I'm not actually trying to argue anything except to not overinterpret data. When I look for studies, everything I find is small, flawed, inconclusive, overinterpreted, etc.. But people (e.g. you) sound very convinced, so I'm always hopeful that I've missed something. I'm not very adroit at searching social science literature.

I'm a proper scientist in this regard: I don't believe things until I'm shown the evidence. It doesn't matter what it is. Climate change, whether guns are used to prevent crime, impact of parenting on violence, radiocarbon dating, dark matter, unfairness of SATs, it doesn't matter. If there are remotely plausible arguments for more than one possibility, or the finding is non-intuitive, I want to see evidence.

So, let's entertain the hypothesis that trans women might have male-pattern criminality, or perhaps a subset of them might, and ask two things: (1) is it highly implausible on the basis of what we know about the mechanism, and (2) is it highly implausible on the basis of what we know from evidence.

When it comes to mechanism, we don't understand very well why people are trans. Since we don't understand that very well, we can't tell which behaviors will go with gender expression and which won't. Maybe they all will. Maybe some will. Maybe which will depends on testosterone or other hormonal factors. Maybe none will. Because we don't have a principled understanding, this really doesn't constrain things at all. If we ask if some other feature that strongly differentiates the population of women from the population of men--like sexual attraction--we find that, no, trans women are not a perfect match to cis women. So that's no help.

When it comes to evidence...well...that's not clear, and it is the pretense that it is clear when it in fact the data hardly constrains the possibilities at all that I am arguing against.

Now, it's entirely valid to ask a methodological question: if we take evidence of this sort to show the criminality of some group, shouldn't it also apply to other groups?

Yep! It should. That's a good point.

Okay, so let's entertain the hypothesis that lesbians, or maybe a subset of lesbians, show male-pattern criminality.

Do we understand in principle why lesbians have sexual attraction that is more commonly expressed by men? Well...no.

Do we have clear studies that establish causal factors in the cases of lesbians? Welllll...not that you linked. I've never looked before. Maybe there are compelling studies.

So again we have the same issue of uncertainty. Of course, sending lesbians to men's prisons isn't a good idea (you have to consider what some of the men will try to do), but on the other hand, sending lesbians to a women's prison could also cause problems. (Indeed the article you linked was arguing that there were problems!)

(In the case of black women specifically, we do have a pretty good hypothesis for what's going on, because pretty careful studies have been done on men, and excess violence is explained by a combination of culture--most notably the presence or absence of senior male authority figures, socioeconomic status, and education, with the residual variation that might be explained by race being very small compared to the other factors and quite possibly zero. So although I haven't seen a study for black women specifically, because we do have a fairly decent understanding of what appears to be the causal factors at least for men, presumably the factors for women are likely to be fairly similar.)

Anyway, I am arguing for trying to understand things, without strong ideological preconceptions, so we can make better policy that meets the needs of as many people as possible without putting at risk more people than need to be at risk.

It's not like if trans women or lesbians or Jesuit nuns or whomever have male-pattern criminality then we should be cruel to trans women or lesbians or Jesuit nuns or whomever. However, we might wish to establish additional precautions in certain cases.