0
$\begingroup$

I am doing some practice problems to prepare for my statistics exam, and I just want to know if my reasoning is correct on one problem, and if not, I want to know how I should reason through this. The question is as follows:

Let X and Y follow a bivariate normal distribution with means (3, 2), variances
(1, 4) and covariance c.

$\begin{bmatrix}X\\Y\end{bmatrix} = N(\begin{bmatrix}3\\2\end{bmatrix}, \begin{bmatrix}1&c\\c&4\end{bmatrix})$

What is the Marginal Distribution of X?

My initial reasoning is that one method I could go about it is the calculation of the joint PDF and integrating it, but that doesn't sound very clean. A cleaner method could possibly be that since the joint is determined by two normal distributions, then $f(X,Y) = f(X)f(Y)$, and $f_x(X)$ (the marginal distribution of X) equals $\int_{-\infty}^{\infty}f(X)f(Y)dY = f(X)\int_{-\infty}^{\infty}f(Y)dY = f(X)$ since $f(Y)$ is a normal distribution, with the final answer being $f(X) = N(\mu_1, \sigma_1^2) = N(3,1)$. This seems to make sense but seems a little bit too simplified. Could anyone give me insight on how to tackle this problem, and if I am wrong anywhere, point it out? Thank you.

$\endgroup$
1
$\begingroup$

Bivariate is a special case of jointly normality for 2D, which means these variables are also marginally normal. So, above $X$ and $Y$ are normal RVs. From the mean vector and the covariance matrix, we know that $\mu_X=3$ and $\sigma^2_X=1$, so $X\sim N(3,1)$. Similarly $Y\sim N(2,4)$. So, your final answer is correct.

But, you cannot say $f(x,y)=f(x)f(y)$ when $c\neq 0$, since it implies independence. Check the bivariate distribution formula here.

And, it is always true that $\int_y{f(x,y)dy}=f(x)$, not only when you factorize the joint PDF out, i.e. like $f(x,y)=f(x)f(y)$.

$\endgroup$

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.