I read a paper that makes an assumption that $E(x\mid xy)$ linear in $xy$ where $x$ and $y$ are independent.
I was wondering what is fundamental about this assumption, or in other words, what specifics about the density function do we assume to get this linearity.
Thank you!