Handwaving Prison Stats
Dismissing the problems of men in women's prisons with cherry-picked, incomplete and misrepresented data
Note: about five minutes after I originally published this post I noticed a huge error on my part requiring an immediate unpublishing followed by a substantial rewrite. So if you are a subscriber who read the version that originally appeared in your inbox - sorry!
Under the headline “Why are trans rights in prison so rarely defended?” in The Guardian, Zoe Williams made the following statements:
There were 97 sexual assaults in women’s prisons between 2016 and 2020 – seven involved trans prisoners. […] It seems pretty obvious that if the majority of sexual assaults in the women’s estate are committed by prisoners who are not trans, then a relentless focus on trans prisoners is not going to keep women safe.
The intent here is to dismiss that this is a significant problem, since after all, 7 is a much smaller number than 97, right? All that “bad faith discourse” must be responsible for making a mountain out of a molehill, when we can clearly see women are sexually assaulting each other in prison in far greater numbers than trans inmates. Or, as Peter Tatchell so charmingly put it, “non-trans women are the bigger threat”:
Not only does this claim fall apart on even a cursory inspection, but digging a little deeper reveals the staggering mess that has been made of prison statistics by the reckless disregard for accurate reporting of sex.
Since there are vastly more female inmates than trans ones, finding that 7% of sexual assaults were committed by that tiny minority of trans inmates is not reassuring.
The data on these assaults was not published by the MoJ and had to be extracted through Freedom of Information requests.
The figure of 7 assaults does not distinguish by sex, ie it includes assaults by both male and female inmates who identify as trans, and thus says little about the real risk posed by male inmates.
Crucially, the data on number of assaults - sexual or otherwise - by inmates who have legally changed sex simply doesn’t exist. Offences committed by men holding a Gender Recognition Certificate are unknown and unknowable.
Rate of sexual assaults is a very narrow standard for safety and wellbeing, given that merely being incarcerated with men is detrimental to the mental health of female inmates irrespective of whether any physical violence takes place.
In 2019, the female prison population was on average about 3,800, and so it is tempting to use this figure to determine the prevalance of sexual assault in the female prison population. However, the figure of 97 is not an annual average, but a total for the period 2016-2019. This means that the true comparison should be the total number of women incarcerated during that time.
There are around 7,000 female admissions to prison annually, so the number of women who passed through the system from 2016-2019 was in the region of 28,000. We don’t have comparable turnover figures for trans inmates as the numbers are so low, but according to the 2019 equalities report, there were 34 transgender prisoners in the female estate. This is doubtless underestimating the total number of trans prisoners to pass through during that time period, but it is unlikely to significantly change the picture.
On its face, rather than providing reassurance, it should be alarming that a demographic comprising roughly 0.1% of the female prison population were responsible for 7% of the sexual assaults over this time period. It puts the incidence of sexual assault among the general female population at about 1 in 300, and among the transgender population at closer to 1 in 5.
However, little conclusion can be drawn from this since the way the 34 transgender prisoners in the equalities report are listed makes it impossible to fathom the exact mix of sex and identity covered:
There were 34 transgender prisoners in women’s prisons: 30 reported their legal gender as female and 4 as male. When asked about the gender with which the prisoner identified, 11 identified as female, 20 as male and 3 did not provide a response.
Reading that, is it at all clear how many of these are men? Is it safe to assume that there are 20 legal females identifying as male? Are the 10 remaining legal females in fact men who hold a Gender Recognition Certificate (GRC)? It is impossible to tell.
But the picture is even more garbled than this. It turns out that the assault figures cited in The Guardian aren’t from any routine official statistical publication, but in fact had to be gleaned from a Freedom of Information request, which says the following:
In relation to the 97 sexual assaults in the female establishments between 1st Jan 2016 and 31st Dec 2019, 7 were incidents where prisoners who identify as transgender were involved.
For the purposes of this data, transgender prisoners are defined as those individuals known within prison or supervised by probation services to be currently living in, or presenting in, a gender different to their sex assigned at birth and who have had a case conference (as defined by PSI 17/2016 The Care and Management of Transgender Offenders). Prisoners or probation service users who have already transitioned and have a full Gender Recognition Certificate are excluded from this dataset.
This data is not actually separated by sex, it just groups together anyone who is “transgender”, who has not acquired a GRC, and who is housed in the female estate. What that means is that the total of 7 sexual assaults may have been committed by inmates of either sex, but excludes any committed by a male inmate who has a GRC, and who is thus is considered “female” for statistical purposes.
Things become even less clear if we refer to a different FOI request, which says:
In respect of question 1 of this part of your request i.e. number of trans offender involved in the 97 incidents of sexual assault whose legal gender is male, I can confirm MoJ holds the information you have requested. However, the figure is in the low range which constitutes personal data and as such exempt from disclosure under section 40(2) of FOIA.
