Double blind studies are great for testing something such as paranormal abilities (e.g., telekinesis, psychic powers, etc.). For example, neither the researchers nor the participants knowing which item was hidden in the envelope or box might be a double-blind type of study for testing somebody's psychic prediction abilities.
With ghosts it's much more problematic. As a researcher you can't run an experimental protocol on ghosts like you can with psi, using experimental apparatus you can manipulate and physical responses that you can directly measure. You can't even detect ghosts, so it's impossible to "measure" them or their influence in a scientific way in the first place. You'd still end up saying something like "the needle moved for some reason, let's blame a ghost".
It's kind of like running a double-blind study on Bigfoot. You first need a bonafide, verified Bigfoot at your disposal to do it. That's the hard part.