I mean, maybe! But in the spirit of transparency, that assumption (that transfeminine people in AGP erotica-sharing interest groups are representative of transfeminine people in general) is something that that should be made explicit in the paper, so that readers who don't already share that assumption can think for themselves about how to interpret the results, rather than being stuck trusting the authors' interpretations.
-We do have some indications that recruitment method matters to some extent. In Bailey and Hsu 2022 ["How Autogynephilic Are Natal Females?"](http://unremediatedgender.space/papers/bailey_hsu-how_autogynephilic_are_natal_females.pdf), Sample 1 was recruited from AGP erotica-sharing groups, whereas Sample 3 was recruited by asking people to take a survey about atypical sexual interests and filtering for males who answered Yes to whether they had ever wondered whether they might be transgender—and as one might have expected, Sample 1 had a significantly higher mean on the Core Autogynephilia Scale (7.00) than Sample 3 (4.27). Does that 7.00 vs. 4.27 difference matter? What does it mean?
+We do have indications that recruitment method matters. In Bailey and Hsu 2022 ["How Autogynephilic Are Natal Females?"](http://unremediatedgender.space/papers/bailey_hsu-how_autogynephilic_are_natal_females.pdf), Sample 1 was recruited from AGP erotica-sharing groups, whereas Sample 3 was recruited by asking people to take a survey about atypical sexual interests and filtering for males who answered Yes to whether they had ever wondered whether they might be transgender—and as one might have expected, Sample 1 had a significantly higher mean on the Core Autogynephilia Scale (7.00) than Sample 3 (4.27).
-[...]
+Does that 7.00 vs. 4.27 difference matter? What does it mean? (The authors suggest that Sample 3's methodology picked up males who aren't AGP, but that's not the only possible interpretation.) I can't claim to have the answers—but it matters that readers of Bailey and Hsu 2022 were provided the information needed to ask the question.
-In light of this, the discussion of sampling bias in the paper seems inadequate. The "Limitations and Future Directions" section says (emphasis mine):
+In light of this, the discussion of sampling bias in Hsu et al. 2025 seems grossly inadequate. The "Limitations and Future Directions" section says (emphasis mine):
> Because we wanted to maximize our sample sizes of male cross-dressers and transfeminine individuals as much as possible, our recruitment strategy involved indiscriminately posting advertisements to a wide range of Facebook groups, Reddit communities, Discord servers, and other websites or forums where they could be found and _where we also had access and permission to recruit_. Considering that this recruitment strategy was not likely to produce representative samples of male cross-dressers and transfeminine individuals, we acknowledge the possibility that any differences found between those two groups and control samples might be confounded by differences in recruitment strategy. _Those who visit online communities through Facebook, Reddit, and Discord might differ from those who complete studies on Prolific_ in ways that are unrelated to whether they are male cross-dressers, transfeminine, or cisgender.
To be fair, if it was difficult to recruit from "mainstream" trans groups (because people who are loyal to the trans-activist subculture consider the study of AGP to be a threat and refuse to take or share a survey administered by Michael Bailey's student), then that's not the authors' fault ... but whether the result generalizes beyond the surveyed population doesn't depend on whose fault it is! If MtFs as a group are _making themselves harder to study_ for political reasons, I think that's something that the paper needs to dicuss explicitly rather than ignoring, even if it's not usual for psychology papers as a genre.
+
+
+----
I might want to write a blog post about this, but I thought I'd check with the authors and SEXNET first. Kevin, James, anyone—thoughts?
projects / Ultimately_Untrue_Thought.git / commitdiff