Discrimination law and the experimental method

I want to apply yesterday’s dazzling insight that peanuts have to be left out of peanut-free meals to the words of the Equality Act and the specific question before the Supreme Court in For Women Scotland v Scottish Ministers, to be heard later this month. This is another fairly short point, though a little more technical than yesterday’s. 

Broadly, the job of the EqA is to prohibit discrimination because of the various protected characteristics. But there are exceptions, so that it remains lawful for director of a play to insist that Juliet is played by a girl and Romeo by a boy, or for a charity to define its beneficiaries by reference to race, national origin or sexual orientation. 

Paragraphs 26 and 27 of schedule 3 make it lawful to provide separate-sex services, and single-sex services, in situations engaging considerations like bodily privacy and dignity. They are expressed in general terms: what’s permitted is providing “single-sex” services or “separate services for persons of each sex”. Obviously that means excluding persons of the other sex. And the exclusion can only be a blanket rule, or the service can’t truthfully be described as separate or single-sex, just as you can’t describe a meal as “peanut free” if you sometimes put peanuts in it. 

That much is straightforward, or ought to be. (There is in fact plenty of dissent to it out there, some of it undeniably heavyweight. Nevertheless, I think the law is clear.)

The question for the Supreme Court in FWS is whether the protected characteristic of sex in the EqA — whether someone is regarded as a man or a woman — is affected by section 9(1) of the GRA, so that a man with a gender recognition certificate declaring him to be a woman counts as a woman for the purposes of discrimination law. In other words, whether “sex” in the EqA just means sex; or whether it means sex except for people with GRCs, in which case it means the sex they are deemed to be because of their GRI. We can call these two possibilities “sex” and “certificated sex” for short. 

The answer to this question determines what kind of discrimination a man with a GRC declaring him to be a woman is subjected to if he is excluded from a women-only service. 

The law has developed a thought experiment, complete with imaginary “experimental control”, as the way of finding out whether someone has suffered discrimination for a particular reason. You don’t have to be a scientist to use experimental controls: we all do it pretty intuitively. 

Suppose your desk lamp isn’t working. Is it the bulb? Is the socket it’s plugged into live? Is it the fuse in the plug? Is the switch in the “on” position? You  find out which is the culprit by trying different things one by one. You change the bulb, keeping everything else the same. Does it light? If so, the problem was the bulb. If not, you put the old bulb back, and try the switch in the other position. Still no light? Switch the switch back, and plug the lamp into a socket you know is live. 

Similarly, if Chris is refused entry to the women’s changing room on account of his obviously male appearance, is that because of his sex? The common sense answer is “yes”. But a GRC transcends (or confounds) common sense: if it operates in the context of the EqA, what matters is Chris’s certificated sex, not his actualy sex, so it tells us that Chris is a woman. To find out whether Chris has been excluded because of his sex, we have to compare him with someone who is of the opposite sex, and ask whether that person would have been excluded, too. 

So, obedient to the pretence required of us by Chris’s GRC, we set to work constructing a comparator. We say “Chris is a woman, so a person of the opposite sex is a man, let’s call him Christopher. This is a women-only space, so Christopher would have been excluded just like Chris. So Chris wasn’t excluded because of his (deemed female) sex, because a person of the opposite sex would also have been excluded.” (In truth, the chances are no-one will ever do anything to Chris because of his “female” sex, because it’s almost certainly obvious that he’s a bloke.)

If we run the same thought experiment on the different PC of gender reassignment, we get a different answer. The law tells us that Chris is a woman (even while our senses tell us different). Chris is a woman with the PC of gender reassignment: although legally a woman, he is a woman not by physiology, but by legal deeming. Obviously a woman without the PC of gender reassignment — that is, an actual female woman — would not have been excluded. So we have our answer: the reason Chris has been excluded is because of his PC of gender reassignment, not his sex. 

