I'm having trouble showing that the 2nd central moment is finite. I have $X_1,\ldots,X_n \overset{iid}{\sim} f(x)$ with $E[X_1]=\mu$ and $E[X_1^k]$ exists and is finite for any integer $k \geq 1$.

I would like to use Law of Large Numbers, so I need to show that either $E[|X_1|]$ is finite or that $E[(X_1-\mu)^2]$ is finite. I tried proving the first one with Jensen's but got stuck since absolute value is convex, not concave.

So now I'm stuck trying to show second central moment is finite.

  • $\begingroup$ It's good you're having trouble, because the conclusion is false unless $k$ has some special values. What do you know about $k$? $\endgroup$ – whuber Jan 25 '13 at 16:53
  • $\begingroup$ If $E[X_1] = \mu$ where $\mu$ is finite, then $E[|X_1|]$ is also finite. $\endgroup$ – Dilip Sarwate Jan 25 '13 at 17:12
  • $\begingroup$ You need to clarify exactly what the the $E[X^k]$ statement means. Is it all $k$?, all $k<M$?, some fixed $k$?... $\endgroup$ – Dave Jan 25 '13 at 22:31
  • $\begingroup$ For $k \geq 1$ an integer. $\endgroup$ – bdeonovic Jan 26 '13 at 20:15
  • 1
    $\begingroup$ You say you want to use the LLN. For what and on what quantity? Note that the statement $\mathbb E X_1 = \mu$ for $\mu \in \mathbb R$ automatically implied by definition that $\mathbb E |X_1|$ is finite! (@Dilip also alludes to this in a previous comment.) So, there is nothing to prove in that respect. That $\mathbb E X_1^k$ exists and is finite for any integer $k \geq 1$ is a very strong statement. $\endgroup$ – cardinal Jan 26 '13 at 22:37

Isn't the second central moment just the variance? So $$E[(X_{1}-\mu)^{2}] = E(X_{1}^{2})-(E(X_{1}))^{2}$$

Since we know that $E[X_1] = \mu$ and $E[X_{1}^{2}]$ exists and is finite it follows that the variance is finite.


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.