Maximum likelihood estimation is investigated in the context of linear regression models under partial independence restrictions. These restrictions aim to assume a kind of completeness of a set of predictors "Z" in the sense that they are sufficient to explain the dependencies between an...