In many situations it's much more an art than a science. Oftentimes you see "we multiplied independent variable X by independent variable Y and that is what we attribute this increase in Z to"...

But it's wrapped up in such complicated rhetoric that no one questions it (perhaps less cares...)

I shouldn't berate it though. It pays the bills..