4
$\begingroup$

If $X$ and $Y$ are two (not necessarily independent) Bernoulli's with success probabilities $a$ and $b$ resp., how do we construct the joint dist. in terms of $a$,$b$, and $\rho$---the correlation?

I can get $\mathbb{P}(X=1,Y=1)$ by manipulating the expression for $\rho$, but lost for the other three...

$\endgroup$
2
$\begingroup$

Hint: $P(X=1) = P(X=1,Y=0) + P(X=1,Y=1)$

$\endgroup$
2
$\begingroup$

If you write $c:= P(X=1, Y=1)$, then you have, as others said, by definition of correlation coefficient: $$ \rho = \frac{c - ab}{ \sqrt{a(1-a)b(1-b)} } $$ and $$ c = ab + \rho \sqrt{a(1-a)b(1-b)} $$ Therefore the joint distribution is:

$P(X=1, Y=1) =c$

$P(X=1, Y=0) =a-c$

$P(X=0, Y=1) =b-c$

$P(X=0, Y=0) =1-a-b+c$

(it is simply filling in the contingency table).

$\endgroup$
0
$\begingroup$

The correlation coefficient is equal to $$\frac{E(XY)-ab}{\sqrt{a(1-a)}\sqrt{b(1-b)}}.$$ If you know the correlation coefficient, and $a$ and $b$, then you know $E(XY)$.

But $E(XY)=\Pr(X=1 \cap Y=1)$. From this, using Robrt Israel's hint, you can calculate the rest of the $\Pr(X=i \cap Y=j)$.

$\endgroup$

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.