There are multiple studies on this and judging just from abstracts it seems well-established. I think the data were from forensic samples for pedophiles but since it was the same for heterosexual pedophiles and for homosexual pedophiles it’s not really an issue. (Although it would be good to also have data from non-forensic samples and it’s a kind of thing one can ask about in online surveys.)
It makes little sense considering the difference in prevalence of attraction towards same sex among pedophiles and teleiophiles.
Furthermore, it’s quite confusing considering results of Bailey’s survey: The prevalence of self-reported attraction towards adults among homosexual pedophiles* was high/moderate/none 5 %/12 %/66% for attraction to men but 5 %/13 %/58 % for attraction towards women. (*Quoting Bailey: “If a respondent’s highest attractions were to members of two different categories (e.g., a prepubescent girl and a prepubescent boy), he would be counted as a member of all relevant categories (in this example, as both a heterosexual pedophile and a homosexual pedophile).” It would be good to have separate numbers for homosexual pedophiles and bisexual pedophiles as well as to know how many reported attraction to both men and women.)
Finally, I remember reading a paper (unfortunately, I don’t remember the title) which mentioned study reporting that while relatively more gay men reported being bullied in childhood for being “sissy”, that wasn’t the case for homosexual pedophiles.
Currently there is MRI research in progress in Czechia, Canada and South Carolina (using same methods), together they should have a pretty robust sample (meaning compared to typical size of MRI studies, >50 pedophiles). It will be very interesting to see the results.