0
$\begingroup$

I'm dealing with this problem for a while, and although I have some intuition, I think I'm missing something, please take a look:

We are given a random sample $X_1, X_2,...,X_n$ from Poisson$(\theta)$ distribution. I'm asked to prove that the conditional distribution of: $X_1, X_2,...,X_{n-1}$ given $Y=\sum_{i=1}^n X_i$ is multinomial.

What I have tried so far:

$P(X_1, X_2,...,X_{n-1} \mid Y = y) = \frac{P(X_1, X_2,...,X_{n-1},Y = y)}{P(Y=y)}$

Then, although I know the $n-1$ samples of $X_i$ are independent among each other, and that I could get their joint pdf easily, the problem comes up when I need to related such joint pdf to the pdf of $Y$. Since, I'm not sure whether they are independent, otherwise, the product of the pdfs would suffice.

I know that $Y \sim \text{Poisson}(n\theta)$.

Any clue? Thanks!

$\endgroup$
1
$\begingroup$

Hint

We have $$ P(X_1=x_1,\ldots, X_{n-1}=x_{n-1}, Y=y) = P(X_1=x_1,\ldots, X_{n-1}=x_{n-1}, X_n=y-\sum_{i=1}^{n-1}x_i) \\=f(x_1)f(x_2)\ldots f(x_{n-1}) f\left(y-\sum_{i=1}^{n-1}x_i \right)$$ where $$f(x) = \frac{\theta^xe^{-\theta}}{x!}$$

$\endgroup$
  • $\begingroup$ Oh.. I see, Good hint. I'm gonna work on it.. and come back. Hopefully would be able to get some help from you. Thanks! $\endgroup$ – pkenneth81 Apr 16 '18 at 1:52
  • $\begingroup$ OK. I worked out the conditional distribution: $ \frac{P(X_1=x_1,X_2=x_2,...,X_{n-1}=x_{n-1},X_n = y - \sum_{i=1}^{n-1} x_i)}{P(Y=y)}$ And obtained: $\frac{y!}{\prod_{i=1}^{n-1} x_i!} \cdot \frac{1}{n^y (y- \sum_{i=1}^{n-1} x_i) !}$ Do you guys recognize the multinomial distribution somehow? $\endgroup$ – pkenneth81 Apr 16 '18 at 2:39
  • $\begingroup$ @htennek2k That doesn't look wrong math-wise, but can't say I do. Are you sure you aren't actually supposed to compute $$ P(X_1=x_1,\ldots, X_n=x_n\mid Y=y)$$ where $Y=\sum_{i=1}^n X_i$? $\endgroup$ – spaceisdarkgreen Apr 16 '18 at 3:02
  • $\begingroup$ Well, literally, the problem says: Show that the conditional pdf of $X_1,X_2,...,X_{n-1}$, given $Y=y$, is multinomial. This is from Introduction to Mathematical Statistics of Hogg, Problem 7.6.6. $\endgroup$ – pkenneth81 Apr 16 '18 at 3:23
  • $\begingroup$ Thanks for your reply though. What I got seems similar to the multinomial distribution, but cannot see the relation with those odd factors I get in the denominator. $\endgroup$ – pkenneth81 Apr 16 '18 at 3:27

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.