2
$\begingroup$

Let's say you have a random variable $X$, which is normally distributed according to $X \sim \mathcal{N}(1,2)$. With $1$ being the mean and $2$ being the variance. Now let's say that there is another random variable $Y$, which is defined as $Y = 10X + 5$. How do you now come up with the distribution for $Y$?

For a normal distribution, expected value equals the mean. So for $X$, the expected value is $1$. Can we now say that the $E(Y) = 10 \ E(X) + 5$, and therefore, the expected value of $Y$ is $15$?

Similarly, can we say that the variance for $Y$ is $25$?

Thus, $Y \sim \mathcal{N}(15,25)$?

$\endgroup$
1
  • $\begingroup$ As far as I remember, the expected value $E$ is a linear operator, so $E(Y)$ will be $15$ indeed, but the variance is not linear, it is quadratic, so $\mathrm{var}(Y)$ is not $25$. It has something to do with the covariance between $X$ and $Y$, I think, but I'll leave to someone who knows it better to give you a definite answer. $\endgroup$
    – Ivo Terek
    Aug 24 '14 at 21:23
2
$\begingroup$

The calculation of $E(Y)$ is correct. If $a$ and $b$ are constants, then $E(aX+b)=aE(X)+b$.

For the variance, use the fact that if $a$ andd $b$ are constants, then $$\text{Var}(aX+b)=a^2\text{Var}(X).\tag{1}$$

So the variance of $aX+b$ is independent of $b$. Intuitively, this is because the variance is a measure of variability, so is unchanged if you shift the distribution by a constant.

Remark: Formula (1), and generalizations, will come up quite often in applications.

Note that the expressions for $E(aX+b)$ and $\text{Var}(aX+b)$ are correct whatever distribution $X$ has.

But to get the additional conclusion that $aX+b$ has normal distribution, we need $X$ to have normal distribution.

Note that in the above calculations, we did not show that $Y$ has normal distribution. How to show it depends on the details of how the normal distribution is defined in your course. We will need to use the fact that $\Pr(Y\le y)=\Pr(10X+5\le y)=\Pr\left(X\le \frac{y-5}{10}\right)$.

$\endgroup$

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.