1
$\begingroup$

Let's say $A$, $B$ and $C$ are three different events. Is there any mathematical relation between these conditional probabilities: $\Pr(A\mid B,C)$, $\Pr(A\mid B)$ and $\Pr(A\mid C)$?

Note: In the relation there could be expressions such as $\Pr(A)$, $\Pr(B)$, $\Pr(C)$, $\Pr(B\mid C)$, etc. but not conditional probabilities in which $A$ is the condition, e.g., $\Pr(B\mid A)$. In other words, we only observe the events $B$ and $C$, and we would like to estimate the probability distribution of the event $A$.

$\endgroup$
1
  • 1
    $\begingroup$ I changed Pr(A|B) to \Pr(A\mid B) so you see $\Pr(A\mid B)$ instead of $Pr(A|B)$. ${}\qquad{}$ $\endgroup$ Jan 26, 2015 at 17:36

1 Answer 1

0
$\begingroup$

As an example of the issue, consider $B$ and $C$ independently having probability $\frac12$. Then consider $A$ as $B \text{ XOR } C$, i.e. $A$ occurs if one but not both of $B$ and $C$ occur. Now look at the conditional probabilities:

  • $\Pr(A\mid B)=\frac12$
  • $\Pr(A\mid C)=\frac12$
  • $\Pr(A\mid B,C)=0$

In general it is possible for $\Pr(A\mid B,C)$ to take any value (including $0$ and $1$) given that $\Pr(A\mid B)$ and $\Pr(A\mid C)$ have any values (except perhaps $0$ and $1$).

$\endgroup$

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.