefine the problem as a RV following the pdf of Max. Of the the RV and then I can use NP lemma to solve the problem. In this case, can we say that my redefined problem and the above stated one are the same?

1

There are 1 best solutions below

3
On

What you have to do is, find $c$ such that \begin{equation} P_0\left(\max\limits_{1\leq i\leq n}X_i>c\right)=0.05. \end{equation} Now, under hypothesis $H_0$, we know that $n=5$. Thus, the probability term on the left hand side of the above equation is \begin{align} P\left(\max\limits_{1\leq i\leq 5}X_i>c \right)&=1-P\left(\max\limits_{1\leq i\leq 5}X_i\leq c \right)\\ & \stackrel{(a)}{=} 1-P(X_1\leq c)^5\\ & =1-(1-e^{-c})^5, \end{align} where in writing $(a)$ above, I have assumed that $X_1,\ldots,X_n$ are independent. Equate the above expression to find out the value of $c$.