So it looks like the noise patterns in effects tend to closely match those in the 'cause' datasets. Of course, when the cause is a filter, get ready to be disappointed.

While a useful heuristic, it's still not proof. Never will be.