0
$\begingroup$

Assume we have three independent random variables A, B and C. The distribution for A,B and C are known(In this case just assume they are all beta distribution with different parameters). I want to compare the three random variables.

Is this equation ${\rm Pr}(C > A\ and\ C > B) = {\rm Pr}(C > A){\rm Pr}(C > B)$ true or false?I don't know if $C>A$ and $C>B$ are independent of each other or not. Please let me know how to judge that.

Also, ${\rm Pr}(C > A\ or\ C > B) = {\rm Pr}(C > A) + {\rm Pr}(C > B) - {\rm Pr}(C > A\ and\ C > B)$should always be correct since $C>A$ and $C>B$ are not mutually exclusive. Please let me know if I'm right.

$\endgroup$
1
  • $\begingroup$ In general, the information that "$C>A$" is evidence to suggest that $C$ is large...hence evidence to suggest $C>B$ as well. So, no...those events are not independent. $\endgroup$
    – lulu
    Jun 6 '18 at 21:29
0
$\begingroup$

No, $C > A$ and $C > B$ are generally not independent, and so $$Pr(C > A \ \text{and}\ C > B) \ne Pr(C > A) Pr(C > B)$$ Examples are easy (e.g. suppose $A$ and $B$ are almost surely constant).

As for your second question, $Pr(E \ \text{or}\ F) = Pr(E) + Pr(F) - Pr(E \ \text{and}\ F)$ is always true, and does not depend on $E$ and $F$ not being mutually exclusive. Of course if they do happen to be mutually exclusive, $Pr(E \ \text{and}\ F) = 0$.

$\endgroup$
1
  • $\begingroup$ Very clear explanation, thank you! $\endgroup$
    – Ben Wu
    Jun 6 '18 at 21:44

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.