n=4 (countries) is not enough to draw any robust conclusions.

That's pretty good for studies where we are counting "nations" to come up with our N.

Counting the reported amount of sexual abuse is problematic. It can a sign that people are less likely to report crimes that is in the case of the data for Japan particularly concerning as he suggests: "in these latter years the rapist was less likely to be known to the victim; proving lack of consent became easier."

He is certainly aware of the issue. I think the passage you quote strengthens rather than weakens his conclusion in that case.

The paper doesn't look like a regular academic paper. It has no abstract. The journal in which is published is named: "Porn 101: Eroticism, Pornography, and the First Amendment". 101 isn't a usual name for a journal.

Right, it's a book, not a journal. When access to journal articles requires payment, citing them is problematic.

The fact that first amendment comes up in a journal name suggest that the journal is politically motivated.

There may be some bias in the book. Social science research in general is very politicized, and sex research more than most. Since these findings have potential implications that run counter to received wisdom on child pornography, the most eminent researchers who don't want to lose their grants might be reluctant to do this sort of work. All sex research has to be examined keeping in mind the political goals of the authors, including all the work on the harm done by pornography.