Biases. I have already touched on this. I suspect that non-western, vulnerable and marginalized groups are underrepresented.
First, the sex positive community, a highly marginalized group, is explicitly excluded from the study. NSFW material is not counted. Regardless of one’s opinion of this community and its products, posts, and comments, excluding it skews the data.
None Western users are underrepresented. I have shown how the platforms (Misskey/Pawoo) and habits of Japanese and other Asian groups are underrepresented by the dependance on hastags, attention-seeking behaviors, and Asian dominated platforms. These groups are much less likely to call attention to them self than the self aggrandizing west.
It is likely that female acculturated individual are underrepresented due less attention drawing behavior and language. This has been demonstrated to be the case in many studied elsewhere and I propose it is true here.
While I have no data to back this up, I would theorize the same is happening with POC populations.
Reliance on hashtags certainly biases the study against less technical individuals. While a place like the Fedi draws heavily from a more heavily technical population, there are other users.
Reliance text dominated platforms minimizes the visibility of individual who are graphically oriented, such as the photographic community. At least a portion of this population uses PixelFed. It is unclear the extent PixelFed is included in this study.
I question many of the conclusions in the study. That a small minority of individual contribute and drive discussion on the Fedi is hardly surprising. The break down of interests, however, seems have questionable validity since it excludes non-western, minority and less technically savvy individuals. Incomplete inclusion of all Fedi platforms also reduces the visibility of those who engage in non-text information sharing.
I may be incorrect in my assessment since these questions: hash tag dependance; inclusion or exclusion of none-mastodon instance; and measure to insure inclusion are not clearly stated.
At the very least, I hope that any future study will discuss how they are dealing with these kinds of biases and validity issue.