0
$\begingroup$

I have to find the conditional pmf and show it sums to one. Using the formula for conditional probability, I have written down $$\mathbb{P}(X_1=a_1,...,X_n=a_n \mid S_n=k) = \frac{\mathbb{P}(X_1=a_1 ,..., X_n=a_n, S_n=k)}{\mathbb{P}(S_n=k)}$$

I know $S_n$ follows a $Binomial(n,p)$ distribution. What I don't understand is how to deal with the top. If i just multiply each Bernoulli pmf I get $\frac{k!(n-k)!}{n!}$, which I'm pretty sure is not the right answer. Then I have to deduce whether $X_1,...,X_n$ are conditionally independent given $S_n=k$. This confuses me for the same reason, I'm not sure how to deal with $\mathbb{P}(X_1=a_1,...,X_n=a_n \mid S_n=k)$ or how to get $\mathbb{P}(X_1=a_1\mid S_n=k)$. Thanks in advance.

$\endgroup$
1
$\begingroup$

You have the right conditional pmf provided $a=(a_1,\ldots,a_n)$ is in the set $$ S:=\Big\{a:a_i\in\{0,1\},|\{i:a_i=1\}|=k\Big\},$$ and the pmf is zero otherwise. To show that the pmf sums to $1$ you just have to count the number of elements in $S$, which you should be able to do.

To test for independence, try a checking events of the form $A=\{X_1=X_2=\ldots=X_k=1\}$ and $B=\{X_{k+1}=\ldots=X_n=0\}$. If $(X_1,\ldots,X_n)$ were independent conditioned on $\{S_n=k\}$ then you would have $\mathbb P(A\cap B\,|\,S_n=k)=\mathbb P(A\,|\,S_n=k)\mathbb P(B\,|\,S_n=k)$. Is this the case here?

$\endgroup$
  • $\begingroup$ k of the n $a_i$'s have to be 1, so should it seems like there are n choose k elements. However, I don't see how this connects to the pmf summing to 1. $\endgroup$ – N Dizzle Sep 12 '18 at 2:29
  • $\begingroup$ Since $\mathbb P(X=a\,|\,S_n=k)$ does not depend on $a$, one simply has that the sum is equal to the number of elements in $S$ times the probability of one of them, that is, ${{n}\choose{k}}\frac{k!(n-k)!}{n!}$. Now expand ${{n}\choose{k}}$ and simplify. $\endgroup$ – Jason Sep 12 '18 at 13:26
  • $\begingroup$ A gentleman and a scholar. Thank you kind sir. $\endgroup$ – N Dizzle Sep 13 '18 at 0:02

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.