Did they study a lot of professions? Just by chance one of them can incur into a type I error.
Personally the most suspicious part about this is the O-LIFE questionnaire; it tests for things like "unusual experiences", "cognitive disorganisation", "impulsive nonconformity", "Introvertive anhedonia", and a bunch of other things that are reminiscent of all the non-replicable personality research from a decade ago.
O-LIFE was developed in 1995, well before replication crisis questions started popping up, so who knows.
However, the article does say it was a "study of magicians", not a study of medical data from the general population.