1
$\begingroup$

$\newcommand{\Cov}{\operatorname{Cov}}$

I have a practice question I'm trying to answer in studying for an upcoming exam:

$X\sim N(0,1)$ and $Y\sim N(0,1)$ and I have $\rho(X,Y)=0.4$. Define a random variable $Z = X + wY$ that also has a normal distribution. What value of $w$ makes $Z$ independent of $X$?

My thoughts is solve for when $\Cov(X,Y) = 0$, so I was going to set it up as: $\rho(X,Y) = \frac{\Cov(X,Y)}{\sigma_X \sigma_Y}$ and solve $0.4 = \frac{\Cov(X,Y)}{1}$ so $\Cov(X,Y) = 0.4$ and then I'd have $\Cov(X,Y) = \Cov(X,wY) = w\Cov(X,Y) = 0.4w$ and solve $0.4w = 0$ so $w = 0$ is when $Z$ is independent of $X$? But I have a feeling I'm not approaching this correctly and am doing something wrong. Can someone please help guide me in the right direction?

$\endgroup$
  • $\begingroup$ $w=0$ doesn't make sense at all, since this would mean $Z=X$ and $X$ is clearly not independent of $X$. $\endgroup$ – saz Mar 18 '13 at 17:38
1
$\begingroup$

$\newcommand{\Cov}{\operatorname{Cov}}$

You need $\Cov(Z,X)=0$, not $\Cov(X,Y)=0$.

Then $$ 0=\Cov(Z,X)= \Cov(X+wY,X)=\Cov(X,X)+w\Cov(Y,X), $$ and recall that $\Cov(X,X)=1$, etc.

Also, you omitted to add that the pair $(X,Y)$ is jointly normally distributed. Without that, your conclusion that $X+wY$ is necessarily normal is not justified.

$\endgroup$
  • $\begingroup$ Thank you so much for the help! $\endgroup$ – Abi Mar 18 '13 at 18:43

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.