# Constant times a random variable question

1. Sep 26, 2007

### jimmy1

A random variable X follows a certain distribution. Now say I multiply the random variable X by a constant a. Does the new random variable aX follow the same distribution as X?

2. Sep 26, 2007

### EnumaElish

Well, let's take it through the basics. If X ~ FX, then FX(x) = Prob{X < x} for x < X < $\bar x$. Let Y = aX for ax < Y < a$\bar x$; then FY(y) = Prob{Y < y} = Prob{aX < y} = Prob{X < y/a} = FX(y/a). Therefore FY(y) = FX(y/a) for ax < Y < a$\bar x$.

Is FY the identical distribution as FX? No. Does it belong to the same family as FX? Yes, up to a scaling factor.

Last edited: Sep 26, 2007
3. Sep 28, 2007

### EnumaElish

Note: in the previous post I assumed a > 0. If a < 0, then FY may not even belong to the same family as FX.

The case of a = 0 is trivial, but you should bear that in mind, too.

4. Aug 9, 2011

### mheena

we know that if X is normally distributed, then so cX for any nonzero real number c.
also X + d is normally distributed, for any real number d.
can anyone please show me the proof?thanks

5. Aug 10, 2011

### chiro

If you want a systematic way to figure this out, use moment generating functions and if the structure of the mgf is the same as the unscaled distributions mgf, then you know that the distribution doesn't change and you can see how the parameters of the distribution have changed.

6. Aug 10, 2011

### chiro

Like I said to the previous poster, use moment generating functions. Let U = cX + d. Use E[e^(tU)] be the mgf of U and you will find that this will give the mgf of a normal distribution with different parameters which will tell you what scaling and translating a normal does to its parameters. (Translating adds to the mean, scaling changes the variance by multiplying it by c^2).