Letter to the Office for Statistics Regulation on criminal justice statistics: update (18 August 2020)

This letter may also be downloaded here

By email

18 August 2020

Dear

Many thanks for taking the time to speak to us last week, and updating us on the criminal justice statistics review. Thanks too for talking us through OSR’s proposed statement regarding the collection of data on sex and gender identity. We have had time to reflect on our discussion and write now with some further comments on both topics.

Criminal justice statistics

We were pleased to hear that the Scottish Government intends to review their statistics. We assume that officials might themselves connect with the Chief Statistician’s Sex and Gender Data Working Group. If that isn’t something they have mentioned to you, it might be worth raising with them. We also thought that HMICS might provide useful insights, particularly in relation to police data collection practices, on the basis of their crime audit work.    

In relation to the publication Statistics on Women and the Criminal Justice System we appreciate that the Ministry of Justice acknowledges that the data reflects ‘a mixture of physiological and personal identity’ (it appears that this qualification was introduced in the 2015 edition). In many ways however, this is one of our core concerns: that data producers have already, or will in future, simply relabel their outputs as ‘self-defined sex’, ‘lived sex’ or ‘gender identity’, rather than address the main problem: which is the loss of sex as a long-established core demographic variable, and the impact of this loss on data reliability. None of these terms are clearly defined. Indeed, most are contested by academics and other relevant experts working with quantitative data.

In relation to criminal proceedings data, we remain concerned that the replacement of sex with self-identified gender identity means that it is already no longer possible to draw reliable conclusions about patterns of female offending over time, and thus to make sense of the data. For offences such as homicide, rape and attempted rape and indecent assault which are rarely committed by females, it is not possible to ascertain whether spikes or inconsistencies in the data already are outliers, or result from charges where a male defendant was classified as female. For example, in Scotland, female homicide figures are volatile, varying inconsistently between 7 and 17 between 2004/15 and 2018/19. In 2016/17 the number of females charged with indecent assault (five) was more than double the annual average between 2004/5 and 2018/19 (two). There are also two rapes by females recorded, in 2006/7 and 2018/19 respectively. Losing the ability to tell whether these are like-for-like pattern by sex represents an important loss of information, which as far as we know has never been discussed with or justified to the wider community (policy makers, service providers, academics, media) with an interest in this data.

More broadly, we remain very concerned about some of the arguments being put forward by data producers, including the Scottish Government and ONS, to justify the replacement of sex with self-defined gender identity, and an unwillingness to properly address the concerns raised by data users.

Comments on OSR’s proposed statement regarding data on sex and gender identity  

We very much welcome the news that OSR intends to issue a statement setting out your expectations with regard to the collection of data on sex and gender identity. As we said on the call, this will legitimise others to question the approaches taken by different data producers.

We appreciate that the draft statement will not be seeking to make the case for collecting data on sex, rather that OSR wants to make clear to data producers what processes they must follow when collecting and presenting data on sex and gender identity.

However, we remain extremely concerned about the nature of the arguments we are seeing being levelled by data producers to justify a shift away from collecting robust data on sex and wonder whether there might be a way of addressing the following points in the draft statement.

Sensitivity and legality of asking about sex at birth

We are increasingly seeing the suggestion that it is hurtful, or even potentially unlawful, to ask a person about their sex at birth. If it is widely accepted that it is wrong, in principle, to ask a person about their sex at birth, this will result in the eventual loss of data on a core demographic variable. Data collection exercises frequently ask respondents about sensitive or personal subjects, including health, disability and sexuality. It would, for instance, be unthinkable for a similar argument to be levelled about collecting data on age, another core demographic variable. We think that the onus should be on data producers to explain why for example, questions on age, disability, health or sexuality are not deemed sensitive, whereas questions on a person’s sex are.   

On the point about legality, it would be worth OSR considering recent statements by both the Equality and Human Rights Commission and ONS. Both have claimed that to ask for someone’s sex at birth might be a breach of Article 8 of the European Convention on Human Rights (ECHR). Section 22 of the Gender Recognition Act 2004 has also been raised, although this can only apply to those with a Gender Recognition Certificate (GRC). We think it is implausible that legislators in 2004 intended to render sex unaskable of anyone, because anyone might in theory be one of the small number of individuals with a GRC (around 5,000 across the UK). We also note that the UK Government’s guidance to applicants for a GRC states that “You should bear in mind that privacy does not mean absolute secrecy… Although the gender recognition process seeks to safeguard your privacy, you do not have a right never to disclose the fact that you obtained a GRC.” We would therefore be very surprised if there were an absolute privacy bar, or anything close to that, on asking for information on sex either in administrative or research settings, in the UK or by implication within all countries covered by the ECHR. The ramifications of that would be considerable. Might OSR consider commissioning an independent legal opinion on this?

