ps. the study linked is what I would say is poor evidence in part because they have 3 populations: two from Australia (1990, 2008) and one from Sweden (2010). Not exactly a broad sample.