Contra Dolly on Trans Criminality

10/7/2022 update: See the addendum at the bottom for some obvious points I should have included before.

I generally find the 21st Century Salonnière’s (also known as Dolly) writing to be thoughtful and insightful, despite the controversial arena she roams around in. Back in April (I have a terrifying reading backlog, ok?) she published the post Sexual Offenses Are More Common Among Transwomen Than Men, a title as provocative as it is unambiguous.

Using prison data from both the UK and the US, Dolly finds that about 50% of trans prisoners are there for a sexual crime, in contrast to 11% - 19% of the general prison population. Hopefully it’s obvious that prisoners are overwhelmingly male by a huge margin, especially for violent offenses, and that the number of trans prisoners is so miniscule as to be almost a rounding error (in the UK there were only 142 trans prisoners out of a total prison population of 92,949 as of 2017).

Dolly’s overall argument is structured thus:

Premise- Trans prisoners are more likely to be there for a sexual offense

Conclusion- Trans people are more likely to commit a sexual offense

I’m not the only one who noticed some glaring omissions in this argument, and a few commentators pointed this out. I would hope my criticism is seen as constructive, but the main feedback I would give is to be more transparent about implicit assumptions. To be clear, Dolly’s operating assumptions (whether stated or not) may be perfectly reasonable, but the discussion is clouded when they're kept shelved away.

The two main assumptions implicitly made are:

1. Trans prisoners are representative of the trans general population

2. The amount of law enforcement attention spent on any particular crime is representative of that crime's frequency

I might be missing others, but those are at least the most important. Setting aside the validity of the prison statistics (I’m assuming they’re legit and have no reason to think otherwise) I remain skeptical these are reasonable assumptions for many of the same reasons “And I for truth” listed in their comment to the post:

It could be the case, for instance, that transwomen commit non-sex crimes at lower rates than other males but sex crimes at the same rates. It could also be that males in prison for sex crimes are more likely to claim a trans identity in the hopes of getting transferred to a woman's prison, either for nefarious reasons (access to women in prison) or self-protection (recognition of the higher risk that other male prisoners might commit violence against them). Both of those situations are consistent with the data you cite.

I’m also skeptical based on my experience as a public defender.

Sex criminals are by far the people treated the worst by the criminal justice system, both by the legal punishments but also informally by retribution from other inmates. When I'm dealing with a sex offense case, I make informal requests to the prosecutor and the judge to not read out their charges out loud, or to take their case last when the courtroom is emptier. Those same clients routinely ask me to not give them any paperwork about their case for fear that it would be discovered by others. All of this is done to protect them, by limiting the number of people who find out about the nature of their charges.

Inmates in general are almost by definition more violent that the general public, and beating the shit out of someone accused of a sex crime is the kind of violence most likely to be implicitly condoned by the powers that be. Correctional officers genuinely have a thankless and very difficult job, and the last thing on their mind is worrying about kiddie diddler getting shanked. Didn’t see it happen, and even if they did, it was self-defense because the rapist provoked it. So yes, the idea that someone accused of a sex offense is more likely to identify as trans once in prison solely for the purposes of better accommodations makes a lot of sense to me. This says less about the authenticity of trans gender identity and much more about the horrific conditions our criminal justice system casually tolerates.

There would also be an added filter at the investigation stage. Sex crimes are notoriously difficult to prosecute, absent clear evidence of violence or coercion. Law enforcement is put in a very difficult situation because they WANT to be receptive of complaints, but they often can't do anything with them due to shoddy evidence. In most of the rape cases I've handled, there is rarely any dispute whatsoever that the people involved had sex, but then the only evidence of a crime is conflicting testimony from the only two witnesses (he said she consented, she denies it). For example, I once had a case where a guy in his 40s had a friends-with-benefits relationship with a 21 year-old for at least two years. The day she reported a rape to police, she was also seven months pregnant with his child. I am not at all saying it’s impossible to rape someone you’re in a relationship with (no matter how casual), but good luck convincing a jury of that beyond a reasonable doubt.

