Arsenic Lupin posted:

Here is the thing. When you've got self-reported retrospective data from a secretive organization, that's your dataset and you're stuck with it. However, once you have that dataset it is a scientist's responsibility to acknowledge the study's shortcomings. In 2004 you can't say there are fewer child abusers than in the 1970s-1980s. You can't say that because (1) the Church is known to have a history of underreporting child abuse until there are criminal indictments and (2) children tend not to report abuse until they reach adulthood. The children from the 1990s forward turn out to have had the same problems with Church denial -- see Kansas City and Minneapolis -- and thus any retrospective prediction in 2004 about the continuing rate of child abuse is statistically meaningless.

Data on child abuse have a built-in latency, because many, many children don't realize they were abused because the abuse was normative in their household/school whatever. They don't report because they won't be believed. They *do* report and aren't believed. They don't report because they think it's only them. All of these motives for underreporting become less important -- although not unimportant -- as the children mature. The John Jay study simply didn't have enough reliable data to make forward predictions.