Here's the problem: Let $X_{1},\dots,X_{n}$ be a sequence of independent random variables with $X_{i}\sim P_{i}$. Let $E\left[X_{i}\right]=\mu_{i}$ and $Var\left[X_{i}\right]=\sigma^{2}$. Suppose that

$\frac{1}{n}\sum_{1\leq i\leq n}\sigma_{i}^{2}\to0$

We want to show that $\bar{X}_{n}-\bar{\mu}_{n}\overset{P}{\to}0$, where

$\bar{\mu}_{n}=\frac{1}{n}\sum_{1\leq i\leq n}\mu_{i}$

I'm absolutely sure that we have to use Markov's inequality for this and link it up with the facts about the variance in the set up

$P\left\{ \left|\bar{X}_{n}-\bar{\mu}_{n}\right|>\varepsilon\right\} \leq\frac{E\left[\left(\bar{X}_{n}-\bar{\mu}_{n}\right)^{2}\right]}{\varepsilon^{2}}$

But I'm not sure where to proceed from here. I'm also asked to show that if $\bar{\mu}_{n}\to\mu$ then $\bar{X}_{n}\overset{P}{\to}\mu$, but I'm also confused about how to show this.

  • $\begingroup$ Can you elaborate on the $X_i \sim P_i$ note? $\endgroup$ – Aaron Montgomery Oct 4 '17 at 2:58
  • $\begingroup$ I says that each $X_i$ is drawn from a different probability distribution $P_i$. In this case the $X_1, \dots, X_n$ are not iid. $\endgroup$ – kittykat Oct 4 '17 at 3:05
  1. Using independence $$ \mathsf{E}(\bar{X}_n-\bar{\mu}_n)^2=\frac{1}{n^2}\sum_{i=1}^n\sigma_i^2. $$
  2. $$ |\bar{X}_n-\mu|\le |\bar{X}_n-\bar{\mu}_n|+|\bar{\mu}_n-\mu|. $$
  • $\begingroup$ Thank you very much. This was very helpful. $\endgroup$ – kittykat Oct 4 '17 at 3:34

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.