Correcting the Record on Rape Statistics

Correcting the Record on Rape Statistics

In August 2021, I made a submission to the Scottish parliament, in support of Petition PE1876/H: “Accurately record the sex of people charged or convicted of rape or attempted rape”. The submission is available online, and is also included as an appendix to this document.

I am on record supporting the view that sex should be accurately recorded in all statistics, not just crime statistics, and not just rape statistics. But my submission on this particular topic appears to have hit a nerve. Gender-identity activists have attempted to paint me as a devil, quite literally in the case of one activist who made a Youtube video defacing a photo of my face with devil horns.

As a general policy, I try to ignore social-media attacks. But I think it is worth correcting a number of falsehoods and misunderstandings which have been circulated and uncritically reproduced. These errors are found in a number of sources, including blogs and twitter threads, and I do not highlight the original sources here. My aim is not to draw attention to any individual, but simply to correct the record. I rebut the five main arguments against my submission below.

  1. The claim that the data I rely on in my submission are not designated as official statistics, and that this undermines my analysis.

National Statistics which have been assessed by the Office for Statistics Regulation as fully compliant with the Code of Practice for Statistics are deemed official statistics. The criteria are trustworthiness, quality and value. The whole basis of my submission was that the Crown Prosecution Service (CPS) statistics that I cite are not trustworthy, in the sense that they do not accurately and consistently record sex. It is unsurprising that they fail to meet the standard of official statistics. The CPS is reliant on individual police forces to provide crime data broken down by sex. Police forces do not currently provide this in a consistent way according to a common standard. Many police forces record male perpetrators as female if the perpetrator requests it. There is a lack of transparency about how police forces collect data, and how their recording practices have changed over time. The Ministry of Justice explicitly refers to a range of different recording processes for ‘sex’, including self-identified and officer-identified.

Far from undermining my analysis, the lack of trustworthy data is the point that I aim to highlight.

  • The claim that I said (or somehow implied) that all 436 rape defendants recorded as female are male.

I clearly note in my submission that, in Britain, females can be charged with rape as an accomplice. Some commentators may have assumed that all 436 defendants were male. This assumption is not justified. The point of my submission was to highlight that we cannot tell how many are male. I subsequently submitted a Freedom of Information Request asking the CPS how many of these cases were males and how many females. They were unable to provide the information.

  • The claim that (i) data collection switched from sex to gender-identity at a single point in time following the Equality Act 2010, and that therefore (ii) the lack of a clear trend in the number of rapes recorded as being perpetrated by females between 2011/12 and 2017/18 tells us that this change in data collection has had no distorting effect on the data.

Claim (i) is simply false. Sex remains a protected characteristic under EA2010. The Public Sector Equality Duty mandates that public sector bodies should monitor and publish data on the protected characteristic of sex. In addition, the Criminal Justice Act 1991 includes a statutory duty for the government to report crime data on both race and sex, in order to facilitate the administration of justice and avoid discrimination (Ministry of Justice 2019). As stated above, reporting practices on sex are inconsistent, and appear to have changed over time in a haphazard way, with no public transparency or clear policy decision-making process, and apparently in violation of existing duties which remain in place.

Claim (ii) is a non sequitur. If recording practices had changed in 2010, a time-series starting in 2011/12 would not help us to assess whether this change in recording had led to substantive bias.

  • The claim that I have suggested that all 436 rape defendants recorded as female must be males with Gender Recognition Certificates (GRCs), and that therefore I must think that an implausibly high proportion of males with GRCs are rapists, given that it is estimated that only 6-7 thousand individuals in the UK hold a GRC.

There is no basis for the claim that I (or anyone else) made this suggestion, which would indeed be implausible. Gender self-id in data collection implies that individuals do not need a GRC in order to be recorded as the opposite-sex.

  • Objections to the law on rape in England and Wales.

Rape is classified as a male crime in England and Wales, in that it requires non-consensual penetration with a penis, and females can only be charged with rape as an accomplice. Some commentators, particularly from the US, have expressed affront that this should be the case, because it means that the definition of rape is not gender-neutral. This objection to the fact that sexed bodies are incorporated into the definition of the crime of rape is of course a normative rather than an empirical objection, and has no bearing on the facts.

