I came across this information in Statistical Inference by Casella and Berger. The problem was the authors didn’t explain how they derived the second equality (highlighted). I never saw such an expression as $P(X > s, X > t)$. Also, why was it divided by $P(X > t)$? Was it because this was the case of conditional probability? Could you please tell me the logic of that derivation? Thanks.
1 Answer
$\begingroup$
$\endgroup$
2
You are probably familiar with the rule and just don't recognize it in this setting. It applies to densities...
$f(x|y)=\frac{f(x,y)}{f(y)}$
...but also to sets (which is your case)...
$P(A|B)=\frac{P(A \cap B)}{P(B)}$
In words, a conditional thing equals the joint thing over the marginal of the condition. This follows from the "chain rule".
-
$\begingroup$ Thanks, Alvaro. To clarify, did the rule as mentioned in your first sentence refer to conditional probability? If so, I wonder why the authors didn't say it, and should had used the expression $P(A \cap B)$ instead? $\endgroup$– NemoCommented Jan 23, 2020 at 12:03
-
1$\begingroup$ Yes, it's a factorization of conditional probability. $P(A\cap B)$ is set notation and $P(A,B)$ is probability notation (I fixed that in my answer now, sorry), but they signify the same thing. $\endgroup$ Commented Jan 23, 2020 at 12:10