Ambiguity in calculating mean of normal distribution

21 Views Asked by At

Let X be a random variable with distribution $N\left(0,1\right)$

$$\therefore f_X\left(x\right) = \dfrac{1}{\sqrt{2\pi}}e^{-\left(x^2/2\right)}$$

Y be another random variable such that,

$$f_Y\left(x\right) = \dfrac{1}{\sqrt{2\pi}}e^{-\left(x - \mu\right)^2/2\sigma^2}$$

and Z is such that,

$$f_Z\left(x\right) = \dfrac{1}{\sigma \sqrt{2\pi}}e^{-\left(x - \mu\right)^2/2\sigma^2}$$

By location-scale transformation I can write Y as $\sigma X + \mu$.

Simple calculations yield,

$$EX = 0$$ thus, $$EY = E\left(\sigma X + \mu\right) = \mu$$

and,

$$EZ = \dfrac{EY}{\sigma} = \dfrac{\mu}{\sigma} $$

But $EZ$ should be equal to $\mu$. Where am I going wrong?

1

There are 1 best solutions below

2
On

There is no such random variable $Y$ since $f_Y$ is not a density function. What is true is $Z$ has same distribution as $\mu +\sigma X$ and $EZ=\mu+\sigma EX=\mu+0=\mu$.