Simplifying a conditional expectation using LIE

40 Views Asked by At

Is it possible to simplify the following expression, using LIE or otherwise?

$\mathbb{E}\left[(1-X)\mathbb{E}(Y|X)+X\mathbb{E}(Z|X)\right]$

where $X, Y \text{and } Z$ are random variables with $X, Y$ and $X, Z$ correlated. At a glance, it looks like it could work down to something like

$\mathbb{E}(1-X)\mathbb{E}(Y)+\mathbb{E}(X)\mathbb{E}(Z)$

or

$\mathbb{E}[(1-X)Y+XZ]$

but neither seems possible. If all these expressions are different, what would be an intuitive way of understanding the difference between them all?

Thanks in advance for the help.

1

There are 1 best solutions below

1
On BEST ANSWER

By pushing $1-X$ and $X$ into the inner expectations (check that you understand why this is possible), we have $$E[E[(1-X)Y \mid X] + E[XZ \mid X]] = E[(1-X)Y] + E[XZ] = E[(1-X)Y + XZ].$$

If $E[Y\mid X] = E[Y]$ and $E[Z \mid X] = E[Z]$ (e.g., if $Y$ and $Z$ are each independent of $X$), then you could rewrite your original expression as $E[1-X] E[Y] + E[X] E[Z]$. But this does not hold in general.