$\operatorname{Var}(X+Y\mid X+Y)=\operatorname{Var}(X\mid X+Y)+\operatorname{Var}(Y\mid X+Y)$?

41 Views Asked by At

Is this equality true for independent random variables $X$ and $Y$? Or maybe we should also assume the same distribution?

1

There are 1 best solutions below

3
On

The conditional distribution of a random variable $Z$ given $Z=z$ is just the unit mass at $z$. This has variance $0$. Therefore $\text{Var}(X+Y \mid X+Y) = 0$. On the other hand, $\text{Var}(X \mid X+Y) \ge 0$; equality means $X+Y$ determines $X$ almost surely. That can happen, but usually is not the case. So your equality is not true in general.