If a request is made for information and the total figure amounts to five or fewer, the MoJ must consider whether this could lead to the identification of individuals and whether disclosure of this information would be in breach of our statutory obligations under the General Data Protection Regulation and/or the Data Protection Act 2018.
Still on request 1 but in respect of question 2 i.e. number of trans offender involved in the 97 incidents of sexual assault whose legal gender is female, I can confirm this figure is 6.
Prisoners or probation service users who have already transitioned and have a full Gender Recognition Certificate are excluded from this dataset.
So here, when asked how many assaults were committed by trans inmates who are legally male, the Ministry of Justice refused to respond because the number involved was low enough to allow personal identification. We can surmise that means it must be at least one, but because of the previous FOI response we know it cannot be more than seven.
The second part indicates that 6 individuals were involved who are legally female, ie, not men, but women identifying as men. Is there a one-to-one relation between individuals and incidents? We don’t know, and once again: inmates who are male GRC holders - who are legally female - are excluded from these results. Male inmates who have legally changed their sex to female are invisible in the statistics. There is no way of know how many - if any - sexual assaults were actually carried out by male inmates, because acquiring a GRC makes that prisoner invisible in the stats.
A third FOI request attempting to address this specifically and identify sexual assaults in this time period committed by those who hold a GRC was refused because the MoJ did not hold the data:
Information on whether a prisoner holds a Gender Recognition Certificate (GRC) is not routinely collected and those with GRCs are under no obligation to disclose this. For this reason, our electronic records include the current legal gender of prisoners, but not their sex recorded at birth. In order to identify the GRC status of each individual we would be required to manually examine all records held about that person in order to identify whether or not we hold information that would allow us to identify both those who did, and those who did not hold a GRC, at the time of the assault.
So given the official published reports and three FOI requests, the best we can say is that at least 1 sexual assault in the female estate between 2016 and 2019 was committed by a man who had not legally changed his sex. It is only by turning to media reports do we see that actually two sexual assaults were carried out in 2017 by the same man.
Which leaves us guessing whether the 5 remaining sexual assaults are completely accounted for by 6 female perpetrators identifying as men, while also still in the dark as to how many assaults were committed by GRC-holders. Is that reassuring? What about when the prisons minister proudly announced that since 2019 “there have been no assaults or sexual assaults committed by transgender women in women’s prisons”? Does that include males who hold a GRC? How can anyone know?
The female estate contains a mix of male and female prisoners, some of whom are considered transgender, and some of whom have legally changed their sex and are sometimes considered transgender, sometimes not, depending on the situation, maybe.
We can’t say much about any of this for certain except that the statistics are utter garbage, and that it is dismal to see, so many years after we first started housing men in women’s prisons, journalists at The Guardian handwaving away the risks without even an ounce of critical thinking, or even a nod to the people whose Freedom of Information requests exposed this mess. The claim that male inmates pose no risk to female inmates is one that cannot be substantiated, because the data simply doesn’t exist.
At best we can say a tiny number of transgender inmates have been disproportionately responsible for sexual assaults in the female prison estate, but the obfuscation of the offender’s sex in the way these have been recorded and reported and the exclusion of males who hold a GRC from the data makes it impossible to see exactly how much additional risk of sexual violence this policy has caused.
Using low incidence of sexual violence as the benchmark for the success of these policies is also a very narrow criteria for success, given that it says nothing about increased risk of assault generally, never mind the impact on the mental health of female inmates forced to share space with men.
Besides which it is absolutely damning that these barely coherent figures have to be cobbled together from a mess of FOI requests, rather than clearly published in official data. The failure to accurately record the sex of both offender and victim in the statistics on assaults in prison - sexual or otherwise - has meant that the risk these men pose to women has been completely hidden from view.
And most importantly of all none of this is new.
As confirmed by BBC reporting in 2021 after judgment was handed down in a judicial review of this policy:
Between 2016 and 2019, 97 sexual assaults were recorded in women's prisons, the judgement said. Of these, it appears that seven were committed by transgender prisoners without a GRC. It is not known whether any were committed by transgender women with a GRC.
To reiterate: two years ago, the MoJ could not produce figures when requested in court to say whether male inmates who are recorded as legally female are responsible for any of the remaining 90 sexual assaults between 2016 and 2019. The data does not exist. They may have committed none of them, or all of them, or anything in between. The data which could quantify the risk that this policy poses has deliberately not been kept. In whose interest is that?
This corruption of data and the harm it has concealed has been pointed out time and time and time again, for years. Later statistical bulletins have started separating out the male-but-legally-female GRC holders thanks to campaigners who have pressed for this clarity. Women in particular have been meticulously poring over these numbers for years, unpaid, in far greater detail than I have here, trying to make sense of the insanity and build a case for change. Gradually, prison policy is being dragged towards sanity, but it is absolutely no thanks whatsoever to those who have used their platforms to insist there was no issue, who have denigrated the work of those pushing for clarity and reform as “confected outrage”, and who now happily benefit from those efforts while completely overselling what meagre data exists and insisting that it was all a lot of fuss about nothing.