I think you have posted other related articles before. Its fairly terrifying to think about. Maybe it was you that also posted about the rarity of repeatable results, that most studies are either never checked or when checked, unverified.
"Perhaps half" seems like a huge amount, and I want to say that there is no way its true, but it unfortunately doesn't seem that unlikely. "Poor methods get results" sums it up very nicely. In a society obessed with instant gratification, there are just too many drivers for bad habits.