Expectation conditioned on multiple independent variables

122 Views Asked by At

I'm trying to figure out whether:

$$ \mathbb{E}\left[X\mid Y, Z\right] = \frac{\mathbb{E}\left[X\mid Y\right] * \mathbb{E}\left[X\mid Z\right]}{\mathbb{E}\left[X\right]} $$

if $Y$ and $Z$ are independent. This is true for probabilities with Bayes' Rule, and seems intuitively plausible, but I'm having trouble proving it here for expectations. Any thoughts?

1

There are 1 best solutions below

1
On BEST ANSWER

No, it generally doesn't hold. Let $Y$ and $Z$ be two i.i.d. Bernoulli(1/2) random variables and let $X = Y+Z$. Then,

$$E[X|Y,Z] = Y+Z \neq (Y+1/2)(Z+1/2) = \frac{(Y+1/2)(Z+1/2)}{1} = \frac{E[X|Y]E[X|Z]}{E[X]}.$$

It does hold when $X,Y$ and $Z$ are all mutually independent. Or more generally when $X$ is independent of $Z$ and also conditionally independent of $Z$ given $Y$.

Edit: fixed a typo in the last sentence