> So cherry-picked cases that range from 15-78% of the data set.
That's a fair point, although what kind of case would be a "cherry" is not clear to me. Both a foreign company winning (implying that patent infringement happened) and losing (possibly feeding suspicion that the judicial system is biased against foreigners) could end up making China look bad.
> 1.6k patent cases over multiple years seems -extremely- low to me.
My mistake. When I checked the blog post by the author of the paper for a date range, I thought it was since 2014 (and until publication in 2018), but actually those cases were all in 2014. (See page 10 of the paper.)