The ability to name male violence and the role of male-bodied people as perpetrators appears central to the outrage that discussion of men and rape can provoke. One male commentator opined that “…the presence of a penis is not a very useful explanation of why rape is so heavily gendered”.


My submission to the Scottish parliament simply supported the demand that the sex of rape suspects should be recorded accurately. As I said “Sex and gender identity are distinct variables. We can respect people’s gender identities without denying the material reality of sex. Ideally, crime statistics should record both variables. It is vital that information on sex is recorded, given the importance of sex in itself, and its intersection with other variables, including gender identity.”

If we had accurate data on the sex of defendants, we would have no need to debate these figures. None of the attacks on my submission have explained why my critics think it is better not to collect accurate data on sex.  

Appendix: Submission in Support of Petition 1876: Accurately record the sex of people charged or convicted of rape or attempted rape

Professor Alice Sullivan, Professor of Sociology, University College London, Social Research Institute.

Submitted 27th August 2021.

Quantitative social scientists, including criminologists, have made clear that accurate data on sex remains fundamentally important. The quantitative social science community have made this case in an open letter to the census authorities, signed by 80 UK quantitative social scientists, and in a submission to Roger Halliday’s consultation on draft guidance issued on the collection of data on sex and gender in Scotland, signed by 91 UK quantitative social scientists. The signatories to these letters include some of the most eminent scholars in their fields, including leaders of major studies.

Sex is a fundamental demographic variable, which is fundamental to analysis across the human and social sciences (Sullivan 2020, 2021). While sex is an important predictor of outcomes across the board, crime represents a particularly extreme example. The overwhelming majority of individuals convicted of violent crime are male, and females represent a tiny minority of those convicted of sexual assault of any kind. Rape is a male crime by definition, requiring non-consensual penetration with a penis. The vast majority of victims of rape are female.

It may be argued that, as the number of people who identify as other than their biological sex is small, so the data error generated by recording gender identity rather than sex for this group will also be small. This argument may seem intuitive, but is in fact incorrect. Small numbers of misallocated cases can have a large effect on research findings in any sub-group analysis where one sex is dominant. Crime statistics in general, and sexual crime statistics in particular, provide a clear example of this.

Office for National Statistics (ONS) figures for England and Wales show that 94 per cent of convicted murderers and 97 per cent of individuals prosecuted for sexual offences other than rape are male. Meanwhile, trans identification among men convicted of custodial offences is certainly high enough to affect the data: one in 50 male prisoners in England and Wales identify as trans.

The issue of rapists identifying as women is not hypothetical. The table below shows Crown Prosecution Service data for England and Wales. Rape is classified as a male crime in UK law, and females can only be charged with rape as an accomplice. Documented instances of females being charged with rape as an accomplice are rare. Yet, between 2012 and 2018, the proportion of rape defendants classified as women varied between 1.2 per cent and 1.8 per cent. During this seven-year period, 436 individuals prosecuted for rape were recorded as women.

 Apr ’11 to Mar ’12Apr ’12 to Mar ’13Apr ’13 to Mar ’14Apr ’14 to Mar ’15Apr ’15 to Mar ’16Apr ’16 to Mar ’17Apr ’17 to Mar ’18
CPS Table 1: Sex of defendants and victims in rape prosecutions, year ending March 2012 to year ending March 2018 – Scroll to see all columns

Source: Crown Prosecution Service (CPS)  2018,

Sex and gender identity are distinct variables. We can respect people’s gender identities without denying the material reality of sex. Ideally, crime statistics should record both variables. It is vital that information on sex is recorded, given the importance of sex in itself, and its intersection with other variables, including gender identity.

Ultimately, we need to remember what crime statistics are for. Such statistics do not exist to affirm the identities of the perpetrators of crime. They exist to further our understanding of crime. The accuracy of such statistics is vital, both as a foundation for data analysis, and to ensure that public trust is not undermined.


Sullivan, A. (2020). Sex and the census: why surveys should not conflate sex and gender identity. International Journal of Social Research Methodology23(5), 517-524. [Open access version SSRN]

Sullivan, A. (2021). Sex and the Office for National Statistics: A Case Study in Policy Capture. The Political Quarterly. [Open access]

Leave a comment