There's no reason it can't be both at the same time: IQ helps for education and education helps raise IQ.

Since they are controlling for the other effects, that means this analysis doesn't measure them. (Assuming, that is, the controls worked.)