4
$\begingroup$

Let $X_1,X_2, \dots$ be iid RVs with mean $\alpha$ and variance $\sigma^2$ , and let $Y_1,Y_2, \dots$ be iid RVs with mean $\beta(\neq 0)$ and variance $\tau^2$. Find the limiting distribution of $$Z_n=\frac{\sqrt{n}( \bar X_n - \alpha)}{\bar Y_n}$$ ,where $\bar X=\frac{1}{n}\sum_{i=1}^{n} X_i$ and $\bar Y=\frac{1}{n}\sum_{i=1}^{n} Y_i$

Trial: By Lindeberg_Levy CLT , $$ \bar X_n \text{~} N(\alpha,\frac{\sigma^2}{n})$$ and $$ \bar Y_n \text{~} N(\beta,\frac{\tau^2}{n})$$ I know Slutsky's theorem but unable to use. Please help.


From Max's hint I think my answer will be :

Since $ \bar X_n \text{~} N(\alpha,\frac{\sigma^2}{n})$ we have $ \sqrt{n}(\bar X_n -\alpha)\text{~} N(0,\sigma^2)$

and here $ \bar Y_n \text{~} N(\beta,\frac{\tau^2}{n})$ where $\beta \neq 0$

So now by using Slutsky's theorem we have

$ Z_n =\frac{\sqrt{n}( \bar X_n - \alpha)}{\bar Y_n}\text{~} N(\frac{0}{\beta},\frac{\sigma^2}{\tau^2/n})=N(0,\frac{n\sigma^2}{\tau^2}) $

Here my variance comes out $\frac{n\sigma^2}{\tau^2}$.

Please tell me where I make mistake? Thank you in advance.

$\endgroup$
  • 2
    $\begingroup$ Why do you believe you are unable to use Slutsky's theorem? $\endgroup$ – cardinal Jan 22 '13 at 17:59
  • 3
    $\begingroup$ This is just a shot in the dark so I'm sure someone else will give a better answer. I believe $\sqrt{n}\left(\bar{X}_n-\alpha\right)$ is normally distributed with zero mean and variance $\sigma^2$. As $n\rightarrow\infty$, $\bar{Y}_n$ goes to the constant $\beta$. I would expect that in the limit, $Z_n$ is normally distributed with zero mean and variance $\left(\frac{\sigma}{\beta}\right)^2$. $\endgroup$ – assumednormal Jan 22 '13 at 18:07
  • 1
    $\begingroup$ @Max: If that is a shot in the dark, it's a pretty good one... :) $\endgroup$ – cardinal Jan 22 '13 at 20:44
  • $\begingroup$ Is a better title available for this question? $Z_n$ could refer to almost everything; a description of the ratio we are finding the limit of might be helpful. $\endgroup$ – Silverfish Nov 24 '14 at 21:33
  • $\begingroup$ @Silverfish Changed the title to suit the question in the post itslef. $\endgroup$ – Alecos Papadopoulos Nov 25 '14 at 3:47
8
$\begingroup$

Since user Max never turned his comment into an answer, to lay this one to officially rest:

By Lindeberg-Levy CLT indeed

$$\sqrt{n}( \bar X_n - \alpha) \xrightarrow{d} N(0,\sigma^2)$$

By the Law of Large Numbers

$$\bar Y_n \xrightarrow{p} \beta$$

Then we can apply Slutsky's theorem

$$Z_n=\frac{\sqrt{n}( \bar X_n - \alpha)}{\bar Y_n} \xrightarrow{d}\frac 1{\beta}N(0,\sigma^2) = N(0, \sigma^2/{\beta^2})$$

$\endgroup$

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.