Half of studies have been shown to be non-repeatable though through p-hacking :) Maybe it’s all a dice roll.