Prosecutors are thus more likely to pursue charges like this if there are other factors to tip the balance. In the case above, the guy had a domestic violence history from a very long time ago. I also have to assume that their age gap also played a role in moving the needle towards “prosecute”. It would not be surprising if a sex crime garners more lurid attention from the law enforcement apparatus just because the suspect is trans. Prosecutors openly advertising their preferred pronouns does not disavow them of any potential bias against trans people they may harbor, and either way they can only file charges when a beat cop or detective cares enough to forward a report. We’re not pulling from a progressive crowd here.

The overall methodology is complicated by the severe dearth of data on a population this miniscule — how many conclusions can you draw from a sample that is 0.15% of the population? This is further hampered by the relatively incoherent framework of gender identity, particularly when it melds into non-binary territory — how do you determine who should be counted in this group?

One of my hobby horses is being a spoilsport for when someone tries to hoist a heavy conclusion on some flimsy stilts. We can shore this up with some figurative gravel — perhaps some reasonable assumptions to close the gaps, but these should at least be stated outright and explicitly. Short of that, sometimes we just don’t have enough evidence to come to a conclusion, and it’s ok to admit when we don’t and can’t know something. This isn’t a call to give up on trying to answer questions, because even a failed attempt to resolve an inquiry can leave us with a useful blueprint for the future.

ADDENDUM: There’s a few more points I’m embarrassed I didn’t think about before.

I should have focused primarily on the population denominator issue, as it’s the most straightforwardly obvious problem with Dolly’s analysis. The best way to illustrate this is to look at offense type for women (and full credit to gdanning for bringing this to my attention) since we know that women are more or less 50% of the overall population.

This study by the Sentencing Project breaks down offense type by gender, and page 4 includes this table:

So of the women in prison, 40% of them are there for a violent offense, and 26% for a drug offense, compared to 60% and 13% for the men in prison respectively. Based on Dolly’s argument, the conclusion to draw from this data is that women in general are more likely than men to commit a drug offense, and about 67% as likely as men to commit a violent offense. This is quite obviously wrong, because women commit crimes FAR less often than men, as Dolly herself plainly established.

According to FBI arrest data, men get arrested for violent offenses at almost 4x the rate women do. So a proper comparison would need to take into account the relevant denominator. Going back to the trans prisoner comparison, we know that 0.15% of UK prisoners are listed as trans. Whether this is a lot or a little depends entirely on what denominator we use for the trans population. If trans individuals somehow constitute 10% of the general population (a highly implausible figure), then a 0.15% prisoner rate would indicate that group is astonishingly law-abiding, even if 50% of those prisoners end up there for a sexual offense. The opposite implication would be true if we find out trans individuals constitute less than 0.15% of the general population (all this of course, assumes that incarceration rate is a good proxy for criminality, which I laid out my reasons to doubt above).

We can't draw a conclusion one way or another without knowing what denominator of the baseline population to use, and we have no idea how many trans people are in the general population, because none of the operating definitions make any sense (and almost every study I see nowadays casually lumps in non-binary within that survey category, a classification which is even more incoherent). Statistical significance (of the 142 UK prisoners sample) would not be relevant when there are so many serious problems with the sample selection.

Lastly, I readily admit I have no direct evidence demonstrating that any significant portion of prisoners are claiming to be trans for opportunistic reasons. I think the theory is very plausible given how much I know about the carceral system, and how much institutions have beholden themselves to accepting self-ID as the only qualifier. I also have no idea how one would even investigate this evidence, as "I'm only pretending to be trans" is not something a prisoner would have a reason to divulge to a researcher collecting data within a correctional institution.

But there is a corollary phenomenon that is much better documented, and it’s situational homosexuality in prisons. This is where otherwise heterosexual men have sex with other men while they’re incarcerated. It wouldn't make sense for me to tally up the numbers of prisoners who have sex with men in prison and then claim it's a proxy for criminal proclivity in the general homosexual population. That sample is not representative of the whole.