4
$\begingroup$

Given a random variable $X_1$ drawn from a distribution with cdf $F$, and random variables $X_2, \cdots,X_n$ drawn from another distribution with cdf $G$, what is the formula for the probability that $X_1$ is the $k$-th out of the $n$ variables, when they are sorted in order. If it helps, assume that $F$ and $G$ are both defined on [0,1], with positive density everywhere.

$\endgroup$
6
$\begingroup$

By computing the probability that $k-1$ of the $G$-distributed $X$'s are less than or equal to $X_1$, conditional on the value of $X_1 = t$, and integrating over all $t$ we obtain

$$\int_{-\infty}^{\infty}{\binom{n-1}{k-1} G(t)^{k-1}(1-G(t))^{n-k}dF(t)}.$$

This of course assumes independence of all the $X_i$. Given no assumptions about any relationship about $F$ and $G$, this expression should not simplify in general.

$\endgroup$

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.