Thread: need help to construct UMPT (uniformly most powerful test)

1. EDIT: Whoops

2. Originally Posted by Volga
Would you mind showing how one would approach the second part of the question, viz "Suppose n=10, sample mean is 2.42, and =1.5. Will you reject Ho at a significance level of 5%?" I am still hoping to have this completed as this is a good practice question for the exam. I promise never to attempt constructing a UMPT in real life after I am done with this Stat Inference exam )))
How can the sample mean be 2.42? That implies $\sum X_i = 24.2$ which is impossible.

Anyways, $P_{\lambda = 1.5} (\sum X_i > 24.2) = .01$

so yes, you would. I used R to calculate the probability, since it's a tedious calculation to do exactly (essentially summing the pmf). In other words, the appropriate k that gives the size of desired test is less than 24.2.

I don't worry about you having to calculate UMP tests in real life because, in general, they don't exist Even throwing in a single nuisance parameter is typically going to ruin things for you, so most of the tests that people use in practice aren't UMP.

3. I used R to calculate the probability
what's that?

Because, I wouldn't know how to calculate $P_{\lambda = 1.5} (\sum X_i > 24.2)$.

4. Originally Posted by Volga
what's that?

Because, I wouldn't know how to calculate $P_{\lambda = 1.5} (\sum X_i > 24.2)$.
R is (free) software. Explicitly you calculate it as

$
P(\sum X_i > 24.2) = 1 - P(\sum X_i \le 24.2) = 1 - \sum_{x = 0}^{24} \frac{15^x e^{-15}}{x!}
$

because $\sum X_i$ is Poisson(15) when $\lambda = 1.5$.

5. Oh I see! I thought "R" was some approximation formula )))

Page 2 of 2 First 12