3
$\begingroup$

If a r.v. $X < a$, does it imply $EX < a$?

  1. If not, why is it different from what I know: If a r.v. $X \leq a$, it implies $EX \leq a$, proved by replacing $X$ with $a$ as the integrand.
  2. Note that $a \in \mathbb R$. If it allows that $a= \infty$, the answer is no: if $X$ has a Pareto distribution with $α=1$, then $X < \infty$, but $EX = \infty$, from http://en.wikipedia.org/wiki/Pareto_distribution and https://stats.stackexchange.com/a/91515/1005
$\endgroup$
  • $\begingroup$ @whuber: I saw some text about how $\infty$ is involved in Lebesgue integrals, and if i am correct, the meanings of $\infty$ as values taken by a random variable $X$ and taken by its expectation $EX$ are the same. So I am still confused. $\endgroup$ – Tim Nov 3 '14 at 21:58
  • $\begingroup$ @whuber, thanks for reminding me about Rudin's book (I forgot where I saw about $\infty$ in Lebesgue integral). According to that book, the definition of Lebesgue integral of a measurable function $f$ is $\int f d \mu := \int f_+ d \mu - \int f_- d \mu$ where $f_+$ and $f_-$ are the positive and negative parts of $f$. $\int f d \mu$ is well-defined iff $\int f_+ d \mu$ and $\int f_- d \mu$ are not both $\infty$, because $\infty - \infty$ is undefined. So I don't see why you said $\infty$ has different meanings for $X$ and $EX$. I see that it has the same meaning in both cases. $\endgroup$ – Tim Nov 3 '14 at 22:51
  • $\begingroup$ I have looked over that section again (1.22 in the second edition) and believe you are correct. I apologize for misleading you: Rudin (rather elegantly) does use the symbol "$\infty$" in a consistent way. $\endgroup$ – whuber Nov 3 '14 at 23:02
4
$\begingroup$

\begin{align} EX &= \int x\cdot \mathbf{1}_{x< a} + x\cdot\mathbf{1}_{x\geq a} \ dP \\ &= \int x\cdot \mathbf{1}_{x< a}\ dP\\ &< \int a\cdot \mathbf{1}_{x< a}\ dP \\ &= a, \end{align}

so yes.

$\endgroup$
  • $\begingroup$ I guess I should add, strict inequality follows because $P\{x<a\}=1$, and this set can be rewritten: $\{x<a\}=\cup_{n> 0}\{x<a-1/n\}$. So some $\varepsilon > 0$ must have $P\{x<a-\varepsilon\}>0$, hence the one integral is strictly less. $\endgroup$ – enthdegree Jun 18 '17 at 18:20

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.