Why we Need Data on Both Sex and Gender Identity

The Office for National Statistics (ONS) is among many bodies to move away from a conventional understanding of sex in the face of lobbying. Sex is a fundamental demographic and explanatory variable, yet in recent years, sex has been overwritten by gender identity across a range of data collection exercises.

Possible arguments in favour of this move include the fact that demographic variables often change over time, and the claim that sex is less important in shaping outcomes of interest than gender. But do these arguments stand up to scrutiny?

ONS have claimed that “within today's society the traditional view of gender as a binary classification, male or female, is changing”. No one would deny that gendered roles have changed over time. But does that really imply that humans no longer have just two sexes, male and female? Or that sex is no longer relevant to our lives?

Underpinning ONS’s claim is a confusion between three distinct concepts: sex, gender and gender identity. Sex is a biological variable, determined at conception, referring to the two human reproductive classes, male and female. Gender is a social category, referring to the stereotypes and roles that may be imposed on males and females, according to their sex. Gender identity refers to how individuals see themselves, rather than how others see them. The concept of gender identity is evolving, and includes non-binary as well as cross-sex and other identities.

To add to the confusion, not only is gender often used as a polite synonym for sex, but gender and gender identity are often used interchangeably, generating an unhelpful elision between the social and the psychological.

It is certainly the case that demographic variables can change over time. For example, the ethnic group categories used in a society at any given time will tend to reflect the groups that are numerous enough to capture meaningfully, and so will change in line with migration flows and fertility. Occupational categories and social-class schemas are revised due to changing labour markets. New questions are added, for example the addition of questions on sexuality and gender identity as these categories have become politically visible. No one has argued against collecting data on gender identity, however. The contested issue is the attempt to remove sex via redefinition in terms of gender identity.

Categories like race and social class are socially constructed, implying a degree of arbitrariness. These categories change as society changes. But sex is different. The social implications of being male or female have certainly changed over time and differ between cultures. But the biological categories remain constant, and have been recognised in all societies throughout history, for the simple reason that they are the basis for human reproduction.

A second argument used for replacing sex with gender identity in data collection is that it is not sex itself that affects a host of outcomes, from health to earnings, but ‘gender’, so why would we collect data on sex? This argument similarly rests on a confused view of the relationship between sex and gender, and a conflation of gender and gender identity.

The idea that ‘gender’ affects outcomes is uncontroversial. Those who argue for the importance of sex are often accused of “biological essentialism”, that is of believing that social forces play no role, but this is a straw man. No social scientist would deny that the impacts of sex are often socially mediated. To illustrate, sex selective abortion means that, in countries such as China and India, more males than females are born. This is due to the higher status of males in these societies. This is a clear example of the effects sex being mediated by gender (the social structures around sex) but not by gender identity. To acknowledge that foetuses are selected for abortion on the basis of their sex does not imply biological essentialism. This example demonstrates that, to understand the way that gendered hierarchies operate in any given society, we need data on sex.

But is it possible that it is really gender identity, and not sex, which affects people’s lives? This argument suggests that sex is a mere proxy for gender identity, as the true underlying explanatory variable. The claim that gender identity should replace sex is very different from the (uncontroversial) claim that gender identity matters in addition to sex.

For some outcomes, such as the risk of becoming pregnant or contracting testicular cancer, the evidence against the claim that gender identity is always the salient variable, is, to say the least, overwhelming. In others, such as patterns of criminality, there is growing empirical evidence suggesting that sex trumps identity.

Nevertheless, the importance of gender identity in relation to sex may differ across domains in complex ways. There is no reason to expect the same pattern of difference across the wide range of outcomes that are examined by human and social scientists, such as physical and mental health, sport, education, the labour market, crime, and social attitudes.

Importantly, sex and gender identity are likely to intersect in ways that mean we cannot fully understand some people’s lives and outcomes without collecting both variables. We can acknowledge people’s identities without denying their sex.

The hypothesis that gender identity is a more powerful predictor than sex for any particular outcome cannot be tested without data on both sex and gender identity. Those who use the claim that gender identity trumps sex as a justification for not collecting data on sex seek to make it impossible to test their assumptions. This demonstrates a radically anti-scientific approach to evidence.

Indeed, objections to the collection of data on sex appear quasi-religious rather than empirical, often taking the form of moral denunciation rather than argument. A recent letter, organised and predominantly signed by North American academics and students, accused advocates for the retention of sex in UK Census data of taking an “abominable moral position”, likening this to "slavery, eugenics, forced sterilisation, the denial of women’s suffrage".

This righteous zealotry has been accompanied by a more general stifling of discussion of sex and gender. Accurate data on sex is fundamental to any analysis of the differences and inequalities between women and men, girls and boys. The drive to undermine sex-based data collection is fundamentally a form of silencing, designed to make certain facts unknowable and unspeakable.