That means that excluding Chris can’t be justified under ¶26 or 27 of schedule 3, because those operate to permit sex discrimiation. But it may still be lawful to exclude Chris, because ¶28 of schedule 3 provides that it’s not unlawful to discriminate on grounds of gender reassignment in relation to the provision of single or separate-sex services, provided “the conduct in question is a proportionate means of achieving a legitimate aim”. 

Remember, all this reasoning is proceeding on the assumption that that a GRC changes Chris’s sex for the purposes of the EqA. The weird thing about ¶28, on this assumption, is that it seems to say you have to work out whether the thing you did — excluding Chris — was a proportionate means of achieving a legitimate aim. But it’s always going to be — because, well, peanuts. If the space or service is single-sex, you can’t let a man in (even a man with a certificate), or in every sense that matters it’s no longer a single-sex space. A legal fiction can deem a man married to another man to be in a heterosexual marriage, or deem him to be female for pension purposes, etc, but  it doesn’t actually change the reality or the real consequences of a male body (or even the real consequences of the theoretical possibility of a male body). It won’t affect the trauma reaction of the already-traumatised female user of that space, or the justifiable outrage and affront of the non-traumatised woman who looks up when taking her knickers off to meet the eye of a man in a space she was told was for women only. The fact that the man in question has a secret certificate at home in a drawer won’t — even if somehow she knows about it — make her feel any less embarrassed, angry or alarmed. 

So ¶28 seems to call for  “case by case” decision-making in a situation in which only a blanket rule will do. I explored the practical impossibility of that here: https://www.legalfeminist.org.uk/2022/02/16/admission-to-women-only-spaces-and-case-by-case-assessment/. 

There are a lot of reasons why the Supreme Court should find for FWS, and this is only a relatively small one. But I think it’s pleasingly neat. 

The reason I say that is that the “certificated sex” assumption leads you into this weird, artificial, counter-factual reasoning about when you can and can’t exclude Chris — and you end up apparently having to make a case-by-case assessment of something that can only be satisfactorily dealt with by way of a general rule, precisely because the “single-sexness” of the space is about what you tell the female users of the space, and whether they can trust you. It’s not really about Chris and his individual characteristics at all. 

But as soon as you remove the GRA spanner from the works of the EqA, this bit of the machine starts running smoothly and rationally. 

On that assumption — that s.9(1) of the GRA doesn’t affect the EqA — this bit of the law can recognise Chris as the man he looks like, and is. He’s excluded because of his sex, which for these purposes remains male. And that’s lawful under ¶26 or 27 if it’s lawful to run a single-sex space at all. 

So what’s ¶28 for, on this hypothesis? Good question! I’m glad you asked it, because the answer is elegant and satisfying. The point of ¶28 is to make it lawful, where appropriate, to exclude not men, but some women from the space, because of their PC of gender reassignment. 

Mostly, women who say they are men (“trans men”) will be perfectly welcome in women-only spaces. That’s because they are women, with female bodies. Their presence won’t affront, humiliate or alarm anyone, and they are likely to have the same needs as any other woman. 

But some “trans men” have taken extreme steps to look like men. Women who do this can often do it quite successfully, for precisely the same reason that men who say they are women almost always remain very visibly male. The reason is testosterone. Testosterone is a powerful drug, and a one-way ticket. A man who has gone through male puberty will almost never be able to disguise its effects successfully in later life. But when a woman takes testosterone, she’s likely to acquire a much more male-looking physique, a broken male-sounding voice, facial hair and male-pattern baldness. So some women with the PC of gender reassignment really do look and sound pretty much like men, and there will be circumstances in which it is genuinely necessary to exclude them from women-only spaces for the sake of the other women in them. 

Obviously, this is a fact-sensitive judgement which will depend on the particular nature of the space or service, who else is likely to be using it, how it is organised, and how convincingly masculine is the appearance of the trans-identifying woman in question. In other words, it calls for precisely the kind of “case by case” decision-making that ¶28 seems to envisage. The difference is that on this hypothesis — that “sex” in the EqA simply means “sex” —“case by case” makes perfect sense.