The methodology is pretty weak.
and more importantly is it possible to do better with the available resources available to a "regular" researcher?
is there even some hard data on ghost job openings? (ie. from court cases or labor board cases or ...?)
The methodology is pretty weak.