Defining and responding to ‘societal need’

You stated that your expectation would be that data producers should collect data according to ‘societal need’ and we spoke briefly about how this might manifest itself. Again, we are concerned that some data producers are already arguing that society has changed and that data on self-defined gender identity is more relevant and important than data on sex. Indeed, this was the claim made by the Scottish Government in its Policy Memorandum on the Census (Amendment) (Scotland) Bill:

“The 2011 Census recognised that society‘s understanding of sex has changed and guidance provided explained that the question was being asked in terms of self identified sex.”

We fully support the need to collect data on self-defined gender identity to help identify where there may be new information needs for this. We are not however aware of any evidence to indicate that a person’s sex is no longer a key determinant of their life experiences and opportunities, from birth onwards, across a range of policy fields, including health, education and criminal justice. As you are aware, the Scottish Household Survey no longer collects data on respondents’ sex but has supplanted the sex question with one on gender identity (by contrast we note the Northern Irish equivalent collects both). Guidance from the UK Government Equalities Office advises employers to collect data on their employees’ gender identity (but not their sex) for the purposes of reporting on their gender pay gap, a statutory requirement. Given the current direction in data collection trends, we think those data producers that have already replaced or are intending to replace, rather than supplement, sex with self-defined gender identity need to explain how they have come to the conclusion that this information no longer matters.    

Data reliability

We are aware that data producers are relying on the untested assumption that the number of cases of people who identify as the opposite sex is too small to be statistically significant. In the absence of any robust data on the number and demographic characteristics of people that identify as the opposite sex, we would suggest that this assumption is misplaced and that there are substantial risks around reliability at the subgroup level. For instance, a 2018 Swedish population study reported that 6.3% of 22 to 29 years olds wanted to be treated as members of the opposite sex (compared to a 2.8% sample average) and there are good grounds for assuming that this will not be a uniform effect by sex. Again, we think that the onus should be on data producers to investigate and test the effects of data that conflates sex and gender identity in different subpopulations ahead of making changes to data collection practices.

Clarifying data collection practices

Since 2015 the MoJ has included information on how sex is recorded by different criminal justice agencies in the Technical Report of Women in Criminal Justice Statistics (Table G.01). This is essential in order to understand what the data represents, and we think that all official data producers should clarify the data collection practices that underpin any published data on sex and gender identity in a similar way. We would however suggest that this information is made much more obvious and accessible to readers, and not placed in Technical Reports or appendices.     

Explaining change

We appreciate that there is already a general expectation under V1.6 of the Code of Practice that discontinuing a set of statistics should be done in discussion with stakeholders and other users. We think it would be helpful if the statement said explicitly that OSR expects data producers to make a clear statement on the public record justifying in detail, with evidence of the process and underlying analysis, of their rationale for ceasing to seek to collect data on sex as an objectively defined variable, including where they are replacing it or conflating it with data on gender identity. A clear statement also needs to be made when new information about data collection practices emerges, which we assume explains the acknowledgment made in the 2015 edition of Women in Criminal Justice Statistics (rather than new data collection practices across multiple CJS organisation in the same year). Where changes have previously taken place without a clear acknowledgement, we think that a retrospective statement and explanation should be published.   

Engagement with data users

Lastly, we are concerned about the nature of data producers’ engagement with data users, and how the views of particular advocacy groups continue to be privileged over expert data users and those who might advocate alternative views. This has been particularly stark in the context of the census. Our scrutiny of the planning processes undertaken by the census authorities in preparation for the 2021 census has revealed that the views of data users, such as social scientists working with population level data, have been overlooked. We would go further and argue that it can now be shown from the public record that their input on anything like equal terms to others has been resisted. Meantime, interest groups who claim to represent particular respondent cohorts have had a disproportionate influence on the development of the questions on sex and gender identity.

Again, we appreciate that engagement with relevant data users is a feature of the Code of Practice but we wonder whether a way needs to be found to make it clear in the statement that OSR has a clear expectation that data producers will not engage only or mainly with representatives of those advocating for moving away from collecting data on sex, which is clearly what has happened in the context of the census, and we strongly suppose was the case with the Scottish Household Survey. We understand this may be difficult to draft, but are concerned any less frank instruction will be read in ways which are used to support more of the unbalanced process that we and others can now show has happened with the census.

We are grateful to OSR for involving us in these discussions. Please don’t hesitate to get back in touch if we can provide you with any further information or if there is anything above that you would like to clarify with us.

Yours sincerely

Dr Kath Murray                       Lucy Hunter Blackburn                       Lisa Mackenzie